WEBVTT

1
00:00:03.399 --> 00:00:07.719
<v Speaker 1>Welcome to Bedtime Astronomy. Explore the wonders of the cosmos

2
00:00:07.759 --> 00:00:12.279
<v Speaker 1>with our soothing Bedtime Astronomie podcast. Each episode offers a

3
00:00:12.359 --> 00:00:16.320
<v Speaker 1>gentle journey through the stars, planets, and beyond, perfect for

4
00:00:16.399 --> 00:00:20.239
<v Speaker 1>unwinding after a long day. Let's travel through the mysteries

5
00:00:20.239 --> 00:00:22.440
<v Speaker 1>of the universe as you drift off into a peaceful

6
00:00:22.480 --> 00:00:23.800
<v Speaker 1>slumber under the night sky.

7
00:00:26.960 --> 00:00:30.079
<v Speaker 2>Imagine staring at the night sky without blinking, like, not

8
00:00:30.120 --> 00:00:33.439
<v Speaker 2>for a minute, not for an hour, but for four straight.

9
00:00:33.200 --> 00:00:37.359
<v Speaker 3>Years, just an unyielding, relentless gaze locked onto.

10
00:00:37.119 --> 00:00:40.439
<v Speaker 2>The dark exactly, no glancing away, no resting your eyes.

11
00:00:41.079 --> 00:00:45.640
<v Speaker 2>That is exactly what NASA's Transiting Exoplanet Survey satellite that

12
00:00:45.719 --> 00:00:48.640
<v Speaker 2>are known as TESS has been doing. It has been

13
00:00:48.759 --> 00:00:54.079
<v Speaker 2>staring completely unblinking at over two point two million stars.

14
00:00:54.479 --> 00:00:56.840
<v Speaker 2>And it's not just looking at them in a passive,

15
00:00:56.920 --> 00:00:58.079
<v Speaker 2>romantic sort of way.

16
00:00:58.320 --> 00:01:02.880
<v Speaker 3>No, it's watching for the absolutely faintest, most microscopic dimming

17
00:01:02.960 --> 00:01:03.640
<v Speaker 3>of their starlight.

18
00:01:03.840 --> 00:01:06.560
<v Speaker 2>Yeah, we are talking about drops and lights so small

19
00:01:06.599 --> 00:01:09.719
<v Speaker 2>that they push the absolute boundaries of our technological limits.

20
00:01:09.840 --> 00:01:12.640
<v Speaker 3>To really appreciate the scale of that microscopic dimming, we

21
00:01:12.719 --> 00:01:15.680
<v Speaker 3>basically have to throw out our everyday understanding of brightness.

22
00:01:15.840 --> 00:01:18.519
<v Speaker 3>How so, well, we are talking about a drop in

23
00:01:18.560 --> 00:01:22.840
<v Speaker 3>stellar luminosity that is so minuscule it would be entirely

24
00:01:22.959 --> 00:01:25.799
<v Speaker 3>imperceptible to the human eye, even to you are floating

25
00:01:25.799 --> 00:01:28.480
<v Speaker 3>in a spacesuit right next to the telescope lenses. Yeah.

26
00:01:28.599 --> 00:01:32.439
<v Speaker 3>Tests is essentially this collection of ultra sensitive cameras designed

27
00:01:32.439 --> 00:01:36.200
<v Speaker 3>to capture incredibly subtle dips in light, dips that occur

28
00:01:36.239 --> 00:01:38.719
<v Speaker 3>when a planet passes directly in front of its host

29
00:01:38.879 --> 00:01:42.719
<v Speaker 3>star from our specific vantage point here in the Solar System.

30
00:01:42.760 --> 00:01:44.799
<v Speaker 2>So it's kind of like a game of shadows played

31
00:01:44.840 --> 00:01:47.000
<v Speaker 2>across trillions of miles of empty space.

32
00:01:47.079 --> 00:01:48.159
<v Speaker 3>That's exactly what it is.

33
00:01:48.200 --> 00:01:51.760
<v Speaker 2>And because TESS has been maintaining this unblinking vigil for

34
00:01:51.840 --> 00:01:55.680
<v Speaker 2>its first four years of operation, it has accumulated a

35
00:01:55.760 --> 00:01:59.680
<v Speaker 2>mountain of data that is frankly almost impossible for a

36
00:01:59.760 --> 00:02:01.239
<v Speaker 2>human brain to comprehend.

37
00:02:01.319 --> 00:02:05.120
<v Speaker 3>Oh, the volume is staggering. Yeah, we are talking about continuous,

38
00:02:05.280 --> 00:02:09.520
<v Speaker 3>unbroken light curves for two point two million distinct fusion

39
00:02:09.560 --> 00:02:11.439
<v Speaker 3>reactors scattered across the galaxy.

40
00:02:11.560 --> 00:02:14.759
<v Speaker 2>But today we are jumping straight into a monumental breakthrough

41
00:02:14.879 --> 00:02:17.280
<v Speaker 2>that just emerged from that exact mountain of.

42
00:02:17.280 --> 00:02:18.599
<v Speaker 3>Data, right the new research.

43
00:02:18.800 --> 00:02:22.280
<v Speaker 2>Yeah, astronomers at the University of Warwick have successfully waded

44
00:02:22.319 --> 00:02:27.120
<v Speaker 2>through that digital ocean and validated over one hundred exoplanets. Specifically,

45
00:02:27.199 --> 00:02:29.719
<v Speaker 2>they've handed us one hundred and eighteen validated planets, and

46
00:02:29.759 --> 00:02:32.439
<v Speaker 2>the really exciting part is the thirty one newly detected ones.

47
00:02:32.280 --> 00:02:36.080
<v Speaker 3>Right, Exactly thirty one of those are completely fresh discoveries

48
00:02:36.319 --> 00:02:40.840
<v Speaker 3>that literally nobody on Earth knew existed until the publication

49
00:02:40.919 --> 00:02:43.599
<v Speaker 3>of this research, which is wild, it really is. Thirty

50
00:02:43.639 --> 00:02:46.199
<v Speaker 3>one brand new worlds is a staggering yield for a

51
00:02:46.240 --> 00:02:50.199
<v Speaker 3>single study, especially when you consider how thoroughly the low

52
00:02:50.240 --> 00:02:54.639
<v Speaker 3>hanging fruit in exoplanet research has already been picked.

53
00:02:54.719 --> 00:02:56.719
<v Speaker 2>Yeah, we've been at this for a while now, we have.

54
00:02:57.080 --> 00:03:00.280
<v Speaker 3>But the Warwick team didn't stop there. Alongside those one

55
00:03:00.360 --> 00:03:04.520
<v Speaker 3>hundred and eighteen validated planets, they also produce this massive,

56
00:03:04.719 --> 00:03:09.840
<v Speaker 3>highly refined catalog of over two thousand high quality planet.

57
00:03:09.520 --> 00:03:12.840
<v Speaker 2>Candidates, and nearly one thousand of those candidates are entirely

58
00:03:12.840 --> 00:03:13.800
<v Speaker 2>new to science too.

59
00:03:14.479 --> 00:03:18.800
<v Speaker 3>They've basically handed the global astronomical community a highly detailed,

60
00:03:19.120 --> 00:03:20.280
<v Speaker 3>curated treasure map.

61
00:03:20.360 --> 00:03:22.719
<v Speaker 2>Okay, let's unpack this, because to really wrap your head

62
00:03:22.719 --> 00:03:25.159
<v Speaker 2>around how difficult it is to find these things. Let's

63
00:03:25.199 --> 00:03:28.199
<v Speaker 2>think about the method they're actually using, the transit method.

64
00:03:28.319 --> 00:03:31.879
<v Speaker 2>Right in astrophysics, this is called the transit method. Imagine

65
00:03:31.919 --> 00:03:33.759
<v Speaker 2>trying to spot a moth flying in front of a

66
00:03:33.800 --> 00:03:34.719
<v Speaker 2>distant street light.

67
00:03:34.960 --> 00:03:37.680
<v Speaker 3>I love this analogy, thanks, But the street.

68
00:03:37.520 --> 00:03:39.719
<v Speaker 2>Light isn't down the block at the end of your street.

69
00:03:40.120 --> 00:03:43.000
<v Speaker 2>It is hundreds of light years away. You can't see

70
00:03:43.000 --> 00:03:46.039
<v Speaker 2>the moth itself. The moth is entirely invisible.

71
00:03:46.080 --> 00:03:47.960
<v Speaker 3>You're just looking for the shadow exactly.

72
00:03:48.240 --> 00:03:51.360
<v Speaker 2>You are solely trying to measure the exact fraction of

73
00:03:51.400 --> 00:03:53.960
<v Speaker 2>a shadow that the moth casts on your eye as

74
00:03:54.000 --> 00:03:57.520
<v Speaker 2>its tiny, fragile body blocks a microscopic sliver of the

75
00:03:57.520 --> 00:04:00.280
<v Speaker 2>bulb's light for just a few hours.

76
00:04:00.400 --> 00:04:02.439
<v Speaker 3>It's a great way to picture it, but I'd actually

77
00:04:02.439 --> 00:04:05.759
<v Speaker 3>take it a step further to highlight the sheer statistical

78
00:04:05.840 --> 00:04:07.599
<v Speaker 3>improbability of what test is doing.

79
00:04:07.680 --> 00:04:08.439
<v Speaker 2>Okay, go for it.

80
00:04:08.759 --> 00:04:11.479
<v Speaker 3>So the distant street light is the host star and

81
00:04:11.520 --> 00:04:15.479
<v Speaker 3>the moth is the exoplanet. Yes, but remember planets don't

82
00:04:15.520 --> 00:04:18.639
<v Speaker 3>just fly around randomly. They orbit in a flat plane,

83
00:04:18.839 --> 00:04:21.160
<v Speaker 3>right like a record player exactly, So for us to

84
00:04:21.199 --> 00:04:24.399
<v Speaker 3>see that shadow, that temporary drop in the star's light curve.

85
00:04:24.839 --> 00:04:27.439
<v Speaker 3>The geometry has to be absolutely perfect. The edge of

86
00:04:27.439 --> 00:04:30.160
<v Speaker 3>that planetary system has to be perfectly aligned with our

87
00:04:30.199 --> 00:04:30.720
<v Speaker 3>line of sight.

88
00:04:30.839 --> 00:04:31.920
<v Speaker 2>Oh, I say yeah.

89
00:04:31.920 --> 00:04:34.199
<v Speaker 3>If the system is tilted even a few degrees up

90
00:04:34.279 --> 00:04:37.600
<v Speaker 3>or down relative to Earth, the planet just orbits above

91
00:04:37.759 --> 00:04:40.519
<v Speaker 3>or below the star from our perspective, and we see

92
00:04:40.560 --> 00:04:41.480
<v Speaker 3>absolutely nothing.

93
00:04:41.680 --> 00:04:43.680
<v Speaker 2>So not only are we looking for a moth's shadow

94
00:04:43.759 --> 00:04:46.399
<v Speaker 2>from light years away, we are only able to see

95
00:04:46.399 --> 00:04:48.759
<v Speaker 2>the tiny fraction of moths that happen to be flying

96
00:04:48.800 --> 00:04:52.079
<v Speaker 2>on the exact right trajectory across our line of vision.

97
00:04:51.920 --> 00:04:55.800
<v Speaker 3>Precisely, which means for every planet test actually catches, there

98
00:04:55.839 --> 00:04:59.600
<v Speaker 3>are likely dozens, if not hundreds of others orbiting just

99
00:04:59.680 --> 00:05:01.040
<v Speaker 3>out of geometric alignment.

100
00:05:01.160 --> 00:05:04.040
<v Speaker 2>That's humbling. The universe is teeming with these things, but

101
00:05:04.120 --> 00:05:07.199
<v Speaker 2>we are only catching the ones that cross our incredibly narrow.

102
00:05:06.959 --> 00:05:10.920
<v Speaker 3>Tripwire, and we are looking for a highly specific periodic

103
00:05:11.000 --> 00:05:13.920
<v Speaker 3>dip in brightness. The moth has to circle the street

104
00:05:14.000 --> 00:05:16.279
<v Speaker 3>light and block the light again and again on a

105
00:05:16.279 --> 00:05:17.279
<v Speaker 3>predictable schedule.

106
00:05:17.399 --> 00:05:20.680
<v Speaker 2>Right. That periodicity is the first major clue it is.

107
00:05:20.800 --> 00:05:24.480
<v Speaker 3>But you know, the universe is a remarkably messy, chaotic place.

108
00:05:25.000 --> 00:05:26.959
<v Speaker 3>A lot of things can cause a star's light to

109
00:05:27.000 --> 00:05:30.920
<v Speaker 3>flicker or dim like what, well, stars have massive internal

110
00:05:30.959 --> 00:05:33.800
<v Speaker 3>weather systems, They have sunspots larger than the Earth that

111
00:05:33.920 --> 00:05:36.199
<v Speaker 3>rotate in and out of you. They have violent flares.

112
00:05:36.399 --> 00:05:39.120
<v Speaker 2>Oh right, so it's not a steady bolt exactly.

113
00:05:39.560 --> 00:05:43.240
<v Speaker 3>And the instruments on the satellite itself experienced thermal shifts

114
00:05:43.319 --> 00:05:47.480
<v Speaker 3>that introduce noise into the data. Distinguishing the delicate, repeating

115
00:05:47.600 --> 00:05:51.079
<v Speaker 3>shadow of our proverbial moth from all the other violent

116
00:05:51.199 --> 00:05:54.920
<v Speaker 3>cosmic noise is the real challenge of modern astronomy.

117
00:05:54.720 --> 00:05:57.279
<v Speaker 2>Which brings us to why this matters to you listening

118
00:05:57.360 --> 00:05:59.959
<v Speaker 2>right now. The focus of this breakthrough isn't just about

119
00:06:00.360 --> 00:06:02.959
<v Speaker 2>finding a few new dots in the sky or adding

120
00:06:02.959 --> 00:06:06.879
<v Speaker 2>a list of unpronounceable alphanumeric names to a stellar catalog.

121
00:06:07.120 --> 00:06:07.720
<v Speaker 3>Definitely not.

122
00:06:07.800 --> 00:06:11.920
<v Speaker 2>It's about a revolutionary new artificial intelligence tool that is

123
00:06:11.959 --> 00:06:15.519
<v Speaker 2>completely changing how we map the universe. We are witnessing

124
00:06:15.560 --> 00:06:18.639
<v Speaker 2>a fundamental shift in our astronomical capabilities.

125
00:06:18.680 --> 00:06:22.319
<v Speaker 3>We really are. The era of humans painstakingly plotting graphs

126
00:06:22.319 --> 00:06:24.519
<v Speaker 3>to find planets is over. It had to end, didn't

127
00:06:24.519 --> 00:06:27.240
<v Speaker 3>the Oh absolutely, when you were looking at continuous light

128
00:06:27.319 --> 00:06:30.720
<v Speaker 3>readings from two point two million stars, taking measurements every

129
00:06:30.720 --> 00:06:33.480
<v Speaker 3>two minutes. In some cases, the data volume becomes so

130
00:06:33.720 --> 00:06:36.839
<v Speaker 3>vast that it is physically impossible for human eyes to

131
00:06:36.879 --> 00:06:37.639
<v Speaker 3>process at all.

132
00:06:37.800 --> 00:06:40.680
<v Speaker 2>Not even massive international teams of human eyes.

133
00:06:40.600 --> 00:06:43.480
<v Speaker 3>Not a chance. You can't just throw a room full

134
00:06:43.480 --> 00:06:46.160
<v Speaker 3>of tired graduate students at this problem and ask them

135
00:06:46.199 --> 00:06:48.319
<v Speaker 3>to stare at squiggly lines on monitors all day.

136
00:06:48.480 --> 00:06:49.920
<v Speaker 2>Right, that sounds like torture.

137
00:06:50.319 --> 00:06:52.480
<v Speaker 3>It is. You need a way to separate the true

138
00:06:52.480 --> 00:06:55.519
<v Speaker 3>planets from the cosmic illusions, and you need to do

139
00:06:55.560 --> 00:06:59.240
<v Speaker 3>it at an industrial scale. And that absolute necessity is

140
00:06:59.319 --> 00:07:02.120
<v Speaker 3>exactly what birth this new AI pipeline.

141
00:07:02.160 --> 00:07:04.519
<v Speaker 2>You know. I always find it fascinating how the biggest

142
00:07:04.600 --> 00:07:08.639
<v Speaker 2>roadblock in a scientific field often isn't the technology used

143
00:07:08.639 --> 00:07:11.920
<v Speaker 2>to gather the data, but our capacity to actually read it.

144
00:07:12.160 --> 00:07:14.079
<v Speaker 3>That's very common issue in modern science.

145
00:07:14.360 --> 00:07:17.600
<v Speaker 2>So what exactly is the bottleneck here? If tests is

146
00:07:17.639 --> 00:07:21.120
<v Speaker 2>so incredibly sensitive, why is it so hard to just

147
00:07:21.199 --> 00:07:23.560
<v Speaker 2>look at a dip in the light curve and confidently

148
00:07:23.560 --> 00:07:26.399
<v Speaker 2>announced to the world, yep, we found another Earth. Pack

149
00:07:26.439 --> 00:07:26.920
<v Speaker 2>your bags.

150
00:07:26.959 --> 00:07:30.000
<v Speaker 3>Well, the massive bottleneck in modern astronomy right now isn't

151
00:07:30.000 --> 00:07:34.079
<v Speaker 3>detection at all. It's conformation. Okay, missions like tests or

152
00:07:34.079 --> 00:07:38.519
<v Speaker 3>its legendary predecessor, the Kepler Space Telescope, are incredibly proficient

153
00:07:38.560 --> 00:07:42.759
<v Speaker 3>at flagging anomalies. They routinely identify thousands of possible planet candidates.

154
00:07:42.800 --> 00:07:44.639
<v Speaker 2>They say, hey, the light dipped tier, you should look

155
00:07:44.680 --> 00:07:46.279
<v Speaker 2>at this exactly.

156
00:07:46.360 --> 00:07:50.160
<v Speaker 3>But confirming which of those signals are real physical planets

157
00:07:50.199 --> 00:07:54.639
<v Speaker 3>and which are illusions is painstakingly slow. The primary culprit

158
00:07:54.680 --> 00:07:57.639
<v Speaker 3>holding everything up in the astronomical community is something we

159
00:07:57.720 --> 00:07:59.279
<v Speaker 3>call false positives.

160
00:07:58.920 --> 00:08:01.240
<v Speaker 2>Because the universe loves to play tricks on us.

161
00:08:01.279 --> 00:08:04.439
<v Speaker 3>Oh, it really does. It generates phenomena that look exactly

162
00:08:04.480 --> 00:08:07.279
<v Speaker 3>like what we are searching for but are entirely different beasts.

163
00:08:07.399 --> 00:08:10.120
<v Speaker 2>Right, and the most common, the most frustrating trick the

164
00:08:10.199 --> 00:08:14.720
<v Speaker 2>universe plays on exoplanet hunters is the eclipsing binary star system.

165
00:08:14.839 --> 00:08:18.160
<v Speaker 3>Right, that is the big one. Yes. To understand this,

166
00:08:18.199 --> 00:08:20.480
<v Speaker 3>we have to remember that our own sun, which is

167
00:08:20.519 --> 00:08:24.519
<v Speaker 3>a solitary star floating alone in space, is actually somewhat

168
00:08:24.560 --> 00:08:27.759
<v Speaker 3>of an anomaly. Really yeah, A huge percentage of the

169
00:08:27.759 --> 00:08:31.199
<v Speaker 3>stars in our galaxy are binary systems, meaning two stars

170
00:08:31.279 --> 00:08:34.480
<v Speaker 3>locked in a gravitational dance orbiting around a common center of.

171
00:08:34.519 --> 00:08:37.519
<v Speaker 2>Mass like a tattooine situation to use the mandatory Sci

172
00:08:37.519 --> 00:08:40.360
<v Speaker 2>Fi reference two suns in the sky exactly that.

173
00:08:40.759 --> 00:08:43.480
<v Speaker 3>Now, imagine those two stars orbiting around each other from

174
00:08:43.480 --> 00:08:45.759
<v Speaker 3>our specific line of sight. If the orbital plane is

175
00:08:45.879 --> 00:08:49.039
<v Speaker 3>edge on to Earth, one star will periodically pass in

176
00:08:49.039 --> 00:08:52.000
<v Speaker 3>front of the other. Okay, I'm picturing it. Sometimes the

177
00:08:52.080 --> 00:08:56.240
<v Speaker 3>stars are of different sizes and temperatures. When a smaller, dimmer,

178
00:08:56.559 --> 00:09:00.639
<v Speaker 3>cooler star passes in front of a larger, brighter, hotter star,

179
00:09:01.279 --> 00:09:03.679
<v Speaker 3>it blocks a portion of the bright star's light.

180
00:09:03.679 --> 00:09:06.200
<v Speaker 2>Which creates a dimming effect in the overall light we

181
00:09:06.279 --> 00:09:07.399
<v Speaker 2>receive in our telescopes.

182
00:09:07.559 --> 00:09:12.120
<v Speaker 3>Right and mechanically, mathematically, that specific dip in light can

183
00:09:12.159 --> 00:09:15.679
<v Speaker 3>perfectly mimic the transit of a large Jupiter sized planet.

184
00:09:15.759 --> 00:09:18.879
<v Speaker 2>Let me make sure I'm visualizing this correctly. The telescope

185
00:09:18.960 --> 00:09:21.440
<v Speaker 2>is basically just a bucket collecting photons.

186
00:09:21.559 --> 00:09:22.559
<v Speaker 3>That's a good way to put it.

187
00:09:22.559 --> 00:09:24.440
<v Speaker 2>It doesn't see a picture of a star. It just

188
00:09:24.559 --> 00:09:27.080
<v Speaker 2>measures the amount of light hitting its sensor. So the

189
00:09:27.120 --> 00:09:31.000
<v Speaker 2>telescope sees the light dropped by say one percent, flags

190
00:09:31.000 --> 00:09:34.159
<v Speaker 2>it as a potential giant planet crossing the star, but

191
00:09:34.240 --> 00:09:37.519
<v Speaker 2>it's actually just a smaller, fainter star getting in the way.

192
00:09:37.679 --> 00:09:40.840
<v Speaker 3>Yes, it's a stellar eclipse, not a planetary transit.

193
00:09:41.039 --> 00:09:43.080
<v Speaker 2>Man. That is deeply frustrating.

194
00:09:43.159 --> 00:09:46.360
<v Speaker 3>That's the core of the false positive problem. And it

195
00:09:46.399 --> 00:09:49.480
<v Speaker 3>gets even more insidious. You have what are called grazing binaries.

196
00:09:49.679 --> 00:09:52.919
<v Speaker 3>Grazing binaries, Yeah, this is where the two stars aren't

197
00:09:52.919 --> 00:09:55.600
<v Speaker 3>perfectly aligned with us, but they just barely clip the

198
00:09:55.679 --> 00:09:57.080
<v Speaker 3>edge of each other's disc as they.

199
00:09:57.039 --> 00:09:58.440
<v Speaker 2>Orbit, just a glancing glow.

200
00:09:58.600 --> 00:10:02.879
<v Speaker 3>Exactly, that tiny little clip that grazing pass creates a

201
00:10:02.960 --> 00:10:06.279
<v Speaker 3>very shallow dip in the light curve, and a shallow

202
00:10:06.320 --> 00:10:09.159
<v Speaker 3>dip is the exact signature we expect from a small

203
00:10:09.559 --> 00:10:14.200
<v Speaker 3>Earth sized rocky planet. So a grazing binary star system

204
00:10:14.399 --> 00:10:19.039
<v Speaker 3>can totally masquerade as a potentially habitable Earth like world.

205
00:10:19.240 --> 00:10:22.080
<v Speaker 2>So how did astronomers handle this before the AI? I mean,

206
00:10:22.120 --> 00:10:25.080
<v Speaker 2>you have a list of thousands of these signals from tests.

207
00:10:25.559 --> 00:10:27.440
<v Speaker 2>How do you figure out if it's a moth or

208
00:10:27.519 --> 00:10:30.759
<v Speaker 2>just another slightly dimmer street light swinging in the wind?

209
00:10:31.120 --> 00:10:36.159
<v Speaker 3>Well? Distinguishing between a planetary transit and eclipsing binary historically

210
00:10:36.200 --> 00:10:39.960
<v Speaker 3>required intense, incredibly time consuming follow up observation.

211
00:10:39.679 --> 00:10:41.279
<v Speaker 2>And pointing other telescopes at it.

212
00:10:41.360 --> 00:10:43.480
<v Speaker 3>Exactly you couldn't just use tests anymore. You had to

213
00:10:43.559 --> 00:10:48.360
<v Speaker 3>requisition time on completely different, massive ground based telescopes on Earth.

214
00:10:48.399 --> 00:10:51.639
<v Speaker 3>Oh wow, You'd use instruments called spectrographs to analyze the

215
00:10:51.679 --> 00:10:55.120
<v Speaker 3>light spectrum of the star looking for minute gravitational wobbles.

216
00:10:55.159 --> 00:10:57.000
<v Speaker 3>It's a technique called radial velocity.

217
00:10:57.200 --> 00:10:58.399
<v Speaker 2>How does that tell them? Apart?

218
00:10:58.960 --> 00:11:01.679
<v Speaker 3>A planet will cause au star to wobble very slightly,

219
00:11:02.240 --> 00:11:05.639
<v Speaker 3>but a massive companion star will cause the primary star

220
00:11:06.000 --> 00:11:08.159
<v Speaker 3>to yank back and forth violently.

221
00:11:08.320 --> 00:11:11.799
<v Speaker 2>But getting time on those giant ground based observatories is

222
00:11:11.960 --> 00:11:15.720
<v Speaker 2>notoriously competitive. You can't just point the Keck telescope in

223
00:11:15.720 --> 00:11:19.360
<v Speaker 2>Hawaii at two thousand different stars just to double check

224
00:11:19.440 --> 00:11:20.240
<v Speaker 2>Tess's homework.

225
00:11:20.519 --> 00:11:23.279
<v Speaker 3>No, it would take decades and cost millions of dollars.

226
00:11:23.000 --> 00:11:26.480
<v Speaker 2>Which is precisely why we have a massive backlog of

227
00:11:26.679 --> 00:11:30.200
<v Speaker 2>unconfirmed candidates just sitting on servers waiting to be vetted.

228
00:11:30.440 --> 00:11:34.279
<v Speaker 3>Exactly. The traditional scientific process simply could not keep pace

229
00:11:34.360 --> 00:11:36.879
<v Speaker 3>with the sheer volume of data the satellite was beaming

230
00:11:36.919 --> 00:11:40.039
<v Speaker 3>down to Earth every month. We were drowning in potential discoveries,

231
00:11:40.200 --> 00:11:42.039
<v Speaker 3>completely paralyzed by the false.

232
00:11:41.840 --> 00:11:45.720
<v Speaker 2>Positive enter raven. Yeah, this is the newly developed AI

233
00:11:45.759 --> 00:11:48.480
<v Speaker 2>pipeline from the team at the University of Warwick that

234
00:11:48.600 --> 00:11:51.840
<v Speaker 2>is cutting through the backlog like a Scythe and raven

235
00:11:51.919 --> 00:11:55.600
<v Speaker 2>stands for a ranking and validation of exoplanets. I always

236
00:11:55.600 --> 00:11:57.519
<v Speaker 2>appreciate a backronym in astronomy.

237
00:11:57.559 --> 00:11:58.519
<v Speaker 3>They are very good at them.

238
00:11:58.639 --> 00:12:02.320
<v Speaker 2>They really are its core function. Its entire reason for

239
00:12:02.399 --> 00:12:06.960
<v Speaker 2>existing is to solve this exact bottleneck. It is a

240
00:12:07.000 --> 00:12:11.759
<v Speaker 2>bespoke software architecture designed to distinguish between actual planetary transits

241
00:12:12.200 --> 00:12:17.200
<v Speaker 2>and these astrophysical impostures without needing years of expensive ground

242
00:12:17.240 --> 00:12:18.879
<v Speaker 2>based telescope.

243
00:12:18.240 --> 00:12:21.399
<v Speaker 3>Time and the way to choose this is an absolute

244
00:12:21.559 --> 00:12:24.600
<v Speaker 3>master class in modern computational astrophysics.

245
00:12:24.720 --> 00:12:25.919
<v Speaker 2>How does it actually work.

246
00:12:26.200 --> 00:12:28.080
<v Speaker 3>It doesn't just look at the overall depth of the

247
00:12:28.080 --> 00:12:31.440
<v Speaker 3>shadow and make a guess. It uses deep machine learning,

248
00:12:31.639 --> 00:12:36.679
<v Speaker 3>specifically advanced neural network architectures to analyze the entire shape.

249
00:12:36.440 --> 00:12:38.440
<v Speaker 2>Of the light curve, the shape of the curve itself.

250
00:12:38.559 --> 00:12:40.679
<v Speaker 3>Right. It looks at the duration of the dip, the

251
00:12:40.679 --> 00:12:43.879
<v Speaker 3>speed at which the light drops the tiny, almost imperceptible

252
00:12:43.960 --> 00:12:46.600
<v Speaker 3>variations at the edges of the shadow as the planet

253
00:12:46.639 --> 00:12:51.120
<v Speaker 3>begins its crossing. It searches for the subtle mathematical signatures

254
00:12:51.360 --> 00:12:55.159
<v Speaker 3>that separate a solid spherical planet from a glowing dynamic star.

255
00:12:55.279 --> 00:12:57.559
<v Speaker 2>Okay, let's unpack this methodology, because this is where I

256
00:12:57.600 --> 00:12:59.720
<v Speaker 2>have to push back a little on how AI usually

257
00:12:59.720 --> 00:13:02.960
<v Speaker 2>operate rates. When I was diving into the research papers

258
00:13:03.039 --> 00:13:06.080
<v Speaker 2>detailing how they built Raven, I noticed something that sounded

259
00:13:06.120 --> 00:13:10.799
<v Speaker 2>completely counterintuitive, almost paradoxical. The training data exactly. Usually, when

260
00:13:10.840 --> 00:13:13.720
<v Speaker 2>you train an AI, you feed it real world examples.

261
00:13:14.080 --> 00:13:16.000
<v Speaker 2>If you want an AI to recognize a cat, you

262
00:13:16.080 --> 00:13:19.519
<v Speaker 2>show it a million photographs of actual cats. Right, But

263
00:13:19.600 --> 00:13:22.879
<v Speaker 2>the Warwick researchers didn't train this AI on the thousands

264
00:13:22.919 --> 00:13:26.080
<v Speaker 2>of real exoplanets we've already discovered and confirmed over the

265
00:13:26.159 --> 00:13:30.240
<v Speaker 2>last twenty years. Instead, they trained Raven on hundreds of

266
00:13:30.320 --> 00:13:34.440
<v Speaker 2>thousands of realistically simulated planets and simulated stellar events.

267
00:13:34.519 --> 00:13:35.080
<v Speaker 3>Yes they did.

268
00:13:35.279 --> 00:13:37.240
<v Speaker 2>I have to admit I was scratching my head. We

269
00:13:37.320 --> 00:13:41.039
<v Speaker 2>are training an AI on fake computer generated planets so

270
00:13:41.080 --> 00:13:44.399
<v Speaker 2>it can find real ones. How does learning from a

271
00:13:44.440 --> 00:13:47.480
<v Speaker 2>massive video game simulation of the universe give us an

272
00:13:47.600 --> 00:13:51.200
<v Speaker 2>edge over just studying the real, messy data we already

273
00:13:51.200 --> 00:13:52.759
<v Speaker 2>have from Kepler and Tess.

274
00:13:53.159 --> 00:13:55.840
<v Speaker 3>What's fascinating here is that your hesitation hits on the

275
00:13:55.960 --> 00:13:59.200
<v Speaker 3>absolute most crucial debate in machine learning. Right now. Your

276
00:13:59.200 --> 00:14:02.320
<v Speaker 3>confusion is exactly why this approach is so brilliant and

277
00:14:02.360 --> 00:14:04.080
<v Speaker 3>why it represents such a leap forward.

278
00:14:04.120 --> 00:14:06.440
<v Speaker 2>Okay, explain that to me, it all comes down.

279
00:14:06.320 --> 00:14:10.440
<v Speaker 3>To how neural networks actually learn. Machine learning models excel

280
00:14:10.480 --> 00:14:14.720
<v Speaker 3>at pattern recognition. They are vastly superior to humans at

281
00:14:14.759 --> 00:14:19.039
<v Speaker 3>finding hidden correlations in high dimensional data. But to learn

282
00:14:19.080 --> 00:14:23.679
<v Speaker 3>those patterns correctly, they need training data that is perfectly categorized.

283
00:14:24.200 --> 00:14:26.759
<v Speaker 3>They need to know, without a shadow of a doubt,

284
00:14:27.120 --> 00:14:27.879
<v Speaker 3>what the right.

285
00:14:27.759 --> 00:14:30.960
<v Speaker 2>Answer is, So they need an infallible answer key like

286
00:14:31.320 --> 00:14:34.000
<v Speaker 2>if you are teaching a child math, you can't give

287
00:14:34.039 --> 00:14:36.320
<v Speaker 2>them a textbook where half the answers in the back

288
00:14:36.360 --> 00:14:39.480
<v Speaker 2>are wrong or you know, probably right, but we aren't

289
00:14:39.480 --> 00:14:40.720
<v Speaker 2>totally sure exactly.

290
00:14:41.159 --> 00:14:44.080
<v Speaker 3>In computer science we call this the ground truth. If

291
00:14:44.080 --> 00:14:46.919
<v Speaker 3>we were to train an AI exclusively on real data

292
00:14:46.960 --> 00:14:50.519
<v Speaker 3>from tests or even our catalog of previously confirmed planets,

293
00:14:50.840 --> 00:14:53.120
<v Speaker 3>we are fundamentally training it on data that is full

294
00:14:53.159 --> 00:14:58.120
<v Speaker 3>of noise, uncorrected instrument artifacts, and most importantly, unknown.

295
00:14:57.759 --> 00:15:00.840
<v Speaker 2>Variables, even for the ones we think or confirmed, even.

296
00:15:00.639 --> 00:15:03.559
<v Speaker 3>Those even with planets we think we've confidently confirmed using

297
00:15:03.600 --> 00:15:06.679
<v Speaker 3>ground based telescopes, there's always a margin of error. There's

298
00:15:06.720 --> 00:15:09.159
<v Speaker 3>always a tiny, nagging percentage of doubt.

299
00:15:09.120 --> 00:15:09.639
<v Speaker 2>Kind of doubt.

300
00:15:09.759 --> 00:15:12.720
<v Speaker 3>Well, what if a faint background star is blending its

301
00:15:12.759 --> 00:15:16.559
<v Speaker 3>light with our target, subtly altering the signal? What if

302
00:15:16.600 --> 00:15:19.840
<v Speaker 3>our understanding of the star's mass is slightly off?

303
00:15:20.639 --> 00:15:23.120
<v Speaker 2>Ah, I see where this is going. If you train

304
00:15:23.200 --> 00:15:28.200
<v Speaker 2>an AI on flawed, messy human data, it just inherits

305
00:15:28.240 --> 00:15:31.440
<v Speaker 2>all of our own blind spots. Precisely, If the astronomical

306
00:15:31.480 --> 00:15:35.840
<v Speaker 2>community has been systematically misclassifying a certain type of weird

307
00:15:36.240 --> 00:15:39.120
<v Speaker 2>grazing binary star as a rocky planet for the last

308
00:15:39.120 --> 00:15:42.120
<v Speaker 2>ten years, and we feed that into the AI, the

309
00:15:42.200 --> 00:15:44.840
<v Speaker 2>AI will just learn to make that exact same mistake,

310
00:15:45.240 --> 00:15:47.679
<v Speaker 2>only it will make that mistake a million times faster

311
00:15:47.879 --> 00:15:49.600
<v Speaker 2>and with terrifying confidence.

312
00:15:49.679 --> 00:15:52.320
<v Speaker 3>And that is the absolute danger of black box AI

313
00:15:52.480 --> 00:15:56.960
<v Speaker 3>in science. It amplifies human error. But by generating simulated events,

314
00:15:57.039 --> 00:15:59.559
<v Speaker 3>the researchers hold the ultimate perfect answer.

315
00:15:59.360 --> 00:16:01.559
<v Speaker 2>Key Because as they built the simulation.

316
00:16:01.399 --> 00:16:05.000
<v Speaker 3>Right, they built incredibly complex mathematical models of stars, They

317
00:16:05.039 --> 00:16:09.200
<v Speaker 3>generated artificial planetary transits across them, calculating the exact geometry

318
00:16:09.200 --> 00:16:12.440
<v Speaker 3>of the shadows. They added synthetic telescope noise that perfectly

319
00:16:12.480 --> 00:16:15.240
<v Speaker 3>mimics the thermal shifts and pixel bleed of the test cameras.

320
00:16:15.320 --> 00:16:16.840
<v Speaker 2>That's incredibly thorough.

321
00:16:16.639 --> 00:16:22.080
<v Speaker 3>And critically they generated artificial eclipsing binary stars and grazing binaries.

322
00:16:22.600 --> 00:16:26.120
<v Speaker 3>Because the researchers created the simulation from the ground up,

323
00:16:26.559 --> 00:16:30.120
<v Speaker 3>writing the laws of physics in code, they know the

324
00:16:30.159 --> 00:16:34.559
<v Speaker 3>absolute ground truth of every single pixel of data.

325
00:16:34.639 --> 00:16:37.799
<v Speaker 2>They know with one hundred percent mathematical certainty, whether a

326
00:16:37.840 --> 00:16:40.960
<v Speaker 2>specific simulated light curve was caused by a fake planet

327
00:16:41.279 --> 00:16:44.279
<v Speaker 2>or a fake binary star exactly. And it makes total

328
00:16:44.320 --> 00:16:46.039
<v Speaker 2>sense when you frame it like that, you aren't just

329
00:16:46.039 --> 00:16:49.120
<v Speaker 2>showing it pictures of shadows. You are teaching the AI

330
00:16:49.320 --> 00:16:53.879
<v Speaker 2>the exact, pure mathematical signature of a planet, completely divorced

331
00:16:53.919 --> 00:16:56.759
<v Speaker 2>from human error and observational bias. Yes, you're giving it

332
00:16:56.799 --> 00:16:59.679
<v Speaker 2>the fundamental physics of an orbital transit isolated in a

333
00:16:59.720 --> 00:17:01.399
<v Speaker 2>stere laboratory environment.

334
00:17:01.639 --> 00:17:04.559
<v Speaker 3>And by simulating hundreds of thousands of these events, they

335
00:17:04.640 --> 00:17:07.680
<v Speaker 3>can expose the AI to scenarios that might be incredibly

336
00:17:07.759 --> 00:17:11.440
<v Speaker 3>rare in our current real world catalogs, but theoretically possible

337
00:17:11.480 --> 00:17:14.440
<v Speaker 3>in the wider universe, Like what well they can train

338
00:17:14.519 --> 00:17:16.640
<v Speaker 3>the AI on what a planet looks like orbiting a

339
00:17:16.720 --> 00:17:20.240
<v Speaker 3>highly active star covered in massive sunspots, or what a

340
00:17:20.279 --> 00:17:22.440
<v Speaker 3>planet theory transit looks like if the planet has a

341
00:17:22.440 --> 00:17:23.960
<v Speaker 3>massive ring system like Saturn.

342
00:17:24.319 --> 00:17:26.039
<v Speaker 2>Oh, that's so smart, right.

343
00:17:26.200 --> 00:17:29.000
<v Speaker 3>Once the neural network has completely mastered the physics of

344
00:17:29.000 --> 00:17:32.920
<v Speaker 3>the simulation, once it can perfectly distinguish the fake planets

345
00:17:32.920 --> 00:17:36.720
<v Speaker 3>from the fake stars, then and only then is it

346
00:17:36.799 --> 00:17:39.759
<v Speaker 3>unleashed on the real, messy tests data.

347
00:17:39.839 --> 00:17:42.799
<v Speaker 2>And because it has internalized the underlying signature of a

348
00:17:42.839 --> 00:17:45.880
<v Speaker 2>true planet so deeply, it can cut through the noise

349
00:17:46.000 --> 00:17:49.960
<v Speaker 2>with incredible precision. It knows what a moth's shadow looks like,

350
00:17:50.279 --> 00:17:52.440
<v Speaker 2>even if the street light is flickering, and even it

351
00:17:52.440 --> 00:17:54.000
<v Speaker 2>fits raining precisely.

352
00:17:54.039 --> 00:17:55.880
<v Speaker 3>And it doesn't just give a yes or no answer.

353
00:17:56.200 --> 00:17:58.400
<v Speaker 3>Raven provides a statistical probability.

354
00:17:58.440 --> 00:18:00.200
<v Speaker 2>Oh, it gives a confidence score exactly.

355
00:18:00.720 --> 00:18:03.160
<v Speaker 3>It looks at a candidate and says, based on my training,

356
00:18:03.319 --> 00:18:05.680
<v Speaker 3>there is a ninety nine point nine percent chance this

357
00:18:05.759 --> 00:18:08.599
<v Speaker 3>is a planetary transit, and only one point one percent

358
00:18:08.680 --> 00:18:12.119
<v Speaker 3>chance it is a background eclipsing binary. When that probability

359
00:18:12.119 --> 00:18:15.559
<v Speaker 3>crosses a certain rigorous threshold, the candidate is upgraded to

360
00:18:15.599 --> 00:18:16.720
<v Speaker 3>a validated planet.

361
00:18:16.960 --> 00:18:19.039
<v Speaker 2>And the other thing that makes Raven such a paradigm

362
00:18:19.119 --> 00:18:21.960
<v Speaker 2>shift is the workflow advantage. I was reading about how

363
00:18:22.000 --> 00:18:24.000
<v Speaker 2>this used to be done and in the past, confirming

364
00:18:24.039 --> 00:18:28.160
<v Speaker 2>a planet was this incredibly fragmented, piecemeal process oosen Nimare.

365
00:18:28.279 --> 00:18:30.480
<v Speaker 2>You'd have one astronomer write a piece of software to

366
00:18:30.480 --> 00:18:33.319
<v Speaker 2>clean the raw data. Then you'd pass that data to

367
00:18:33.359 --> 00:18:37.039
<v Speaker 2>another team using a different algorithm to search for periodic dips.

368
00:18:37.480 --> 00:18:40.400
<v Speaker 2>Then a human might literally eyeball the resulting.

369
00:18:40.039 --> 00:18:42.720
<v Speaker 3>Graphs yep, lots of staring at graphs.

370
00:18:42.799 --> 00:18:45.599
<v Speaker 2>Then if it looks promising, you'd feed it into a

371
00:18:45.599 --> 00:18:49.279
<v Speaker 2>completely different statistical tool to calculate the probabilities of a

372
00:18:49.279 --> 00:18:50.160
<v Speaker 2>false positive.

373
00:18:50.359 --> 00:18:54.240
<v Speaker 3>It was a very patchwork, fragmented ecosystem of tools. Every

374
00:18:54.319 --> 00:18:58.480
<v Speaker 3>research group had their own preferred software packages written in

375
00:18:58.519 --> 00:18:59.680
<v Speaker 3>different programming languages.

376
00:19:00.000 --> 00:19:01.200
<v Speaker 2>That was incredibly inefficient.

377
00:19:01.440 --> 00:19:04.599
<v Speaker 3>It inherently slows down the science, makes it difficult to

378
00:19:04.640 --> 00:19:09.200
<v Speaker 3>reproduce results, and introduces opportunities for errors or data corruption

379
00:19:09.480 --> 00:19:14.039
<v Speaker 3>at every single handoff point. It was well artisanal planet hunting, but.

380
00:19:14.079 --> 00:19:16.920
<v Speaker 2>Raven handles the entire process end to end. It's a

381
00:19:17.000 --> 00:19:18.759
<v Speaker 2>unified ecosystem.

382
00:19:18.200 --> 00:19:19.160
<v Speaker 3>From start to finish.

383
00:19:19.400 --> 00:19:23.319
<v Speaker 2>It takes the raw, messy, noisy data straight from the satellite,

384
00:19:23.559 --> 00:19:28.160
<v Speaker 2>processes the light curves, detects the potential periodic signals, vets

385
00:19:28.200 --> 00:19:31.119
<v Speaker 2>those signals using its highly trained machine learning models to

386
00:19:31.160 --> 00:19:35.079
<v Speaker 2>weed out the astrophysical imposters, and then statistically validates the

387
00:19:35.079 --> 00:19:38.960
<v Speaker 2>surviving planets, all in one seamless automated pipeline.

388
00:19:39.000 --> 00:19:40.079
<v Speaker 3>That's the beauty of it.

389
00:19:40.079 --> 00:19:43.279
<v Speaker 2>It takes the grueling, repetitive heavy lifting out of the

390
00:19:43.359 --> 00:19:47.519
<v Speaker 2>equation and just hands the astronomers a highly reliable, mathematically

391
00:19:47.599 --> 00:19:48.839
<v Speaker 2>sound list of worlds.

392
00:19:48.880 --> 00:19:52.720
<v Speaker 3>It's the difference between a master mechanic trying to build

393
00:19:52.720 --> 00:19:55.359
<v Speaker 3>a car from individual parts they ordered from a dozen

394
00:19:55.400 --> 00:19:58.759
<v Speaker 3>different catalogs versus having a fully automated, state of the

395
00:19:58.839 --> 00:20:03.119
<v Speaker 3>art robotic assembly lie perfect analogy. The efficiency and consistency

396
00:20:03.160 --> 00:20:05.039
<v Speaker 3>are what allows us to look at two point two

397
00:20:05.119 --> 00:20:08.079
<v Speaker 3>million stars and actually makes sense of him in the

398
00:20:08.160 --> 00:20:08.880
<v Speaker 3>human lifetime.

399
00:20:08.960 --> 00:20:11.279
<v Speaker 2>So now that we understand how this AI works and

400
00:20:11.319 --> 00:20:13.920
<v Speaker 2>why training it on a massive simulation was the secret

401
00:20:14.000 --> 00:20:16.079
<v Speaker 2>to its success, we need to look at what it

402
00:20:16.200 --> 00:20:18.519
<v Speaker 2>actually found when they turned it on. The fun part,

403
00:20:18.759 --> 00:20:21.519
<v Speaker 2>because the AI didn't just find a bunch of generic,

404
00:20:21.680 --> 00:20:25.440
<v Speaker 2>boring dead rocks floating in empty space. It uncovered some

405
00:20:25.480 --> 00:20:30.119
<v Speaker 2>of the most extreme, chaotic and alien worlds imaginable. Hidden

406
00:20:30.200 --> 00:20:33.759
<v Speaker 2>in that test data, the thirty one newly discovered planets

407
00:20:33.799 --> 00:20:36.519
<v Speaker 2>are not places you would ever want to visit. Let's

408
00:20:36.519 --> 00:20:38.519
<v Speaker 2>talk about the exoplanet menagerie.

409
00:20:38.720 --> 00:20:41.920
<v Speaker 3>The diversity of planetary systems out there is truly staggering.

410
00:20:42.240 --> 00:20:44.799
<v Speaker 3>Our own Solar system, with its neat division of inner

411
00:20:44.920 --> 00:20:48.279
<v Speaker 3>rocky planets and outer gas giants, is just one tiny,

412
00:20:48.359 --> 00:20:52.960
<v Speaker 3>specific flavor in a cosmic ice cream shop of bizarre configurations.

413
00:20:53.039 --> 00:20:55.079
<v Speaker 2>And it's important to note the specific focus of the

414
00:20:55.119 --> 00:20:58.519
<v Speaker 2>Warwick team's study to understand what Raven was finding. The

415
00:20:58.559 --> 00:21:01.279
<v Speaker 2>pipeline was specifically off demise to look for planets with

416
00:21:01.400 --> 00:21:04.599
<v Speaker 2>very tight orbits. We are talking about planets that complete

417
00:21:04.640 --> 00:21:07.559
<v Speaker 2>a full orbit around their star a full planetary year

418
00:21:07.599 --> 00:21:09.039
<v Speaker 2>in less than sixteen days.

419
00:21:09.079 --> 00:21:11.880
<v Speaker 3>Sixteen days. Earth takes three hundred and sixty five days

420
00:21:11.920 --> 00:21:14.640
<v Speaker 3>to make that trip. Mercury, the closest planet to our Sun,

421
00:21:14.799 --> 00:21:15.559
<v Speaker 3>takes eighty.

422
00:21:15.319 --> 00:21:17.839
<v Speaker 2>Eight days, So a sixteen day year means these planets

423
00:21:17.880 --> 00:21:21.559
<v Speaker 2>are hugging their host stars incredibly uncomfortably closely. They are

424
00:21:21.559 --> 00:21:23.000
<v Speaker 2>practically skimming the surface.

425
00:21:23.279 --> 00:21:27.200
<v Speaker 3>They are deeply embedded in the intense gravitational and radiational

426
00:21:27.279 --> 00:21:30.480
<v Speaker 3>environment of their stars. To give you a sense of scale,

427
00:21:30.839 --> 00:21:33.599
<v Speaker 3>if Earth were on a sixteen day orbit, the Sun

428
00:21:33.640 --> 00:21:36.160
<v Speaker 3>would take up a massive portion of the sky, the

429
00:21:36.240 --> 00:21:37.559
<v Speaker 3>heat would be unimaginable.

430
00:21:37.680 --> 00:21:42.279
<v Speaker 2>And within that already extreme sixteen day parameter Raven managed

431
00:21:42.319 --> 00:21:45.480
<v Speaker 2>to detect a population of what astronomers classify as ultra

432
00:21:45.559 --> 00:21:47.720
<v Speaker 2>short period planets, or usps.

433
00:21:48.119 --> 00:21:50.039
<v Speaker 3>These are worlds that take the concept of a tight

434
00:21:50.160 --> 00:21:52.000
<v Speaker 3>orbit to an absurd extreme.

435
00:21:52.160 --> 00:21:54.480
<v Speaker 2>These are planets that orbit their stars in less than

436
00:21:54.480 --> 00:21:55.559
<v Speaker 2>twenty four hours.

437
00:21:55.319 --> 00:21:57.200
<v Speaker 3>Less than an Earth day to complete a full year.

438
00:21:57.359 --> 00:21:59.519
<v Speaker 2>I want to try and visualize that reality for a second.

439
00:22:00.039 --> 00:22:03.240
<v Speaker 2>Jugene celebrating New Year's Eve. Every single day you wake up,

440
00:22:03.240 --> 00:22:05.119
<v Speaker 2>it's a new year. You go to sleep, the year

441
00:22:05.200 --> 00:22:09.720
<v Speaker 2>is over. Your entire planetary cycle season orbital dynamics is

442
00:22:09.720 --> 00:22:11.680
<v Speaker 2>compressed into a single Earth day.

443
00:22:11.799 --> 00:22:12.519
<v Speaker 3>It's frantic.

444
00:22:12.759 --> 00:22:16.200
<v Speaker 2>A planet moving that incredibly fast, orbiting that perilously close

445
00:22:16.240 --> 00:22:19.720
<v Speaker 2>to its host star is experiencing an environment that completely

446
00:22:19.759 --> 00:22:23.400
<v Speaker 2>breaks the human imagination. At that proximity, the gravitational forces

447
00:22:23.440 --> 00:22:26.960
<v Speaker 2>are immense. The gravity of the star almost certainly locks

448
00:22:27.000 --> 00:22:29.759
<v Speaker 2>the planet in place. A phenomenon we call tidal locking.

449
00:22:30.599 --> 00:22:32.759
<v Speaker 2>It's the same physics that keeps one side of our

450
00:22:32.799 --> 00:22:36.200
<v Speaker 2>Moon always facing Earth right, So for these ultra short

451
00:22:36.240 --> 00:22:39.279
<v Speaker 2>period planets, one side of the planet always faces the

452
00:22:39.359 --> 00:22:41.920
<v Speaker 2>nuclear furnace of the Sun and the other side is

453
00:22:41.960 --> 00:22:43.960
<v Speaker 2>cast in eternal freezing darkness.

454
00:22:44.160 --> 00:22:46.920
<v Speaker 3>And the dayside of a tidally locked ultra short period

455
00:22:46.920 --> 00:22:49.519
<v Speaker 3>planet isn't just hot in the way we can comprehend.

456
00:22:50.000 --> 00:22:54.119
<v Speaker 3>It is a literal physical hellscape, like how hot. Because

457
00:22:54.160 --> 00:22:56.920
<v Speaker 3>they are so close, the surface temperatures on the star

458
00:22:56.960 --> 00:23:00.480
<v Speaker 3>facing side can easily reach several thousands of degrees. At

459
00:23:00.480 --> 00:23:04.559
<v Speaker 3>those temperatures, rock doesn't just melt into magma. It vaporizes

460
00:23:04.839 --> 00:23:05.720
<v Speaker 3>vaporized rock.

461
00:23:05.920 --> 00:23:06.200
<v Speaker 2>Wow.

462
00:23:06.319 --> 00:23:09.559
<v Speaker 3>Yeah, the extreme heat breaks down molecular bonds. You are

463
00:23:09.640 --> 00:23:13.720
<v Speaker 3>likely looking at unimaginable landscapes dominated by global churning oceans

464
00:23:13.759 --> 00:23:16.759
<v Speaker 3>of molten lava, and the atmosphere, if the stellar winds

465
00:23:16.759 --> 00:23:19.720
<v Speaker 3>haven't entirely stripped away, would be composed of vaporized rock

466
00:23:19.759 --> 00:23:20.519
<v Speaker 3>and heavy metals.

467
00:23:20.599 --> 00:23:23.000
<v Speaker 2>So we are talking about planets where the weather forecast

468
00:23:23.079 --> 00:23:26.480
<v Speaker 2>might literally call for scattered showers of liquid iron or

469
00:23:26.559 --> 00:23:28.279
<v Speaker 2>condensation of vaporized glass.

470
00:23:28.319 --> 00:23:28.720
<v Speaker 3>Pretty much.

471
00:23:28.799 --> 00:23:31.480
<v Speaker 2>Yeah, and thanks to this AI pipeline. These aren't just

472
00:23:31.599 --> 00:23:36.240
<v Speaker 2>theoretical mathematical models on a whiteboard anymore. Raven is casually

473
00:23:36.319 --> 00:23:39.720
<v Speaker 2>validating dozens of these nightmare worlds in a single sweep

474
00:23:39.960 --> 00:23:41.000
<v Speaker 2>of the test data.

475
00:23:41.480 --> 00:23:44.359
<v Speaker 3>It's finding them so efficiently that we can start studying

476
00:23:44.400 --> 00:23:48.160
<v Speaker 3>them as a broader population rather than just isolated freaks

477
00:23:48.160 --> 00:23:49.039
<v Speaker 3>of nature.

478
00:23:48.759 --> 00:23:51.359
<v Speaker 2>And studying them tells us so much about the violent

479
00:23:51.440 --> 00:23:55.039
<v Speaker 2>history of solar systems. Plants don't generally form that close

480
00:23:55.079 --> 00:23:56.319
<v Speaker 2>to a star, do they know?

481
00:23:56.759 --> 00:23:59.480
<v Speaker 3>The environment is way too hot and volatile for dust

482
00:23:59.519 --> 00:24:01.799
<v Speaker 3>and gas to clump together and form a planet.

483
00:24:01.920 --> 00:24:03.000
<v Speaker 2>So how did they get there?

484
00:24:03.160 --> 00:24:07.640
<v Speaker 3>Well, which means these ultra short period worlds likely formed

485
00:24:07.720 --> 00:24:10.359
<v Speaker 3>much further out in their solar systems where it was cooler,

486
00:24:10.920 --> 00:24:14.839
<v Speaker 3>and then, over millions of years, gravitational interactions with other

487
00:24:14.880 --> 00:24:18.440
<v Speaker 3>planets dragged them inward. They spiraled closer and closer to

488
00:24:18.480 --> 00:24:21.759
<v Speaker 3>the star until they parked in these extreme roasting orbits.

489
00:24:22.000 --> 00:24:24.160
<v Speaker 2>It's the kind of extreme physics that used to belong

490
00:24:24.319 --> 00:24:26.480
<v Speaker 2>entirely to the realm of science fiction, and now it's

491
00:24:26.640 --> 00:24:29.519
<v Speaker 2>just a Tuesday for an AI algorithm at Warwick. But

492
00:24:29.599 --> 00:24:33.000
<v Speaker 2>the extreme environments Raven found aren't just limited to these

493
00:24:33.119 --> 00:24:37.720
<v Speaker 2>solitary roosting rocks. The pipeline also proved incredibly adept at

494
00:24:37.720 --> 00:24:41.960
<v Speaker 2>detecting close orbiting multiplanet systems. We are talking about previously

495
00:24:42.119 --> 00:24:46.319
<v Speaker 2>unknown planetary pairs, or even entire groups of planets orbiting

496
00:24:46.359 --> 00:24:50.200
<v Speaker 2>the same star, all packed into incredibly tight spaces.

497
00:24:50.480 --> 00:24:53.880
<v Speaker 3>This is where orbital mechanics gets incredibly complex, and where

498
00:24:53.920 --> 00:24:56.839
<v Speaker 3>the AI truly flexes its computational muscle.

499
00:24:56.920 --> 00:24:57.880
<v Speaker 2>How tired are we talking?

500
00:24:58.160 --> 00:25:00.799
<v Speaker 3>Well? When we think of a Solar system picture, our

501
00:25:00.839 --> 00:25:04.759
<v Speaker 3>own planets spread out over vast distances, neatly separated by

502
00:25:04.839 --> 00:25:08.920
<v Speaker 3>millions of miles of empty space, Mercury than Venus than Earth,

503
00:25:09.279 --> 00:25:12.279
<v Speaker 3>all keeping a respectful distance, right, But the universe has

504
00:25:12.319 --> 00:25:15.559
<v Speaker 3>a way of packing planets together in configurations that completely

505
00:25:15.640 --> 00:25:18.160
<v Speaker 3>challenge our traditional models of planetary formation.

506
00:25:18.640 --> 00:25:22.319
<v Speaker 2>A perfect specific example of this, to help visualize exactly

507
00:25:22.359 --> 00:25:25.079
<v Speaker 2>the kind of chaos Raven is trained to untangle, is

508
00:25:25.119 --> 00:25:29.480
<v Speaker 2>the Kepler eleven system. Now, Kepler eleven was discovered before Raven,

509
00:25:29.559 --> 00:25:31.640
<v Speaker 2>but it perfectly illustrates the phenomenons.

510
00:25:31.680 --> 00:25:32.359
<v Speaker 3>A great example.

511
00:25:32.519 --> 00:25:35.680
<v Speaker 2>Kepler eleven is a sun like star relatively similar to

512
00:25:35.720 --> 00:25:38.920
<v Speaker 2>our own, but it has six confirmed planets orbiting it,

513
00:25:39.519 --> 00:25:42.720
<v Speaker 2>and those planets aren't spread out elegantly like ours. Five

514
00:25:42.759 --> 00:25:45.319
<v Speaker 2>of the six planets and Kepler eleven orbit closer to

515
00:25:45.400 --> 00:25:48.960
<v Speaker 2>their star than Mercury orbits our Sun. It's wild to

516
00:25:49.000 --> 00:25:54.079
<v Speaker 2>think about imagine taking Venus, Earth, Mars, Jupiter, and Saturn

517
00:25:54.480 --> 00:25:57.000
<v Speaker 2>and crushing all of their orbits down so they fit

518
00:25:57.079 --> 00:26:01.200
<v Speaker 2>inside the orbit of Mercury. It is an unbelo believably crowded,

519
00:26:01.359 --> 00:26:02.799
<v Speaker 2>claustrophobic neighborhood.

520
00:26:02.920 --> 00:26:07.039
<v Speaker 3>And from our observational perspective, watching the light curve of

521
00:26:07.079 --> 00:26:09.799
<v Speaker 3>a crowded system like that is both chaotic and.

522
00:26:09.720 --> 00:26:12.079
<v Speaker 2>Beautiful because they're all blocking the light right.

523
00:26:12.319 --> 00:26:14.160
<v Speaker 3>Because the orbits are so tight and the planets are

524
00:26:14.160 --> 00:26:17.119
<v Speaker 3>moving so fast, you don't just see one simple neat

525
00:26:17.279 --> 00:26:20.559
<v Speaker 3>planetary transit at a time. You regularly have situations where

526
00:26:20.599 --> 00:26:22.680
<v Speaker 3>two or even three planets are passing in front of

527
00:26:22.720 --> 00:26:24.880
<v Speaker 3>the host stars simultaneously.

528
00:26:24.119 --> 00:26:26.559
<v Speaker 2>Which compounds the starlight dimming. You get a shadow within

529
00:26:26.599 --> 00:26:27.039
<v Speaker 2>a shadow.

530
00:26:27.119 --> 00:26:30.799
<v Speaker 3>Precisely, the light drops when Planet A enters the disk. Then,

531
00:26:30.880 --> 00:26:33.759
<v Speaker 3>while Planet A is still crossing, Planet B enters and

532
00:26:33.799 --> 00:26:34.960
<v Speaker 3>the light drops further.

533
00:26:34.920 --> 00:26:35.759
<v Speaker 2>Like stair steps.

534
00:26:36.000 --> 00:26:39.519
<v Speaker 3>Exactly, Then Planet A leaves and the light goes back

535
00:26:39.599 --> 00:26:42.119
<v Speaker 3>up a bit, but Planet B is still there. The

536
00:26:42.160 --> 00:26:46.839
<v Speaker 3>shadow deepens and lightens in these complex overlapping asymmetrical steps,

537
00:26:47.640 --> 00:26:50.839
<v Speaker 3>untangling a light curve like that. Trying to isolate the

538
00:26:50.960 --> 00:26:55.079
<v Speaker 3>distinct mathematical shadow of three different planets overlapping each other

539
00:26:55.599 --> 00:26:58.599
<v Speaker 3>while accounting for the gravitational tugs they exert on each

540
00:26:58.599 --> 00:27:02.400
<v Speaker 3>other that slightly alter their transit times is a monumental

541
00:27:02.440 --> 00:27:03.400
<v Speaker 3>mathematical puzzle.

542
00:27:03.640 --> 00:27:06.799
<v Speaker 2>It's like listening to a massive symphony orchestra playing a

543
00:27:06.839 --> 00:27:10.160
<v Speaker 2>complex piece of music and being asked to perfectly isolate

544
00:27:10.200 --> 00:27:13.400
<v Speaker 2>and transcribe the notes of three specific violins playing at

545
00:27:13.400 --> 00:27:16.599
<v Speaker 2>the exact same time while the drummer is occasionally hitting

546
00:27:16.680 --> 00:27:17.680
<v Speaker 2>the microphone stand.

547
00:27:18.039 --> 00:27:20.680
<v Speaker 3>That is a very accurate way to describe the noise.

548
00:27:20.599 --> 00:27:24.039
<v Speaker 2>A human brain struggles to separate those signals. But it's

549
00:27:24.079 --> 00:27:27.680
<v Speaker 2>exactly the kind of complex overlapping pattern recognition that a

550
00:27:27.680 --> 00:27:30.640
<v Speaker 2>well trained neural network like Raven is built to decode.

551
00:27:31.079 --> 00:27:33.680
<v Speaker 2>It looks at the tangled mess of shadows and instantly

552
00:27:33.720 --> 00:27:35.359
<v Speaker 2>identifies the individual players.

553
00:27:35.759 --> 00:27:40.559
<v Speaker 3>And discovering these compact multiplanet systems is vital because they

554
00:27:40.559 --> 00:27:45.160
<v Speaker 3>are highly unstable on a cosmic timescale. The gravitational interactions

555
00:27:45.160 --> 00:27:47.839
<v Speaker 3>in the system pack that tightly are immense.

556
00:27:47.480 --> 00:27:49.400
<v Speaker 2>So they're tugging on each other constantly.

557
00:27:49.519 --> 00:27:53.559
<v Speaker 3>Yeah. By cataloging them, we can study orbital resonances, which

558
00:27:53.599 --> 00:27:56.039
<v Speaker 3>is how planets sync up their orbits to avoid colliding,

559
00:27:56.319 --> 00:27:58.519
<v Speaker 3>and we can test our theories on planetary migration.

560
00:27:58.880 --> 00:28:03.240
<v Speaker 2>But amidst all this crowded, chaotic orbital traffic, amidst the

561
00:28:03.359 --> 00:28:06.640
<v Speaker 2>lava worlds and the pack systems, the researchers also use

562
00:28:06.799 --> 00:28:10.640
<v Speaker 2>raven to explore something entirely different. They use this incredible

563
00:28:10.680 --> 00:28:14.000
<v Speaker 2>tool to look at a very specific and surprisingly empty

564
00:28:14.079 --> 00:28:16.960
<v Speaker 2>region of space. I was fascinated by this part of

565
00:28:17.000 --> 00:28:19.960
<v Speaker 2>the study. The AI helped clarify the boundaries of a

566
00:28:20.039 --> 00:28:22.400
<v Speaker 2>rare class of planets found in an area known as

567
00:28:22.440 --> 00:28:23.640
<v Speaker 2>the Neptunian Desert.

568
00:28:23.839 --> 00:28:26.119
<v Speaker 3>The Neptuny Desert is one of the most intriguing and

569
00:28:26.160 --> 00:28:30.440
<v Speaker 3>hotly debated concepts in modern astrophysics. Essentially, the AI mapped

570
00:28:30.440 --> 00:28:33.440
<v Speaker 3>out the demographics of planets and confirmed a region where

571
00:28:33.480 --> 00:28:36.440
<v Speaker 3>theoretical physics strongly predicts planets of a certain size and

572
00:28:36.559 --> 00:28:39.079
<v Speaker 3>composition should be almost entirely non existent.

573
00:28:39.319 --> 00:28:42.759
<v Speaker 2>Okay, this raises an important question, why is it a desert.

574
00:28:43.319 --> 00:28:46.200
<v Speaker 2>If we are finding planets packed in everywhere else, if

575
00:28:46.200 --> 00:28:49.160
<v Speaker 2>we are finding planets hugging their stars on sixteen hour

576
00:28:49.279 --> 00:28:53.440
<v Speaker 2>orbits and magma oceans, why does physics say this specific zone,

577
00:28:53.559 --> 00:28:57.640
<v Speaker 2>for this specific type of planet should be a barren wasteland.

578
00:28:58.480 --> 00:29:01.880
<v Speaker 3>It has to do with the dell violent balance between

579
00:29:01.920 --> 00:29:06.599
<v Speaker 3>a planet's physical size, its internal composition, and its proximity

580
00:29:06.640 --> 00:29:08.880
<v Speaker 3>to the intense radiation of its host star.

581
00:29:09.359 --> 00:29:11.160
<v Speaker 2>Okay, break that down for me to break down.

582
00:29:11.160 --> 00:29:13.920
<v Speaker 3>When we talk about Neptunian planets or sub Neptunes, we

583
00:29:14.000 --> 00:29:16.480
<v Speaker 3>mean worlds roughly the size of our own ice giants

584
00:29:16.680 --> 00:29:19.799
<v Speaker 3>Uranus and Neptune. These are planets that are significantly larger

585
00:29:19.839 --> 00:29:22.039
<v Speaker 3>than Earth, but not quite as massive as Jupiter.

586
00:29:22.200 --> 00:29:23.000
<v Speaker 2>And what are they made of?

587
00:29:23.559 --> 00:29:26.400
<v Speaker 3>Crucially, they are composed of a relatively small rocky or

588
00:29:26.599 --> 00:29:30.160
<v Speaker 3>icy core surrounded by a massive thick envelope of hydrogen

589
00:29:30.200 --> 00:29:31.079
<v Speaker 3>and helium gas.

590
00:29:31.160 --> 00:29:34.200
<v Speaker 2>So they are basically a giant, dense ball of gas

591
00:29:34.200 --> 00:29:36.000
<v Speaker 2>and ice with a little pit in the center.

592
00:29:36.400 --> 00:29:39.119
<v Speaker 3>Think of a peach, where the pit is the rocky

593
00:29:39.160 --> 00:29:42.000
<v Speaker 3>core and the massive amount of flesh around it is

594
00:29:42.079 --> 00:29:45.559
<v Speaker 3>the gas atmosphere. Now, if a planet like that forms

595
00:29:45.599 --> 00:29:47.799
<v Speaker 3>further out in the Solar System. It's fine, it's coal,

596
00:29:47.920 --> 00:29:51.680
<v Speaker 3>the gas is stable. But if that Neptunian planet migrates

597
00:29:51.720 --> 00:29:54.839
<v Speaker 3>too close to its host star, say it gets dragged

598
00:29:54.880 --> 00:29:57.359
<v Speaker 3>into that sixteen day or less orbital zone we were

599
00:29:57.359 --> 00:30:01.079
<v Speaker 3>talking about earlier, it enters a profoundly hostile environment. It

600
00:30:01.119 --> 00:30:06.599
<v Speaker 3>gets cooked, it is subjected to intense, searing stellar radiation.

601
00:30:07.039 --> 00:30:09.640
<v Speaker 3>We are talking about a relentless bombardment of X rays,

602
00:30:09.680 --> 00:30:14.559
<v Speaker 3>extreme ultraviolet light, and powerful stellar winds constantly blasting the

603
00:30:14.559 --> 00:30:15.920
<v Speaker 3>planet's upper atmosphere.

604
00:30:15.960 --> 00:30:18.599
<v Speaker 2>Okay, but a planet like Jupiter is also mostly gas,

605
00:30:18.720 --> 00:30:21.400
<v Speaker 2>and we find hot jupiters orbiting incredibly close to their

606
00:30:21.440 --> 00:30:24.000
<v Speaker 2>stars all the time. Why doesn' Jupiter get destroyed?

607
00:30:24.160 --> 00:30:27.359
<v Speaker 3>That is the crucial distinction. For a massive gas giant

608
00:30:27.359 --> 00:30:30.480
<v Speaker 3>the size of Jupiter, its own internal gravity is immense.

609
00:30:30.799 --> 00:30:33.559
<v Speaker 3>The gravitational pull of the planet is so strong that

610
00:30:33.640 --> 00:30:36.759
<v Speaker 3>it can hold on to its massive gaseous atmosphere despite

611
00:30:36.759 --> 00:30:39.319
<v Speaker 3>the intense radiation trying to blast it away. It's too

612
00:30:39.359 --> 00:30:41.319
<v Speaker 3>heavy to be blown apart easily got it. On the

613
00:30:41.319 --> 00:30:44.720
<v Speaker 3>other end of the spectrum. For a small, dense, rocky

614
00:30:44.799 --> 00:30:47.640
<v Speaker 3>planet like Earth or Venus or one of those lava

615
00:30:47.680 --> 00:30:51.039
<v Speaker 3>worlds we discussed, there isn't a massive thick envelope of

616
00:30:51.119 --> 00:30:53.440
<v Speaker 3>hydrogen gas to strip away in the first place. It's

617
00:30:53.519 --> 00:30:54.480
<v Speaker 3>just a solid rock.

618
00:30:54.720 --> 00:30:57.920
<v Speaker 2>Ah, So a neptune sized planet is caught in the

619
00:30:58.039 --> 00:30:59.079
<v Speaker 2>tragic metal ground.

620
00:30:59.200 --> 00:31:03.160
<v Speaker 3>It's the worst of worlds precisely. Its gravity isn't strong

621
00:31:03.240 --> 00:31:07.079
<v Speaker 3>enough to tightly hold onto that massive, fluffy gaseous envelope

622
00:31:07.240 --> 00:31:10.559
<v Speaker 3>against the relentless assault of the star's high energy photons.

623
00:31:10.759 --> 00:31:13.640
<v Speaker 3>But it has too much gas to just ignore the radiation.

624
00:31:13.759 --> 00:31:14.920
<v Speaker 2>So what happens to the gas?

625
00:31:15.440 --> 00:31:19.759
<v Speaker 3>Theoretical models, specifically mechanisms like photoevaporation, where the light literally

626
00:31:19.799 --> 00:31:22.759
<v Speaker 3>heats the gas until it escapes the planet's gravity, predict

627
00:31:22.839 --> 00:31:25.799
<v Speaker 3>that the intense radiation should literally boil the atmosphere off

628
00:31:25.799 --> 00:31:26.359
<v Speaker 3>into space.

629
00:31:26.480 --> 00:31:26.920
<v Speaker 2>Wow.

630
00:31:27.319 --> 00:31:29.359
<v Speaker 3>Over a span of millions or billions of years, the

631
00:31:29.400 --> 00:31:33.079
<v Speaker 3>Neptune is violently dismantled. The gas is stripped away entirely,

632
00:31:33.359 --> 00:31:36.599
<v Speaker 3>leaving behind nothing but its bare much smaller rocky core.

633
00:31:37.119 --> 00:31:40.119
<v Speaker 2>So the peach gets entirely eaten away by the star,

634
00:31:40.359 --> 00:31:42.839
<v Speaker 2>leaving only the pit, and that pit just looks like

635
00:31:42.839 --> 00:31:46.559
<v Speaker 2>a slightly large, rocky planet. Therefore, according to the physics

636
00:31:46.559 --> 00:31:51.200
<v Speaker 2>of photo evaporation. Finding an intact, fluffy, neptune sized planet

637
00:31:51.240 --> 00:31:54.119
<v Speaker 2>with its atmosphere still attached that close to a star

638
00:31:54.759 --> 00:31:56.839
<v Speaker 2>should be statistically impossible.

639
00:31:56.920 --> 00:31:59.400
<v Speaker 3>It should be a desert, and yet we found some.

640
00:32:00.200 --> 00:32:04.039
<v Speaker 3>Really Yeah, Raven's unparalleled precision allowed it to sift through

641
00:32:04.079 --> 00:32:07.000
<v Speaker 3>the noise and validate a handful of worlds sitting right

642
00:32:07.079 --> 00:32:10.359
<v Speaker 3>in the middle of this supposed wasteland. We are detecting

643
00:32:10.640 --> 00:32:14.160
<v Speaker 3>intact neptunes surviving in a radiation environment that our current

644
00:32:14.200 --> 00:32:17.079
<v Speaker 3>models say should have completely dismantled them eons.

645
00:32:17.079 --> 00:32:19.799
<v Speaker 2>Again, which means our understanding of the universe is broken,

646
00:32:20.119 --> 00:32:23.400
<v Speaker 2>or at least incomplete. If these planets exist where they shouldn't,

647
00:32:23.440 --> 00:32:25.960
<v Speaker 2>there are survival mechanisms at play that we don't fully

648
00:32:26.000 --> 00:32:26.640
<v Speaker 2>grasp yet.

649
00:32:26.720 --> 00:32:30.839
<v Speaker 3>It forces the entire astrophysical community to reconsider the timeline

650
00:32:30.880 --> 00:32:35.000
<v Speaker 3>and mechanics of planetary evolution. It poses massive questions like what,

651
00:32:35.240 --> 00:32:38.799
<v Speaker 3>well are our models of photo evaporation too aggressive? Perhaps

652
00:32:38.880 --> 00:32:42.039
<v Speaker 3>these neptunes only very recently arrived in their tight orbits.

653
00:32:42.559 --> 00:32:45.599
<v Speaker 3>Maybe they migrated inward late in the system's life and

654
00:32:45.640 --> 00:32:47.720
<v Speaker 3>the star just hasn't had enough time to strip them there.

655
00:32:47.839 --> 00:32:49.880
<v Speaker 3>Yet we are catching them in the act of.

656
00:32:49.880 --> 00:32:53.160
<v Speaker 2>Dying, or maybe the planets themselves have defenses we didn't anticipate.

657
00:32:53.480 --> 00:32:56.359
<v Speaker 2>What if they have internal planetary magnetic fields that are

658
00:32:56.480 --> 00:32:59.759
<v Speaker 2>orders of magnitude stronger than anything we've observed, acting like

659
00:32:59.759 --> 00:33:03.279
<v Speaker 2>a mass of deflector shield protecting their atmospheres from the

660
00:33:03.359 --> 00:33:04.160
<v Speaker 2>stellar wind.

661
00:33:04.400 --> 00:33:08.319
<v Speaker 3>Exactly. These are the kinds of profound field altering discoveries

662
00:33:08.680 --> 00:33:12.240
<v Speaker 3>that drive physics forward. When the data contradicts the theory,

663
00:33:12.319 --> 00:33:17.480
<v Speaker 3>the theory must evolve finding these bizarre, extreme individual worlds,

664
00:33:17.559 --> 00:33:21.960
<v Speaker 3>the lava planets, the crowded, chaotic systems, the impossible surviving

665
00:33:22.000 --> 00:33:25.759
<v Speaker 3>neptunes is an incredible feat of engineering and software design.

666
00:33:26.039 --> 00:33:27.720
<v Speaker 2>But when I was looking at the scope of the

667
00:33:27.720 --> 00:33:30.839
<v Speaker 2>Warwick team's work, I realized that the true power of

668
00:33:30.880 --> 00:33:34.680
<v Speaker 2>this AI isn't just acting as a cosmic stamp collector

669
00:33:35.039 --> 00:33:39.079
<v Speaker 2>building a gallery of space oddities and weird planets. It's

670
00:33:39.119 --> 00:33:43.160
<v Speaker 2>about doing planetary census work on a galactic scale. We

671
00:33:43.240 --> 00:33:47.440
<v Speaker 2>are moving from studying individual interesting trees to mapping the

672
00:33:47.440 --> 00:33:51.359
<v Speaker 2>demographics of the entire forest. Let's shift gears and look

673
00:33:51.400 --> 00:33:54.640
<v Speaker 2>at the broader demographics of the galaxy. Because the numbers

674
00:33:54.799 --> 00:33:56.759
<v Speaker 2>Raven generated are mind bending.

675
00:33:56.960 --> 00:33:59.519
<v Speaker 3>This is where the Warwick Team's companion study, which was

676
00:33:59.559 --> 00:34:02.519
<v Speaker 3>also publish recently, really changes the game for theorists.

677
00:34:02.599 --> 00:34:04.039
<v Speaker 2>Yeah, let's get into those numbers.

678
00:34:04.200 --> 00:34:06.599
<v Speaker 3>By having a data set that is so heavily vetted

679
00:34:06.640 --> 00:34:10.440
<v Speaker 3>by Raven and so demonstrably free of those pesky false

680
00:34:10.480 --> 00:34:14.000
<v Speaker 3>positives we discussed earlier, the researchers could confidently step back

681
00:34:14.000 --> 00:34:18.199
<v Speaker 3>and perform massive demographic analysis. They mapped out the prevalence

682
00:34:18.239 --> 00:34:21.800
<v Speaker 3>of these close in planets with unprecedented surgical detail. They

683
00:34:21.800 --> 00:34:23.880
<v Speaker 3>weren't just asking what is out there? But how common

684
00:34:23.960 --> 00:34:24.159
<v Speaker 3>is it?

685
00:34:24.559 --> 00:34:28.119
<v Speaker 2>And the big demographic statistic they unveiled is staggering. It

686
00:34:28.199 --> 00:34:31.719
<v Speaker 2>completely reorients your perspective of the galaxy. They found that

687
00:34:31.760 --> 00:34:34.239
<v Speaker 2>around nine to ten percent of all Sun like stars

688
00:34:34.280 --> 00:34:36.960
<v Speaker 2>host a close end planet, meaning a planet orbiting in

689
00:34:37.000 --> 00:34:38.159
<v Speaker 2>less than sixteen days.

690
00:34:38.559 --> 00:34:41.639
<v Speaker 3>Let's contextualize that number because it is massive. When we

691
00:34:41.679 --> 00:34:45.199
<v Speaker 3>say some like stars, astronomers are specifically referring to what

692
00:34:45.239 --> 00:34:48.480
<v Speaker 3>we call FG and K main sequence stars.

693
00:34:48.599 --> 00:34:50.159
<v Speaker 2>Okay, what does that mean for a lay person?

694
00:34:50.320 --> 00:34:53.679
<v Speaker 3>Stars are classified by their temperature and spectrum. F stars

695
00:34:53.719 --> 00:34:55.920
<v Speaker 3>are a bit hotter and larger than our Sun. G

696
00:34:56.039 --> 00:34:58.559
<v Speaker 3>stars are exactly like our Sun, and K stars are

697
00:34:58.559 --> 00:35:03.000
<v Speaker 3>slightly cooler, smaller or dwarfs. These FGK stars make up

698
00:35:03.039 --> 00:35:06.159
<v Speaker 3>a very significant, stable portion of the stellar population in

699
00:35:06.199 --> 00:35:08.280
<v Speaker 3>the Milky Way. They are the prime real estate for

700
00:35:08.320 --> 00:35:10.320
<v Speaker 3>finding planets and potentially life.

701
00:35:10.559 --> 00:35:13.400
<v Speaker 2>Here's where it gets really interesting. I want you to

702
00:35:13.440 --> 00:35:16.039
<v Speaker 2>go outside tonight, assuming it's a clear night wherever you

703
00:35:16.039 --> 00:35:18.440
<v Speaker 2>are listening from, and just look up at the night sky.

704
00:35:19.280 --> 00:35:22.239
<v Speaker 2>Pick out the stars you can see with your naked eye.

705
00:35:22.320 --> 00:35:24.719
<v Speaker 2>The vast majority of the individual stars you can see

706
00:35:24.719 --> 00:35:29.280
<v Speaker 2>are relatively close to us in the galactic neighborhood. Statistically speaking,

707
00:35:29.639 --> 00:35:32.719
<v Speaker 2>roughly one in ten of the sunlike stars you were

708
00:35:32.760 --> 00:35:35.440
<v Speaker 2>looking at right now has a planet whipping around it

709
00:35:35.480 --> 00:35:39.519
<v Speaker 2>at breakneck speed, completing a year in less than sixteen days.

710
00:35:39.840 --> 00:35:41.920
<v Speaker 2>It's hard to wrap your head around ten percent. It

711
00:35:41.960 --> 00:35:45.440
<v Speaker 2>completely changes how you view the cosmos. The sky isn't

712
00:35:45.440 --> 00:35:48.599
<v Speaker 2>just a collection of lonely isolated fusion reactors burning in

713
00:35:48.639 --> 00:35:52.679
<v Speaker 2>the void. It is teeming with complex planetary systems operating

714
00:35:52.679 --> 00:35:55.880
<v Speaker 2>on hyper speed. Every tenth star you point your finger

715
00:35:55.920 --> 00:35:58.199
<v Speaker 2>at has a world practically touching its surface.

716
00:35:58.519 --> 00:36:01.519
<v Speaker 3>It is a profoundly humbling to know that the night

717
00:36:01.599 --> 00:36:04.559
<v Speaker 3>sky is that crowded. Now to be fair to the

718
00:36:04.599 --> 00:36:07.679
<v Speaker 3>history of astronomy and to provide the proper context. This

719
00:36:07.800 --> 00:36:10.679
<v Speaker 3>nine to ten percent figure isn't entirely new. It is

720
00:36:10.760 --> 00:36:14.519
<v Speaker 3>largely consistent with the earlier estimates produced by NASA's legendary

721
00:36:14.559 --> 00:36:19.079
<v Speaker 3>Kepler mission, which operated in the decade before tests. Kepler

722
00:36:19.119 --> 00:36:22.800
<v Speaker 3>was the great pioneer of exo planet demographics. It stared

723
00:36:22.840 --> 00:36:26.079
<v Speaker 3>at a single fixed patch of sky in the Sickness

724
00:36:26.079 --> 00:36:29.679
<v Speaker 3>Constellation for years, and it gave humanity our first, real,

725
00:36:30.000 --> 00:36:33.079
<v Speaker 3>mathematically sound estimates of planetary occurrence rates.

726
00:36:33.239 --> 00:36:34.960
<v Speaker 2>Right. I'm glad you brought that up, because when I

727
00:36:35.000 --> 00:36:37.000
<v Speaker 2>read the ten percent number, my first thought was, didn't

728
00:36:37.039 --> 00:36:39.519
<v Speaker 2>we already know that from Kepler? If Kepler already gave

729
00:36:39.599 --> 00:36:42.199
<v Speaker 2>us that ten percent estimate a decade ago, what makes

730
00:36:42.199 --> 00:36:45.639
<v Speaker 2>the test data and the raven pipeline so groundbreaking today?

731
00:36:45.679 --> 00:36:47.159
<v Speaker 2>Why is this a new breakthrough?

732
00:36:47.599 --> 00:36:51.159
<v Speaker 3>It fundamentally comes down to precision and confidence. Kepler gave

733
00:36:51.239 --> 00:36:53.599
<v Speaker 3>us the initial estimate, but it was an estimate with

734
00:36:53.719 --> 00:36:55.599
<v Speaker 3>relatively wide aerror bars.

735
00:36:55.480 --> 00:36:57.280
<v Speaker 2>So they weren't totally sure exactly.

736
00:36:57.400 --> 00:36:59.840
<v Speaker 3>Yea, there was still a margin of uncertainty because of

737
00:36:59.880 --> 00:37:03.199
<v Speaker 3>the false positive rates and the limitations of the older software.

738
00:37:04.039 --> 00:37:08.320
<v Speaker 3>Raven processing the incredibly rich test data mapp this exact

739
00:37:08.320 --> 00:37:11.280
<v Speaker 3>same demographic, but with uncertainties that are up to ten

740
00:37:11.320 --> 00:37:11.960
<v Speaker 3>times smaller.

741
00:37:12.360 --> 00:37:16.199
<v Speaker 2>Wow, ten times more precise. That isn't just a minor

742
00:37:16.239 --> 00:37:18.760
<v Speaker 2>incremental update. That's a massive leap and resolution.

743
00:37:18.960 --> 00:37:22.199
<v Speaker 3>Think of it this way. Kepler proved the concept. It

744
00:37:22.280 --> 00:37:25.480
<v Speaker 3>was like an early explorer sailing across the ocean and

745
00:37:25.559 --> 00:37:29.000
<v Speaker 3>sketching a rough pencil outline of a newly discovered continent.

746
00:37:29.199 --> 00:37:31.360
<v Speaker 3>They could tell you the continent was there and roughly

747
00:37:31.360 --> 00:37:35.199
<v Speaker 3>how big it was. But Raven and tests are refining

748
00:37:35.239 --> 00:37:37.079
<v Speaker 3>that map to a degree we've never seen.

749
00:37:37.400 --> 00:37:38.920
<v Speaker 2>They're mapping the actual terrain.

750
00:37:39.199 --> 00:37:42.199
<v Speaker 3>They are like a modern constellation of satellites providing a

751
00:37:42.320 --> 00:37:47.119
<v Speaker 3>high definition, millimeter accurate topographical map of that same continent.

752
00:37:47.760 --> 00:37:51.440
<v Speaker 3>When your statistical uncertainty shrink by a massive factor of ten,

753
00:37:52.000 --> 00:37:55.119
<v Speaker 3>your theoretical models of how solar systems form suddenly have

754
00:37:55.199 --> 00:37:57.920
<v Speaker 3>to become much much more rigorous to fit the new data.

755
00:37:58.039 --> 00:38:01.079
<v Speaker 3>You can't hide sloppy physics inside way airbars anymore.

756
00:38:01.159 --> 00:38:03.280
<v Speaker 2>That makes perfect scent. The tighter the data, the harder

757
00:38:03.280 --> 00:38:05.639
<v Speaker 2>it is for a bad theory to survive, And that

758
00:38:05.800 --> 00:38:09.360
<v Speaker 2>exact precision is what allowed the Warwick researchers to finally

759
00:38:09.440 --> 00:38:12.039
<v Speaker 2>quantify the Neptunian desert. We were just talking.

760
00:38:11.760 --> 00:38:13.199
<v Speaker 3>About, yes, the desert numbers.

761
00:38:13.400 --> 00:38:16.079
<v Speaker 2>We discuss the physics of why those planets are rare,

762
00:38:16.159 --> 00:38:19.360
<v Speaker 2>how their atmospheres get boiled away. But the Raven pipeline

763
00:38:19.360 --> 00:38:21.880
<v Speaker 2>allowed the researchers to move beyond just saying they are

764
00:38:21.960 --> 00:38:26.159
<v Speaker 2>rare and actually provide the first direct, highly precise, empirical

765
00:38:26.199 --> 00:38:30.320
<v Speaker 2>measurement of this phenomenon. They calculated that Neptunian planets surviving

766
00:38:30.320 --> 00:38:33.239
<v Speaker 2>in tight orbits occur around a mere point zero eight

767
00:38:33.320 --> 00:38:34.559
<v Speaker 2>percent of Sun like.

768
00:38:34.519 --> 00:38:37.559
<v Speaker 3>Stars, less than one tenth of one percent. If we

769
00:38:37.599 --> 00:38:40.480
<v Speaker 3>connect this to the bigger picture of astrophysics, that point

770
00:38:40.599 --> 00:38:42.880
<v Speaker 3>zero eight percent isn't just a neat piece of trivia

771
00:38:42.920 --> 00:38:46.559
<v Speaker 3>for a textbook. It is a hard, unyielding mathematical boundary.

772
00:38:47.039 --> 00:38:49.639
<v Speaker 3>It is an empirical constraint on the fundamental physics of

773
00:38:49.679 --> 00:38:52.679
<v Speaker 3>solar system formation and atmospheric escape.

774
00:38:52.920 --> 00:38:55.559
<v Speaker 2>It's the universe drawing a literal line in the sand.

775
00:38:55.880 --> 00:38:59.039
<v Speaker 2>It tells us, with absolute numerical precision, exactly where the

776
00:38:59.079 --> 00:39:01.519
<v Speaker 2>tipping point is betwel in a planet surviving its stars,

777
00:39:01.639 --> 00:39:05.199
<v Speaker 2>radiation and a planet being mercilessly boiled down to its bare,

778
00:39:05.320 --> 00:39:06.440
<v Speaker 2>rocky core, and.

779
00:39:06.440 --> 00:39:09.760
<v Speaker 3>By establishing these hard boundaries the ten percent prevalence of

780
00:39:09.760 --> 00:39:13.159
<v Speaker 3>short period planets the point zero eight percent survival rate

781
00:39:13.199 --> 00:39:16.960
<v Speaker 3>in the extreme radiation of the Neptunian desert, the Raven

782
00:39:17.039 --> 00:39:21.159
<v Speaker 3>team is providing the anchor points for every single theoretical

783
00:39:21.239 --> 00:39:24.760
<v Speaker 3>astrophysicist trying to write the history of the galaxy.

784
00:39:24.320 --> 00:39:26.719
<v Speaker 2>So they have to build their models around these number.

785
00:39:27.000 --> 00:39:30.039
<v Speaker 3>Exactly when a scientist today sits down to build a

786
00:39:30.079 --> 00:39:33.880
<v Speaker 3>complex computer simulation of how dust and gas coalesced to

787
00:39:33.920 --> 00:39:37.440
<v Speaker 3>form planets, they have to ensure that their simulation naturally

788
00:39:37.480 --> 00:39:41.320
<v Speaker 3>produces these exact percentages. If your computer model of a

789
00:39:41.320 --> 00:39:44.280
<v Speaker 3>solar system forms Neptunes in tight orbits five percent of

790
00:39:44.360 --> 00:39:46.599
<v Speaker 3>the time, you know your model is wrong, because the

791
00:39:46.719 --> 00:39:49.599
<v Speaker 3>Raven data says it only happens point zero eight percent

792
00:39:49.639 --> 00:39:53.000
<v Speaker 3>of the time. It keeps our theoretical physics strictly grounded

793
00:39:53.000 --> 00:39:54.360
<v Speaker 3>in observational reality.

794
00:39:54.440 --> 00:39:58.199
<v Speaker 2>The extreme precision of these demographics fundamentally shifts how astronomers

795
00:39:58.199 --> 00:40:00.960
<v Speaker 2>will plan their next decades of research, which brings us

796
00:40:01.000 --> 00:40:03.000
<v Speaker 2>to the final and perhaps the most exciting part of

797
00:40:03.039 --> 00:40:06.119
<v Speaker 2>this breakthrough, where we go from here because this paper

798
00:40:06.239 --> 00:40:08.039
<v Speaker 2>isn't just the end of a study. It's not a

799
00:40:08.079 --> 00:40:10.920
<v Speaker 2>closed book. It is the beginning of a whole new

800
00:40:11.000 --> 00:40:14.800
<v Speaker 2>era of exploration. The Warwick team has embraced what I

801
00:40:14.840 --> 00:40:17.840
<v Speaker 2>think is the absolute best part of modern science, the

802
00:40:17.920 --> 00:40:19.239
<v Speaker 2>open source universe.

803
00:40:19.519 --> 00:40:22.599
<v Speaker 3>It is a vital transformative point. The development of a

804
00:40:22.599 --> 00:40:26.800
<v Speaker 3>pipeline like Raven doesn't just generate astronomical discoveries in a vacuum.

805
00:40:27.360 --> 00:40:31.960
<v Speaker 3>It simultaneously stress tests the absolute limits of artificial intelligence

806
00:40:32.079 --> 00:40:36.000
<v Speaker 3>on incredibly difficult, noisy, real world research problems.

807
00:40:36.000 --> 00:40:37.840
<v Speaker 2>Oh that makes sense. A pushing AI forward to.

808
00:40:37.920 --> 00:40:41.960
<v Speaker 3>Absolutely The complex neural network architectures and mathematical frameworks they're

809
00:40:41.960 --> 00:40:45.760
<v Speaker 3>building to find exoplanets push the boundaries of machine learning

810
00:40:45.920 --> 00:40:49.559
<v Speaker 3>in ways it could eventually benefit completely different, unrelated fields

811
00:40:49.599 --> 00:40:53.000
<v Speaker 3>of science. The way Raven analyzes time series data to

812
00:40:53.000 --> 00:40:55.960
<v Speaker 3>find a tiny drop in light could theoretically be adapted

813
00:40:56.000 --> 00:40:59.920
<v Speaker 3>to look for microscopic anomalies in medical data or seismic waves.

814
00:41:00.239 --> 00:41:04.960
<v Speaker 2>And crucially, the researchers haven't hoarded this AI or the

815
00:41:05.000 --> 00:41:08.639
<v Speaker 2>massive catalogs of data produced. They didn't lock it behind

816
00:41:08.679 --> 00:41:12.239
<v Speaker 2>a patent or a private university server. The team at

817
00:41:12.280 --> 00:41:16.079
<v Speaker 2>WARG has released interactive tools, the underlying code and the

818
00:41:16.119 --> 00:41:20.440
<v Speaker 2>full comprehensive catalogs of there one hundred and eighteen validated planets,

819
00:41:20.719 --> 00:41:23.239
<v Speaker 2>and there are thousands of high quality candidates to the

820
00:41:23.239 --> 00:41:24.519
<v Speaker 2>global scientific community.

821
00:41:24.599 --> 00:41:25.800
<v Speaker 3>It's a huge gift to the field.

822
00:41:25.920 --> 00:41:29.320
<v Speaker 2>They've essentially built a highly accurate AI verified roadmap of

823
00:41:29.360 --> 00:41:31.920
<v Speaker 2>the local galaxy and they've just handed the keys to

824
00:41:31.960 --> 00:41:34.559
<v Speaker 2>anyone on Earth with a telescope and an Internet connection.

825
00:41:34.960 --> 00:41:38.440
<v Speaker 3>And that roadmap is absolutely critical for the next generation

826
00:41:38.519 --> 00:41:41.480
<v Speaker 3>of space observation because the telescopes that we are building

827
00:41:41.480 --> 00:41:44.920
<v Speaker 3>now are marvels of engineering, but their time is precious.

828
00:41:45.239 --> 00:41:48.199
<v Speaker 3>We have massive new missions on the immediate horizon, like

829
00:41:48.239 --> 00:41:52.239
<v Speaker 3>the PLATO mission right yes, For example, the European Space

830
00:41:52.280 --> 00:41:55.800
<v Speaker 3>Agency is preparing for the PLATO mission, which stands for

831
00:41:55.840 --> 00:41:59.159
<v Speaker 3>Planetary Transits and Oscillations of Stars. It is set to

832
00:41:59.199 --> 00:42:01.679
<v Speaker 3>launch in the near field future, and PLATO is a

833
00:42:01.719 --> 00:42:05.320
<v Speaker 3>massive step up. It is specifically designed to search for

834
00:42:05.440 --> 00:42:10.159
<v Speaker 3>potentially habitable terrestrial Earth like planets orbiting in the habitable

835
00:42:10.280 --> 00:42:11.679
<v Speaker 3>zones of Sun like stars.

836
00:42:12.119 --> 00:42:15.320
<v Speaker 2>But launching a space telescope costs billions of dollars and

837
00:42:15.440 --> 00:42:18.800
<v Speaker 2>observing time is incredibly competitive. And expensive. You can't just

838
00:42:19.079 --> 00:42:21.719
<v Speaker 2>point Plato blindly at the sky and hope for the best,

839
00:42:22.280 --> 00:42:24.599
<v Speaker 2>and you certainly can't waste its valuable time looking at

840
00:42:24.639 --> 00:42:27.679
<v Speaker 2>targets that turn out to be false positive grazing binaries.

841
00:42:27.719 --> 00:42:30.800
<v Speaker 3>Exactly, you need a rigorously vetted target list. You need

842
00:42:30.840 --> 00:42:33.760
<v Speaker 3>to know exactly where to point the multi billion dollar instrument.

843
00:42:34.039 --> 00:42:36.960
<v Speaker 3>The one hundred and eighteen planets dalinated by Raven and

844
00:42:37.079 --> 00:42:41.119
<v Speaker 3>the thousands of pristine, high quality candidates that identified become

845
00:42:41.239 --> 00:42:42.480
<v Speaker 3>that ultimate target.

846
00:42:42.159 --> 00:42:43.920
<v Speaker 2>List, so they know exactly where to look.

847
00:42:44.119 --> 00:42:47.360
<v Speaker 3>Ground based observatory is and upcoming space missions like Plato

848
00:42:47.719 --> 00:42:51.239
<v Speaker 3>will use this exact catalog generated by the AI. They

849
00:42:51.280 --> 00:42:55.199
<v Speaker 3>will know precisely which stars warrant the intense expensive follow

850
00:42:55.280 --> 00:42:58.000
<v Speaker 3>up required to measure the exact mass of a planet

851
00:42:58.679 --> 00:43:01.480
<v Speaker 3>or to use spectroscopy to look for the chemical signatures

852
00:43:01.519 --> 00:43:05.880
<v Speaker 3>of water, oxygen, or methane in an alien atmosphere. Raven

853
00:43:05.920 --> 00:43:08.519
<v Speaker 3>has cleared the brush so Plato can see the forest.

854
00:43:09.000 --> 00:43:11.480
<v Speaker 2>So what does this all mean When we take a

855
00:43:11.480 --> 00:43:15.159
<v Speaker 2>step back from the convolutional neural networks, the sixteen day orbits,

856
00:43:15.199 --> 00:43:17.599
<v Speaker 2>and the Neptunian deserts and look at the whole picture,

857
00:43:17.960 --> 00:43:21.440
<v Speaker 2>we are seeing a profound evolution in how humanity interacts

858
00:43:21.440 --> 00:43:24.920
<v Speaker 2>with the cosmos. Artificial intelligence has allowed us to take

859
00:43:24.960 --> 00:43:28.400
<v Speaker 2>the raw, chaotic, impossibly noisy light from two point two

860
00:43:28.440 --> 00:43:31.840
<v Speaker 2>million distant stars and to still it into a high fidelity,

861
00:43:31.960 --> 00:43:34.199
<v Speaker 2>highly precise map of alien worlds.

862
00:43:34.280 --> 00:43:35.400
<v Speaker 3>It's beautiful, really.

863
00:43:35.480 --> 00:43:38.159
<v Speaker 2>We've automated the discovery of the universe. We have outsourced

864
00:43:38.159 --> 00:43:40.320
<v Speaker 2>our sense of wonder to an algorithm, and the algorithm

865
00:43:40.360 --> 00:43:41.920
<v Speaker 2>is showing us things we never could have found on

866
00:43:41.960 --> 00:43:42.320
<v Speaker 2>our own.

867
00:43:42.519 --> 00:43:46.840
<v Speaker 3>It is the perfect synthesis of human ingenuity and machine capability.

868
00:43:47.360 --> 00:43:50.079
<v Speaker 3>We built the magnificent telescope to gather the agent light,

869
00:43:50.440 --> 00:43:53.719
<v Speaker 3>We built the intricate mathematical simulation to teach the machine

870
00:43:53.920 --> 00:43:58.440
<v Speaker 3>the underlying physics of reality, and the machine tirelessly processed

871
00:43:58.519 --> 00:44:02.119
<v Speaker 3>the sheer crushing volume with data that our biological brains

872
00:44:02.159 --> 00:44:05.360
<v Speaker 3>never could have handled alone. Is a partnership that expands

873
00:44:05.360 --> 00:44:06.440
<v Speaker 3>our reach into the dark.

874
00:44:06.960 --> 00:44:09.719
<v Speaker 2>It makes you wonder about the sheer volume of information

875
00:44:09.800 --> 00:44:13.280
<v Speaker 2>we've already collected across all fields of science, which leaves

876
00:44:13.320 --> 00:44:15.840
<v Speaker 2>you with a lingering question to ponder long after we

877
00:44:15.960 --> 00:44:20.159
<v Speaker 2>finish today. If a newly designed AI pipeline train purely

878
00:44:20.199 --> 00:44:24.920
<v Speaker 2>on simulated mathematical realities can uncover over one hundred hidden

879
00:44:25.079 --> 00:44:27.800
<v Speaker 2>exotic worlds hiding in plain sight within a four year

880
00:44:27.800 --> 00:44:31.480
<v Speaker 2>old data set from tests. While what other fundamental earth

881
00:44:31.519 --> 00:44:34.239
<v Speaker 2>shattering truths of the universe are currently sitting quietly in

882
00:44:34.280 --> 00:44:37.559
<v Speaker 2>our existing data archives right here on Earth, just waiting

883
00:44:37.559 --> 00:44:39.960
<v Speaker 2>for the right algorithm to finally open our eyes to them.

884
00:44:40.079 --> 00:44:42.320
<v Speaker 3>The data is already there, resting on our hard drives.

885
00:44:42.519 --> 00:44:44.679
<v Speaker 3>The universe has already sent the message. We just need

886
00:44:44.679 --> 00:44:46.559
<v Speaker 3>to learn how to build the right machine to read it.

887
00:44:46.559 --> 00:44:49.760
<v Speaker 2>It's exactly like looking for the impossibly faint shadow of

888
00:44:49.760 --> 00:44:52.559
<v Speaker 2>a moth against a distant street light hundreds of light

889
00:44:52.639 --> 00:44:56.480
<v Speaker 2>years away. The shadow is always there. Sometimes you just

890
00:44:56.519 --> 00:44:58.719
<v Speaker 2>need a machine to tell you exactly how to look.

891
00:45:08.599 --> 00:45:43.119
<v Speaker 2>Usai