WEBVTT

1
00:00:01.840 --> 00:00:07.759
NBC News on CACAA Lomalinda sponsored by
Teamsters Local nineteen thirty two, Protecting the

2
00:00:07.799 --> 00:00:18.280
Future of working Families Teamsters nineteen thirty
two. Dot org. The information economy

3
00:00:18.359 --> 00:00:23.079
has a rid. The world is
teeming with innovation as new business models reinvent

4
00:00:23.280 --> 00:00:29.239
every industry industry. Inside Analysis is
your source of information and insight about how

5
00:00:29.280 --> 00:00:33.920
to make the most of this exciting
new era. Learn more at inside analysis

6
00:00:34.000 --> 00:00:39.880
dot Comsideanalysis dot com. And now
here's your host, Eric Kavanaugh. Yes,

7
00:00:43.840 --> 00:00:48.920
oh, yes, indeed, folks, Welcome to the future. The

8
00:00:48.960 --> 00:00:52.960
future is here already. That's what
William Gibson once wrote. It's just not

9
00:00:53.119 --> 00:00:57.759
evenly distributed. I have to say
I heard that line years ago and it

10
00:00:57.880 --> 00:01:02.719
just blew my mind. I was
like, that's pretty cool. The future

11
00:01:02.799 --> 00:01:07.359
is here already. It's just not
evenly distributed yet. But that's happening right

12
00:01:07.359 --> 00:01:10.959
now. You see it with cell
phones, for example. You see it

13
00:01:11.000 --> 00:01:15.760
with certain kinds of construction, skyscrapers. You look at the new skyscrapers and

14
00:01:15.760 --> 00:01:19.159
places like Dubai in the UAE and
they're just amazing. And that is the

15
00:01:19.200 --> 00:01:23.359
forefront of the industry. And it's
primarily because they have a lot of money.

16
00:01:23.680 --> 00:01:26.439
I want to build some cool buildings, and in the rest of the

17
00:01:26.439 --> 00:01:30.599
world you have older buildings that are
still standing, obviously, But I bring

18
00:01:30.640 --> 00:01:36.000
in this metaphor just to point out
that we are at a very transformative phase

19
00:01:36.439 --> 00:01:40.480
or moment in time as human beings, quite frankly, and with technology,

20
00:01:40.519 --> 00:01:44.599
of course, it's always out there
improving, usually improving our lives, doing

21
00:01:44.599 --> 00:01:49.239
cool things for us well. Artificial
intelligence it's not new. It's been around

22
00:01:49.280 --> 00:01:52.280
for decades, fifty sixty years if
you really push it all the way back

23
00:01:52.319 --> 00:01:57.879
even further depending upon your definition.
But in terms of widespread adoption and use,

24
00:01:57.519 --> 00:02:01.719
boom, we are in the middle
of that first major explosion. And

25
00:02:01.760 --> 00:02:07.959
you can thank the folks at open
AI for sure with their chat GPT,

26
00:02:07.639 --> 00:02:10.680
which just blew the doors off.
Now that even isn't all that new.

27
00:02:12.199 --> 00:02:15.680
GPT has been around for a number
of years now, but this version that

28
00:02:15.719 --> 00:02:20.360
they unleashed on the public really opened
eyeballs and it forced the game. It

29
00:02:20.439 --> 00:02:23.520
was a forcing function. But I
have someone on the show today who knows

30
00:02:23.560 --> 00:02:27.280
a whole lot about AI and has
known a lot about AI for a long

31
00:02:27.280 --> 00:02:30.199
long time, and he taught me
a lot about this stuff too. Asama

32
00:02:30.280 --> 00:02:34.120
Fayad. He's got a company,
Open Insights, but also he was the

33
00:02:34.159 --> 00:02:38.000
chief data officer at a company called
Yahoo. Some of you all remember those

34
00:02:38.039 --> 00:02:40.639
guys. They were the center of
gravity in like the nineties and the early

35
00:02:40.680 --> 00:02:44.800
aughts. That was where you wanted
to be as an engineer, as an

36
00:02:44.800 --> 00:02:46.919
AI person. And then of course
it kind of shifted to companies like cloud,

37
00:02:46.960 --> 00:02:51.080
Darra and Google of course, and
now open Ai. But they're still

38
00:02:51.080 --> 00:02:53.919
out there. Yeah, he's still
doing cool stuff, and so is usama

39
00:02:53.960 --> 00:02:58.919
A Sama Fayad. He's now the
executive director for the Institute for Experiential AI

40
00:02:58.960 --> 00:03:02.199
at Northeastern University. Sama, welcome
to the show here. Thanks so much

41
00:03:02.240 --> 00:03:07.560
for your time and attention. And
first of all, tell us about the

42
00:03:07.599 --> 00:03:12.960
Northeastern University Institute for Experiential AI.
Where did this idea come from and what's

43
00:03:13.000 --> 00:03:21.199
your focus? Thanks? Thanks,
Eric. Idea came from the president of

44
00:03:21.800 --> 00:03:29.039
the university, Joseph and the provost
David Madigan, who's actually a well established

45
00:03:29.479 --> 00:03:37.280
statistician an AI expert in his own
right, and some donors who gave some

46
00:03:37.360 --> 00:03:39.840
large grants saying hey, AI is
going to be big. This was about

47
00:03:40.120 --> 00:03:46.159
This was five years ago, So
around late two thousand they started saying,

48
00:03:46.199 --> 00:03:51.360
Okay, we've got a lot of
funding, let's recruit for someone to head

49
00:03:51.360 --> 00:03:57.800
it up and define it the way
we I'm the inaugural executive director for this,

50
00:03:58.120 --> 00:04:02.879
as well as joining as afessor in
the Computer Science College, Corey College

51
00:04:02.879 --> 00:04:09.680
for Computer Sciences. So experiential AI. The first thing we need to do

52
00:04:09.759 --> 00:04:14.159
is name it. And we use
the term experiential AI to meet AI with

53
00:04:14.240 --> 00:04:18.120
the human in the loop. Because
the reality of my exposure to AI through

54
00:04:19.199 --> 00:04:26.040
my work, whether it was with
Microsoft, NASA's GPL, my own startups,

55
00:04:27.240 --> 00:04:31.279
Yahoo, later with Barclays, and
definitely with Open Insights, working with

56
00:04:31.680 --> 00:04:36.800
many companies around the world, including
AI in the healthcare space, etc.

57
00:04:39.399 --> 00:04:46.319
Has shown me one theme that is
constant, which is every working AI has

58
00:04:46.360 --> 00:04:49.439
a huge element of the human in
the loop. Whether you're talking about the

59
00:04:49.439 --> 00:04:55.759
core of the Google Search Engine or
we're talking about open AI and chat GPT

60
00:04:55.959 --> 00:05:00.519
in any of these large language models, a lot of feedback all the time

61
00:05:00.560 --> 00:05:04.399
from humans supplying what AI cannot supply. And we, like you mentioned,

62
00:05:04.399 --> 00:05:10.480
we've spent over seven decades trying to
figure out stuff like common sense reasoning.

63
00:05:11.480 --> 00:05:15.360
One of the big learning lessons is
you only can get that from humans today.

64
00:05:15.399 --> 00:05:18.360
You cannot get a machine or an
algorithm to provide you that feedback,

65
00:05:18.399 --> 00:05:23.759
which is why there's a lot of
human in the loop intervention in the operation

66
00:05:23.879 --> 00:05:30.959
of AI the other and that became
a theme because what it means is you

67
00:05:30.000 --> 00:05:38.480
need to think about how AI help
human intelligence and how human intelligence helps machine

68
00:05:38.519 --> 00:05:43.240
intelligence or AI become a reality.
So it's about kind of the collaboration with

69
00:05:43.639 --> 00:05:46.360
algorithms and robots and so forth.
So that's how we define it. And

70
00:05:46.399 --> 00:05:50.519
then we had to go look for, Okay, what are areas that need

71
00:05:50.600 --> 00:05:57.279
focus where as an institute we can
make a large contribution, We can become

72
00:05:57.319 --> 00:06:02.439
the leader in certain areas and quickly
IDENTI responsible AI as one big area of

73
00:06:02.480 --> 00:06:06.519
focus from a research perspective, defining
what it is, how do you do

74
00:06:06.560 --> 00:06:10.480
it, how do you do risk
assessment, et cetera. And then we

75
00:06:10.519 --> 00:06:16.720
also identified kind of what I like
to call new generation large language models or

76
00:06:16.759 --> 00:06:24.560
new generation generative AI. So how
do you reduce this obsession with larger and

77
00:06:24.680 --> 00:06:29.240
larger parameters in the model, Larger
models that can so easily go out of

78
00:06:29.279 --> 00:06:36.480
control, start making errors, hallucinating
as Google calls it or basically are extremely

79
00:06:36.519 --> 00:06:42.800
expensive to train because the more parameters
you have, the more and of course

80
00:06:42.839 --> 00:06:49.480
capital equipment wise that the training infrastructure
and the inference is so Rather than how

81
00:06:49.480 --> 00:06:54.040
can we make models bigger, the
question is how do I get away with

82
00:06:54.079 --> 00:07:00.839
the smallest possible large model and enhance
it by stuff that we know prior knowledge?

83
00:07:01.040 --> 00:07:03.959
Do I have a knowledge graph about
the domain? Do I understand who

84
00:07:03.959 --> 00:07:09.279
the people are, professional graph,
a citation graph, et cetera. It

85
00:07:09.360 --> 00:07:15.279
is silly to try to let the
AI rediscover from pure data, very very

86
00:07:15.319 --> 00:07:18.759
expensive data, what we already know. So can you supplement this? And

87
00:07:18.800 --> 00:07:24.279
we call this better together? And
then we picked certain areas and we are

88
00:07:24.319 --> 00:07:29.600
seeing better results indeed qualitatively and qualitatively
when we apply these two things together.

89
00:07:30.680 --> 00:07:34.800
Finally, we had to pick kind
of areas of focus for applications, which

90
00:07:35.040 --> 00:07:39.800
is critical for us. So applications, in my opinion, have been the

91
00:07:39.920 --> 00:07:45.399
driver in the developments of AI for
the past decade or more. So.

92
00:07:45.439 --> 00:07:49.600
We looked at applied areas in AI
plus health, AI plus life sciences,

93
00:07:49.759 --> 00:07:56.920
drug discovery and molecule design and things
like that, antibody design. And then

94
00:07:57.360 --> 00:08:03.439
the last area we added was AI
for climate and sustainability. How do we

95
00:08:03.560 --> 00:08:07.399
leverage the new generations of data sets
that are kind of hyper local, very

96
00:08:07.439 --> 00:08:11.480
detailed, huge big data, if
you will, in the world of climate

97
00:08:13.040 --> 00:08:18.199
and then use it to model both
the climate and the environment and predict things

98
00:08:18.240 --> 00:08:20.759
like here are the impacts of a
potential disaster or a change or what have

99
00:08:20.879 --> 00:08:26.800
you. So these areas define where
we work and to drive the wheel forward.

100
00:08:26.879 --> 00:08:31.519
To start the flywheel going, we
created the AI Solutions Hub. One

101
00:08:31.559 --> 00:08:37.279
of the big funders of the institute
is Dave Rue, who funded also the

102
00:08:37.360 --> 00:08:43.320
Rue Institute up in Portland, Maine, in order to change the economy in

103
00:08:43.399 --> 00:08:46.879
Maine, and AI in his vision, was a big area. So we

104
00:08:46.960 --> 00:08:50.279
created this AI Solutions Hub, which
is a team, a delivery team of

105
00:08:50.360 --> 00:08:54.480
data scientists, data engineers, AI
practitioners whose job is to engage with our

106
00:08:54.559 --> 00:09:00.840
partners companies, organizations, governments and
work on real child just in the real

107
00:09:00.879 --> 00:09:03.879
world, in order to drive that
activity, figure out what the real research

108
00:09:03.919 --> 00:09:11.399
agenda should be and at the same
time figure out how do we train our

109
00:09:11.480 --> 00:09:18.080
students and the corporate learners on AI
in the wild, AI happening in reality?

110
00:09:18.480 --> 00:09:20.519
What are the real issues? How
do you deal with it. You

111
00:09:20.559 --> 00:09:24.759
know, we all know that dependence
deep dependence of data and generative AI,

112
00:09:26.440 --> 00:09:30.879
of generative AI on data. But
the real issue here is what do you

113
00:09:30.919 --> 00:09:35.480
do? And the data is not
an idealized circumstances. It's got quality problems.

114
00:09:35.799 --> 00:09:37.799
Half of it is missing, a
third of what's there is wrong.

115
00:09:39.440 --> 00:09:41.679
But you've got to make the solution
work, and you've got to make it

116
00:09:41.720 --> 00:09:46.840
work in the constraints of kind of
real life and real businesses and real regulations

117
00:09:46.840 --> 00:09:50.200
and so forth. So all of
this coming together gives us, if you

118
00:09:50.360 --> 00:09:54.559
like, the active lab that generates
the research agenda and generates the chance to

119
00:09:54.600 --> 00:09:58.840
train our students and learners from the
corporations on how to make AI solutions work.

120
00:10:00.279 --> 00:10:03.480
Yeah. I really love the vision
here because you're appreciating the importance of

121
00:10:03.559 --> 00:10:09.120
research but also the importance of training, the importance of understanding use cases.

122
00:10:09.159 --> 00:10:13.159
Hence the solution portfolio or the solution
hub. And you know, I want

123
00:10:13.159 --> 00:10:16.440
to drive into one key issue here
that you brought up, because I think

124
00:10:16.440 --> 00:10:22.639
it's the most important one to make
this stuff enterprise ready and business ready and

125
00:10:22.639 --> 00:10:26.159
consumer friendly, which is the truth. Right, We have one of our

126
00:10:26.600 --> 00:10:31.279
regular listeners Alex Husky, super smart
guy. He works for rex On Mobile.

127
00:10:31.600 --> 00:10:33.320
He sent me an email we did
an webinar a few weeks ago.

128
00:10:33.679 --> 00:10:37.840
He said, we cannot allow these
large language models to pattern recognize their way

129
00:10:37.879 --> 00:10:41.519
to the truth. And that's kind
of what they're doing, is there.

130
00:10:41.720 --> 00:10:46.879
It's basically a pattern recognition engine that
predicts text based upon the prompt and all

131
00:10:46.919 --> 00:10:50.519
the corpus of text that it has
consumed. Well, you mentioned citations for

132
00:10:50.600 --> 00:10:56.080
example, and this whole concept of
embeddings comes into play, and that's what

133
00:10:56.120 --> 00:10:58.960
you need, right, that's the
ground truth our citations. This is where

134
00:11:00.120 --> 00:11:05.240
got the information so you can verify
that it's correct. That's a cornerstone of

135
00:11:05.279 --> 00:11:09.320
making this stuff doable and workable for
business, right, absolutely, absolutely,

136
00:11:09.879 --> 00:11:16.320
And you know, citations are a
beautiful example. You know, this autocomplete

137
00:11:16.320 --> 00:11:22.159
technology that we call large language models, it's trying to you know, build

138
00:11:22.159 --> 00:11:24.879
its output a token at a time, and the token is part of a

139
00:11:24.960 --> 00:11:28.879
word, part of a group of
pixels, depending on what your output looks

140
00:11:28.960 --> 00:11:35.679
like. Well, trying to recollect
a precise citation based on an autocomplete is

141
00:11:35.720 --> 00:11:41.399
a very dangerous exercise and rife with
room for making an error. You can

142
00:11:41.440 --> 00:11:43.480
get the year wrong, the author's
wrong, you can make up, you

143
00:11:43.480 --> 00:11:46.480
can come up with a title,
You could change the title all based on

144
00:11:46.519 --> 00:11:50.159
your recollection. Well, why do
all that? If you already have these

145
00:11:50.200 --> 00:11:54.120
citations in very precise format, how
can you use them? And then tying

146
00:11:54.200 --> 00:11:58.600
the citations to the content, to
the meaning is the other big challenge because

147
00:11:58.639 --> 00:12:03.360
most of these models, I mean
we call them technically stochastic parrots because they

148
00:12:03.440 --> 00:12:07.759
really parrot out. They don't know
anything, they don't understand what they're saying.

149
00:12:07.320 --> 00:12:11.600
So tying that to kind of the
sources of truth, as you say,

150
00:12:11.120 --> 00:12:16.639
is another huge challenge which doesn't exist
today. Well, and also we

151
00:12:16.720 --> 00:12:22.919
have to realize there is a workflow
to these models. And what I'm learning

152
00:12:22.960 --> 00:12:24.440
from folks is you can do one
of two things. Basically, you can

153
00:12:24.480 --> 00:12:30.559
fine tune your model by embedding your
corporate data into the model, so training

154
00:12:30.600 --> 00:12:33.440
the model itself on your corporate data. Or the other very common use case

155
00:12:33.519 --> 00:12:37.200
is to use a vector database,
and that's where you store your embeddings.

156
00:12:37.559 --> 00:12:43.960
And so embeddings would be your financial
reports for your business, or your training

157
00:12:43.039 --> 00:12:48.799
modules or documents related to some process
in your business. You want to populate

158
00:12:48.840 --> 00:12:52.759
all that and ideally just use the
large language model for its text generative and

159
00:12:52.879 --> 00:12:58.960
syntactical knowledge essentially, so it's you're
using the engine to spin up the words,

160
00:13:00.279 --> 00:13:03.799
but the facts, if you will, come from the embeddings. Isn't

161
00:13:03.840 --> 00:13:09.360
that the general idea? Yeah,
and well you're touching on some of the

162
00:13:09.399 --> 00:13:13.279
more interesting aspects of large language models. So let me start by saying,

163
00:13:13.279 --> 00:13:20.799
first of all, no one understands
why they work. Of why a very

164
00:13:20.879 --> 00:13:26.399
large model that traditional statistics will tell
you should never work because it's over parameterized,

165
00:13:26.759 --> 00:13:28.240
it's got too many degrees of freedom
it can fit the data, and

166
00:13:28.279 --> 00:13:33.360
therefore no chance of learning why it
works. That's one interesting mystery, right

167
00:13:33.360 --> 00:13:37.919
that we don't write. The other
mystery is this whole embedding space. Why

168
00:13:39.000 --> 00:13:41.080
is it that if I took a
group of words, or a group of

169
00:13:41.120 --> 00:13:46.440
sentences, or a group of paragraphs
and map them into this very large vector

170
00:13:46.480 --> 00:13:50.639
space, which is what's called embedding
a bunch of numbers. Basically, why

171
00:13:50.720 --> 00:13:58.759
is it that those embeddings capture something
about the domain that allows you to generalize,

172
00:13:58.759 --> 00:14:01.879
that allows you to build on knowledge. So that representation is another mystery

173
00:14:03.000 --> 00:14:07.759
eric that we have people working on
trying to demystify and understand what the heck

174
00:14:07.840 --> 00:14:11.679
is going on. Those two aspects
are some of the deeper aspects of the

175
00:14:11.720 --> 00:14:18.279
technology that represent an academic and research
challenge trying to understand what's going on and

176
00:14:18.320 --> 00:14:22.120
how do these models work? The
fact is though they work, and therefore

177
00:14:22.840 --> 00:14:26.120
how do you best leverage them?
And then how do you add these embeddings

178
00:14:26.159 --> 00:14:31.240
with the other generative stuff which is
straightforward and well understood. The autocomplete technology

179
00:14:31.279 --> 00:14:35.200
if you like, that kind of
takes an input a prompt and produces the

180
00:14:35.240 --> 00:14:39.120
output. The other thing that I'll
mention here, which is the third big

181
00:14:39.159 --> 00:14:46.120
mystery in the field, is the
prompt themselves. How do you come up

182
00:14:46.159 --> 00:14:52.200
with the right prompt. We know
there's a super high sensitivity between prompt variations

183
00:14:52.200 --> 00:14:58.600
and output variations, so small disturbances
of the prompt may produce a completely different

184
00:14:58.759 --> 00:15:01.960
answer. Some people claim that,
hey, these are the new hot jobs.

185
00:15:03.000 --> 00:15:07.360
There was a New York Times and
Financial Times articles on you this is

186
00:15:07.399 --> 00:15:13.480
the highest paid jobs, hottest jobs. My colleague and a core member of

187
00:15:13.519 --> 00:15:20.519
the faculty of the Institute for Experiential
AI at Northeastern Byron Wallace, summarized it

188
00:15:20.679 --> 00:15:26.759
very nicely in our recent conference where
he said, this is not prompt engineering.

189
00:15:26.639 --> 00:15:33.159
This is more like incantations and black
magic, right, and you know,

190
00:15:33.200 --> 00:15:37.480
you might stumble on the right incantation, you might not. And he

191
00:15:37.600 --> 00:15:41.080
has a field of study that has
to do with Okay, how do we

192
00:15:41.159 --> 00:15:45.000
reduce that dependence, how do we
reduce that sensitivity, because at the end

193
00:15:45.039 --> 00:15:48.799
of the day, to make it
practical and useful, you can't have this

194
00:15:48.960 --> 00:15:56.600
huge variation and kind of weird stuff. And he says incantations because sometimes prefixes

195
00:15:56.639 --> 00:16:00.159
to prompts that you might you and
I might think have nothing to do with

196
00:16:00.200 --> 00:16:03.399
it end up being very useful to
guide the model to the right space.

197
00:16:03.639 --> 00:16:07.320
So it's people are adding weird stuff
in the front and in the end of

198
00:16:07.360 --> 00:16:11.720
a prompt just to make it work. Right. That is not a science.

199
00:16:11.759 --> 00:16:15.559
That is not engineering. We got
to figure this out. That's funny.

200
00:16:15.559 --> 00:16:18.000
You know, I'm gonna I'm gonna
close this opening segment here with a

201
00:16:18.039 --> 00:16:25.039
fun story and basically say that this
is not as crazy and different and unique

202
00:16:25.080 --> 00:16:26.840
as some people might think. So, I had a business partner for a

203
00:16:26.879 --> 00:16:30.120
number of years back in the year
two thousand and I think earlier you were

204
00:16:30.120 --> 00:16:33.159
talking twenty men twenty twenty, I
think when they started this whole thing.

205
00:16:33.240 --> 00:16:37.399
So it's five years ago. But
back in two thousand, I was working

206
00:16:37.440 --> 00:16:41.799
for a company Keem Jim, a
couple Russians, and the good Russian was

207
00:16:41.840 --> 00:16:45.200
named Nikola niko Lem like this one
thing. He said to me. The

208
00:16:45.279 --> 00:16:49.080
funny thing about the web is nobody
know how it works. All we know

209
00:16:49.279 --> 00:16:53.960
is it works right, And it's
the same thing. So now what's interesting

210
00:16:55.080 --> 00:17:00.240
is if you look at Internet and
cloud and hybrid architecture and all this that

211
00:17:00.240 --> 00:17:03.160
people are talking about today, containerization, Kubernetes, all that stuff, Well

212
00:17:03.160 --> 00:17:08.279
what happened is Google came along with
Kubernetes as this orchestration layer to sit on

213
00:17:08.319 --> 00:17:12.480
top of Docker as a way to
containerize applications make them really really small,

214
00:17:12.799 --> 00:17:17.400
which is like the vision of service
storryed architecture, but a different bus,

215
00:17:17.440 --> 00:17:19.839
if you will. And what's happening
is now we have all this data,

216
00:17:19.960 --> 00:17:23.559
we have tons and tons. It's
called observability. So what's happening now is

217
00:17:23.599 --> 00:17:27.720
we're finally starting to kind of figure
out how the web actually works. So

218
00:17:27.759 --> 00:17:30.920
I'm not saying we didn't know anything, but there are a lot of stuff

219
00:17:30.920 --> 00:17:34.680
that we didn't really fully understand about
how the web actually works, like which

220
00:17:34.720 --> 00:17:38.799
ports are important in what particular process. I mean, there's all this stuff

221
00:17:38.880 --> 00:17:42.920
going on. So my point view
is that this AI stuff, You're right,

222
00:17:42.960 --> 00:17:45.839
we don't know how it works.
It's not the first time we've found

223
00:17:45.880 --> 00:17:48.680
some engine that works and we're like, well, we don't know exactly how

224
00:17:48.720 --> 00:17:52.400
it works, but it sure does
work. But folks, don't touch that

225
00:17:52.480 --> 00:17:56.119
dollar. We'll be right back talking
to Osama Faiad of the AI Institute from

226
00:17:56.119 --> 00:18:08.200
the Experiential AI Institute, will be
right back. Welcome back to Inside Analysis.

227
00:18:08.680 --> 00:18:15.240
Here's your host Eric Tavanaugh. All
right, folks, back here on

228
00:18:15.359 --> 00:18:21.920
Inside Analysis, your host Eric Kavanaugh
with doctor Usama Fayad from Northeastern Universities the

229
00:18:22.039 --> 00:18:26.960
Institute for Experiential AI, which basically
just means humans in the loop. Humans

230
00:18:26.960 --> 00:18:30.200
are really important. Even with self
driving cars, you have to give correction

231
00:18:30.279 --> 00:18:33.480
to the vehicle. I've actually driven
a couple of these now, and it's

232
00:18:33.559 --> 00:18:37.960
kind of freaky when it starts taking
the wheel from you and like, WHOA,

233
00:18:37.039 --> 00:18:40.839
what are you doing. I didn't
even know it was self driving.

234
00:18:40.880 --> 00:18:42.119
It was a rental car. I'm
like, oh, what's going on here.

235
00:18:42.240 --> 00:18:45.680
Okay, it's taking over for me
and seeing you shouldn't be doing something.

236
00:18:47.480 --> 00:18:49.920
We're getting there. It's I mean, thanks to innovations, thanks to

237
00:18:49.960 --> 00:18:55.799
Elon Musk and Tesla and all these
innovations. Those folks are doing pretty cool

238
00:18:55.839 --> 00:19:00.240
stuff. But doctor Fayad, we
wanted to dive into the critical six sess

239
00:19:00.359 --> 00:19:03.880
factors, and one absolutely positively is
the data. You know, I keep

240
00:19:03.960 --> 00:19:08.039
cautioning people, and we're going to
probably dive into this more in our own

241
00:19:08.079 --> 00:19:12.920
business of consulting with organizations. Be
very careful how you train your models.

242
00:19:12.960 --> 00:19:15.319
You want to train your model on
your data. You want to give it

243
00:19:15.440 --> 00:19:18.759
curated, trusted data. You don't
want to just point it at your whatever

244
00:19:18.799 --> 00:19:22.440
information system you have, because there's
a bunch of junk in there. And

245
00:19:22.519 --> 00:19:27.039
anyone who's ever done data quality initiative
knows. If you get an executive doesn't

246
00:19:27.039 --> 00:19:30.039
believe the data is crap, just
show them. Just open up any database

247
00:19:30.079 --> 00:19:33.400
and point to any corner of it, and you'll find some problem somewhere.

248
00:19:33.440 --> 00:19:37.440
I promise, it's just everywhere,
whether it's data in the wrong fields,

249
00:19:37.480 --> 00:19:41.920
data in the wrong format, data
that was truncated during some ETL script or

250
00:19:41.920 --> 00:19:45.359
something, and it's just everywhere.
There's bad data everywhere. So getting the

251
00:19:45.440 --> 00:19:48.480
right data is really important, doctor
Fiab, How do you do that?

252
00:19:48.519 --> 00:19:52.920
What do you recommend? When people
come to you and ask you what are

253
00:19:52.960 --> 00:19:55.519
the best practices? What do you
tell them? Yeah? Great, great

254
00:19:55.599 --> 00:20:00.400
question. So so data is a
whole ocean to open up. There's the

255
00:20:00.519 --> 00:20:06.440
collecting the data at quality. That's
collecting data at the right level of granularity.

256
00:20:06.799 --> 00:20:10.160
So we have spent decades, as
you well know, Eric, in

257
00:20:10.200 --> 00:20:14.920
the whole area of data warehousing and
kind of information systems management and so forth,

258
00:20:14.920 --> 00:20:22.319
and data management thinking about data for
use by humans. Now what the

259
00:20:22.359 --> 00:20:26.640
AI haves. The AI first companies
or the companies who really are using AI

260
00:20:26.720 --> 00:20:32.240
today, they have a deeper understanding
that, Hey, machines needed data at

261
00:20:32.240 --> 00:20:37.359
a different level. They need high
granularity. Not necessarily stuff understandable by humans,

262
00:20:37.599 --> 00:20:41.039
but every little detail of what happened, every log of every event,

263
00:20:42.039 --> 00:20:48.240
every kind of intervention correction by a
human, et cetera. Captured. This

264
00:20:48.319 --> 00:20:52.000
is not how most organizations or companies
work today. So that is one big

265
00:20:52.039 --> 00:20:56.799
shift that needs to happen. The
second thing is, if you notice a

266
00:20:56.839 --> 00:21:00.799
lot of the literature and lower in
the world of data has to do with

267
00:21:00.839 --> 00:21:04.519
structured data. Well, according to
Gartner and many others, the majority of

268
00:21:04.599 --> 00:21:10.400
data in any organization, in any
company is actually unstructured data. If you

269
00:21:10.480 --> 00:21:15.400
actually talk to the IT teams and
the technology teams and so forth, they

270
00:21:15.440 --> 00:21:18.480
only typically only know how to deal
with structured data. And they hit unstructured

271
00:21:18.559 --> 00:21:22.799
data, they're stuck. It's like
in the headlights. We have to change

272
00:21:22.839 --> 00:21:27.799
this dynamic to understand that, hey, we got to embrace the unstructured data,

273
00:21:27.880 --> 00:21:33.319
the text, the images, the
video, the audio, interactions in

274
00:21:33.319 --> 00:21:40.559
the customer center, all the conversations
and leverage that information properly, and that

275
00:21:40.680 --> 00:21:44.599
needs. There exists many good tools
out there. Most of them also,

276
00:21:44.759 --> 00:21:47.839
fortunately are in the open source,
which is something I'm very big fan of.

277
00:21:48.319 --> 00:21:51.839
But you can put these things together
in solutions that work for businesses,

278
00:21:52.119 --> 00:21:55.839
and this is something we're very passionate
about. How do we partner to figure

279
00:21:55.839 --> 00:21:59.440
out, how we help our partners
figure out how to approach an area,

280
00:21:59.480 --> 00:22:03.799
how to the data, et cetera. Now that leads me to another very

281
00:22:03.920 --> 00:22:10.079
very important data that's information or aspect
that's very related to data, which is

282
00:22:10.680 --> 00:22:14.839
kind of many companies out there.
Think that all we have to do is

283
00:22:14.880 --> 00:22:21.720
wait for Microsoft or open AI or
whoever to refine their models so that they

284
00:22:21.720 --> 00:22:25.519
can meet our needs. What happens
then, is you become users of the

285
00:22:25.559 --> 00:22:30.559
technology. You don't have your own
private language model. You don't have the

286
00:22:30.599 --> 00:22:34.119
ability to capture all these corrections.
You mentioned corrections when you're driving right,

287
00:22:34.559 --> 00:22:37.880
Well, it's great to intervene and
correct, but if you don't capture that

288
00:22:37.960 --> 00:22:42.599
correction, the context for that correction, why you made that correction, your

289
00:22:42.640 --> 00:22:47.400
model is never going to benefit from
it unless you're willing to share it back

290
00:22:47.920 --> 00:22:52.599
with the company who's providing this public
utility, this large language model. Often

291
00:22:52.640 --> 00:22:57.240
this information is sensitive, it's proprietary. I argue it's one of the most

292
00:22:57.359 --> 00:23:03.359
valuable sources of IP than any company
operating in any space has, which is

293
00:23:03.680 --> 00:23:07.440
kind of the secret sauce. How
do our processes work? Where do we

294
00:23:07.480 --> 00:23:11.039
intervene, why do we fix what
is our way of doing things? Today?

295
00:23:11.119 --> 00:23:14.880
Companies let all of that go into
the exhaust by never capturing it.

296
00:23:15.319 --> 00:23:18.440
In reality, you need to be
able to not just capture it, but

297
00:23:18.559 --> 00:23:22.880
feed it back. And feeding it
back leads us to another whole interesting area,

298
00:23:22.240 --> 00:23:26.359
which means you really need to have
your own private language model. That

299
00:23:27.519 --> 00:23:33.440
model can't be too huge, like
the likes of open ai or chag GPT

300
00:23:33.759 --> 00:23:37.319
or any of these. It has
to be as small as possible to maintain

301
00:23:37.359 --> 00:23:44.079
its stability, to maintain its maintainability
and its reliability and so forth. But

302
00:23:45.200 --> 00:23:48.400
it needs to get feedback. So
you need to be able to capture all

303
00:23:48.440 --> 00:23:52.079
that feedback, all those corrections,
and feed them right back and retrain that

304
00:23:52.000 --> 00:23:56.680
language model in an affordable way to
maintain your own IP, your own knowledge,

305
00:23:56.680 --> 00:24:02.720
and so forth. Now, many
of the companies who offer the large

306
00:24:02.759 --> 00:24:07.559
language models will offer you the chance, of course, for money, to

307
00:24:07.680 --> 00:24:11.480
kind of fine tune the model or
create your own tuned version of the model,

308
00:24:11.599 --> 00:24:15.720
and there we can talk about technically
what happens. All they do is

309
00:24:15.720 --> 00:24:19.359
they cut off the output layer,
they throw it out, and retrain an

310
00:24:19.359 --> 00:24:25.599
output layer they call these adapters that
is kind of fine tuned just to your

311
00:24:25.640 --> 00:24:30.519
problem and on your feedback. But
we know that that's less effective than retraining

312
00:24:30.559 --> 00:24:36.319
the model and reintegrating that data into
the whole stack. So coming up with

313
00:24:36.359 --> 00:24:40.279
a large language model, your own
private language model, is not impossible.

314
00:24:40.400 --> 00:24:44.599
It is very doable. Many tools
are available out there, whether you're talking

315
00:24:44.759 --> 00:24:48.680
Falcon which came from Abudawi or Lama
or Lama too. I'm a big fan

316
00:24:48.759 --> 00:24:53.480
of at Facebook or Meta because they
actually not only did they make it open

317
00:24:53.519 --> 00:24:59.079
source, they gave a commercial open
source license. So companies, sorry,

318
00:24:59.240 --> 00:25:03.559
yeah, and that opens up a
whole other approach and a whole other mechanism.

319
00:25:03.599 --> 00:25:06.839
But to do that and to take
advantage of it, you need to

320
00:25:06.880 --> 00:25:10.440
have the talent and then know how
of how do I use that language model,

321
00:25:10.480 --> 00:25:11.720
when do I find tune it?
How large does it have to be?

322
00:25:11.880 --> 00:25:15.920
How small can I make it?
All of these questions are the kinds

323
00:25:15.960 --> 00:25:21.559
of questions we partner with companies on
kind of answering and figure out what is

324
00:25:22.119 --> 00:25:26.799
the right approach here in this very
very confusing, and I might add fast

325
00:25:26.839 --> 00:25:30.039
evolving space. I mean I try
to keep up and I am challenged,

326
00:25:30.400 --> 00:25:36.519
right every week, every month there's
new innovations, new ways of doing things.

327
00:25:36.720 --> 00:25:41.039
Somebody's crapping something old and saying it
didn't work. Actually it was part

328
00:25:41.079 --> 00:25:44.599
of the system, but now we've
got to do something completely new. All

329
00:25:44.680 --> 00:25:48.799
good, all healthy, but it's
a challenge to keep up with the latest,

330
00:25:48.880 --> 00:25:52.279
greatest and the right best practice of
how to use it. Yeah,

331
00:25:52.680 --> 00:25:56.640
start with I just want to remind
everybody that data that you're letting go into

332
00:25:56.680 --> 00:26:02.279
the exhaust by not capturing it systematically. That is gold that is actually more

333
00:26:02.359 --> 00:26:06.960
valuable than gold. It's not just
oil. It's more valuable than gold,

334
00:26:07.039 --> 00:26:11.480
and you're letting it go out in
the exhaust and losing all that ip that

335
00:26:11.599 --> 00:26:15.079
your teams, your employees are working
on day in, day out and keeping

336
00:26:15.119 --> 00:26:17.880
operations going. Yeah. Well,
you know, I just as you were

337
00:26:17.920 --> 00:26:21.000
talking, I thought of an interesting
metaphor here, and I'll throw it at

338
00:26:21.039 --> 00:26:25.440
you. I think you'll you'll find
it interesting. So think about language and

339
00:26:25.480 --> 00:26:29.519
feeding a dictionary of English to something
like one of these large language models.

340
00:26:29.519 --> 00:26:33.240
You feed it. The dictionary knows
all the different words, but people don't

341
00:26:33.279 --> 00:26:37.720
talk the way they write, and
people don't talk the way dictionaries are written.

342
00:26:37.759 --> 00:26:40.720
Necessarily, there's a lot of slang, there are lots of expressions,

343
00:26:40.759 --> 00:26:44.640
idioms, all these different things that
don't fit neatly into the definitions of words.

344
00:26:44.880 --> 00:26:48.480
But it's how you actually speak when
you're talking to someone. And that's

345
00:26:48.519 --> 00:26:52.160
the unstructured data, really, er, that's the unstructured syntax. Almost of

346
00:26:52.240 --> 00:26:56.640
how people actually interact with words on
the phone in person, and certain cultures

347
00:26:56.680 --> 00:27:02.720
and in certain regions, and so
if if you don't absorb that that contextual

348
00:27:02.759 --> 00:27:06.680
information, your engine will never speak
like a normal person does. Right,

349
00:27:06.759 --> 00:27:10.880
So this is the nuance you have
to appreciate, is that what's on paper

350
00:27:11.279 --> 00:27:15.240
often is very different in the spoken
word. What do you think? Absolutely,

351
00:27:15.319 --> 00:27:18.640
and it's critical in many applications,
actually, I would argue in most

352
00:27:18.640 --> 00:27:22.799
of them, if you're dealing with
conversational AI, if you're dealing with dealing

353
00:27:22.799 --> 00:27:26.480
with humans, whether you're doing it
for marketing purposes, whether you're doing it

354
00:27:26.519 --> 00:27:33.200
for customer service, whether you're doing
it for predictive maintenance. And it's not

355
00:27:33.319 --> 00:27:36.759
only kind of the idiom or kind
of what the public, how the public

356
00:27:36.839 --> 00:27:41.039
or the customer speak, it's often
the culture within the company. Your engineering

357
00:27:41.079 --> 00:27:45.319
team has their own terminology where they
use certain terms. If you come up

358
00:27:45.319 --> 00:27:48.000
in the rights, they mean something
of it, but they're using the company

359
00:27:48.319 --> 00:27:52.559
to mean a certain aspect of a
process, a certain aspect of a system.

360
00:27:52.839 --> 00:27:56.680
Learning all of that and making it
work in the right context is a

361
00:27:56.720 --> 00:28:02.759
real interesting challenge, and it's very
possible. I mean, that's why the

362
00:28:02.839 --> 00:28:06.160
data is. That's why I describe
it as gold or more valuable than gold,

363
00:28:06.480 --> 00:28:11.640
because it contains all these instances and
it gives you the keys to allowing

364
00:28:11.680 --> 00:28:15.519
these systems to have the bigger and
correct impact on how you do acceleration,

365
00:28:15.680 --> 00:28:21.359
how you do automation, how do
you leverage generative AI to kind of give

366
00:28:21.400 --> 00:28:29.799
you that unique advantage competitive advantage out
there. I've experienced it personally in healthcare.

367
00:28:29.920 --> 00:28:33.920
Right if you're in an operating room
or in an X ray room or

368
00:28:33.960 --> 00:28:38.400
in an imaging center, you know, the language being used is completely different

369
00:28:38.440 --> 00:28:48.319
than what's being used in you know, a pathology lab or a practice for

370
00:28:48.640 --> 00:28:52.119
physical therapy or what have you.
So all of these things, all of

371
00:28:52.160 --> 00:28:56.559
these aspects are important. What contents
to use, the words in, what

372
00:28:56.599 --> 00:29:00.319
they mean, new meanings that are
not available in the dictionary. All of

373
00:29:00.359 --> 00:29:07.240
that is so important and so critical
technically, socially and marketing Whites. Yeah,

374
00:29:07.279 --> 00:29:11.160
well, and there's also tone of
voice. And I remember, going

375
00:29:11.160 --> 00:29:15.599
all the way back to like third
grade or something, our teacher asked us

376
00:29:15.599 --> 00:29:18.519
a question and it stumped all of
us, and I thought about it later,

377
00:29:18.519 --> 00:29:21.319
I was like, Oh, tone
of voice, that's what he was

378
00:29:21.319 --> 00:29:23.640
talking about You can say something and
mean it honestly, or you can say

379
00:29:23.680 --> 00:29:27.559
it sarcastically. And usually it's the
tone of voice and the context that will

380
00:29:27.559 --> 00:29:30.799
tip off a human being. But
it's going to be harder for a machine

381
00:29:30.839 --> 00:29:34.920
to understand that, but not if
it has the intonation that if you capture

382
00:29:34.960 --> 00:29:38.119
the audio and it can kind of
pick up on that sort of thing.

383
00:29:38.200 --> 00:29:44.799
But these are the sort of deeply
understood aspects of communication that we just know

384
00:29:44.960 --> 00:29:48.319
as grown ups, that we've learned
over time. I mean, I remember,

385
00:29:48.759 --> 00:29:51.079
this is funny, but probably most
people don't remember this, but I

386
00:29:51.119 --> 00:29:56.119
remember the first time I really experienced
sarcasm because I was like two or three

387
00:29:56.200 --> 00:30:00.480
or something and I'd spilled milk and
my mother and I knew it was like,

388
00:30:00.559 --> 00:30:03.039
oh no, this is bad.
And my mother looked at me,

389
00:30:03.319 --> 00:30:06.920
she goes, well, that's just
great, and I was like, hold

390
00:30:06.960 --> 00:30:11.640
on, I know that great is
good, but that tone is not good.

391
00:30:11.240 --> 00:30:15.839
What's happening here? And I was
really defuddled as a child. But

392
00:30:15.920 --> 00:30:19.400
that was my first experience with sarcasm. Right, I'll give you here.

393
00:30:19.799 --> 00:30:23.480
You're touching on one of the biggest
challenges that, for example, search engines

394
00:30:23.519 --> 00:30:27.839
today face, which are memes.
Right, you hit a meme on the

395
00:30:27.880 --> 00:30:32.680
Internet, the search engine thinks it
means one thing. Reality is, it's

396
00:30:32.680 --> 00:30:36.920
something completely different. And the only
way you catch that, The only way

397
00:30:37.000 --> 00:30:41.599
you catch that today is human intervention. Search editorial team saying, wait a

398
00:30:41.640 --> 00:30:44.119
second, this shouldn't be in your
search result. This is a meme.

399
00:30:44.319 --> 00:30:48.480
It means something completely different. It
has nothing to do with biology or whatever

400
00:30:48.599 --> 00:30:51.680
it's it's being used to refer to
a group of people or a group of

401
00:30:51.720 --> 00:30:55.880
practices or what have you. So
those are those are real things that matter

402
00:30:56.000 --> 00:31:00.359
a lot, and you have to
have mechanisms for catching them. And you

403
00:31:00.400 --> 00:31:03.319
know, in a way, you
can argue that a lot of the internal

404
00:31:03.359 --> 00:31:07.160
culture of a company's operations team,
a company's engineering team, a company's business

405
00:31:07.160 --> 00:31:11.079
team, a company's sales team,
a company's marketing team is filled with these

406
00:31:11.119 --> 00:31:15.480
memes. We don't call them memes, but they are there. That's funny.

407
00:31:15.640 --> 00:31:19.000
There's actually a great Seinfeld episode where
George Costanza, he's working for the

408
00:31:19.079 --> 00:31:25.640
Yankees and I think he was talking
with it was like the Texas Rangers or

409
00:31:25.640 --> 00:31:29.599
something, and as a baseball team, and the culture of the executives for

410
00:31:29.680 --> 00:31:33.839
that team loved using curse words.
So they were insulting each other all the

411
00:31:33.880 --> 00:31:37.640
time, but as a compliment.
And then George's boss walks by as he's

412
00:31:37.680 --> 00:31:41.119
on the phone using all these expletives, and the boss is like, oh

413
00:31:41.160 --> 00:31:44.359
no, what are you doing.
This is terrible. In reality, he's

414
00:31:44.359 --> 00:31:48.799
bonding with these guys because the context
was there. Classic dramatic irony, right,

415
00:31:48.799 --> 00:31:52.440
the one guy didn't know the context. But this is the importance of

416
00:31:52.559 --> 00:31:56.640
nuance and understanding context. And to
your point, just getting it back to

417
00:31:56.680 --> 00:32:01.519
training AI your own private language model, which is what most businesses are going

418
00:32:01.599 --> 00:32:06.759
to want to have. I'm pretty
sure you have to bake in that kind

419
00:32:06.799 --> 00:32:08.640
of stuff. You have to have
that human in the loop that goes,

420
00:32:08.680 --> 00:32:13.039
okay, no, this is an
example of sarcasm or these guys are just

421
00:32:13.079 --> 00:32:16.279
being goofy or whatever in their context. Calling you a jerk or a so

422
00:32:16.400 --> 00:32:20.880
and so is actually a compliment.
That's the human invention part, right,

423
00:32:21.319 --> 00:32:28.319
absolutely, absolutely, And I remember
that episode. But the thing I wanted

424
00:32:28.319 --> 00:32:32.319
to mention is, you know,
language, the language use is very different

425
00:32:32.400 --> 00:32:37.680
than kind of the formal language defined. That is why I believe, like

426
00:32:37.759 --> 00:32:44.920
you, most companies will find out
that achieve true competitive advantage to have differentiation,

427
00:32:45.480 --> 00:32:47.960
You're going to have your own private
language models, which opens up a

428
00:32:47.960 --> 00:32:52.160
whole new area. We are a
company that's in the you know, quick

429
00:32:52.200 --> 00:32:57.680
service restaurant business, or we are
a company that's doing manufacturing, what do

430
00:32:57.759 --> 00:33:00.279
we know about large language model AI, et cetera, and how do we

431
00:33:00.319 --> 00:33:05.240
approach it? And part of this
demonstification and kind of working with you on

432
00:33:05.279 --> 00:33:09.839
the AI journey is one thing our
institute is very passionate about because that actually

433
00:33:09.839 --> 00:33:15.839
puts us right up front on the
front lines of getting the applications to work

434
00:33:15.880 --> 00:33:19.400
and helping our partners getting it to
work. And through that we will learn

435
00:33:19.440 --> 00:33:22.920
what the real research problems are,
where the real challenges are, and we

436
00:33:22.960 --> 00:33:28.759
will learn what should we be teaching
our students and our corporate learners in order

437
00:33:28.799 --> 00:33:32.319
to be effective in that world.
So that whole area of the infrastructure,

438
00:33:32.640 --> 00:33:39.319
the talent that know how you will
need to build your own private language models

439
00:33:39.359 --> 00:33:43.920
to capture your own IP and your
own knowledge. It's going to be a

440
00:33:43.960 --> 00:33:46.839
whole, huge new trend that no
one out there is paying attention to in

441
00:33:46.880 --> 00:33:52.799
the business world. They're focused on
being of these public right, No,

442
00:33:52.960 --> 00:33:53.880
that's the very very good point.
Well, folks, we're talking to doctor

443
00:33:53.920 --> 00:33:57.599
Usama Fai will be right back.
Don't touch that doubt. You're listening to

444
00:33:57.640 --> 00:34:09.639
Inside Analysis. Welcome back to Inside
Analysis. Here's your host, Eric Tavanaugh.

445
00:34:13.320 --> 00:34:16.159
All right, folks, back here
on Inside Analysis talking all about experiential

446
00:34:16.280 --> 00:34:22.679
AI. That's humans in the Loop
with doctor Usama Fayad of the Northeastern University

447
00:34:23.519 --> 00:34:28.320
Institute for Experiential AI. Got to
get the all clean and straight in my

448
00:34:28.360 --> 00:34:30.679
head. The Institute for Experiential AI, it just means humans in the loop

449
00:34:30.719 --> 00:34:35.639
basicly. But we're talking about the
power of these models, which is amazing.

450
00:34:35.679 --> 00:34:37.440
We're talking about how a lot of
people don't really know how they work,

451
00:34:37.480 --> 00:34:40.559
including people who design them. We
just see that they work, and

452
00:34:40.639 --> 00:34:45.880
so we're coming up with strategies to
help companies really harness the technology. And

453
00:34:45.880 --> 00:34:49.280
I think that's the key, right
is this is if unbridled, this thing

454
00:34:49.320 --> 00:34:52.599
will go wacko and cause you all
kinds of trouble. But if you harness

455
00:34:52.639 --> 00:34:55.639
it effectively, and that's your governance
structure, it's your embeddings. It's using

456
00:34:55.679 --> 00:35:01.480
ontologies and I mentioned my friends from
Praxy Data who are using ontologies to really

457
00:35:01.920 --> 00:35:07.079
narrow the focus of their work and
thus be able to deliver some tangible business

458
00:35:07.199 --> 00:35:10.639
value. And you know that's always
been the case with machine learning and AI

459
00:35:10.840 --> 00:35:15.000
is you have to have a very
narrow focus and you have to know what

460
00:35:15.039 --> 00:35:19.159
you're trying to accomplish. I mean, you can use deep learning to discover

461
00:35:19.360 --> 00:35:22.440
things, but even then you're trying
to discover things. So you have to

462
00:35:22.519 --> 00:35:27.719
have a purpose and a mission and
an objective upon which you train these models

463
00:35:28.039 --> 00:35:30.639
to be able to understand if you're
getting anywhere right. Tell us a bit

464
00:35:30.679 --> 00:35:36.000
about that and why ontologies matter,
Yosama, Well, I mean, ontologies

465
00:35:36.320 --> 00:35:40.039
give you the basics of understanding domains. There are many things you have to

466
00:35:40.039 --> 00:35:45.079
worry about. And by the way, not coincidentally, a lot of the

467
00:35:45.159 --> 00:35:53.000
work in generative AI now is using
what we call the RAG framework, which

468
00:35:53.079 --> 00:35:59.760
stands for retrieval augmented generation, which
is another form of kind of enforcing the

469
00:35:59.760 --> 00:36:04.719
mode to stay within certain guardrails and
to say here's what we mean. And

470
00:36:04.800 --> 00:36:07.880
it's aided by kind of retrieving the
right references and all that to help it

471
00:36:08.559 --> 00:36:15.000
from straying, both in training and
in answering. But anyway, you know,

472
00:36:15.119 --> 00:36:20.280
working in different domains you need to
figure out ways of bringing in that

473
00:36:20.440 --> 00:36:25.000
domain knowledge, that domain into into
the problem in order to constrain the space

474
00:36:25.360 --> 00:36:29.760
in order to focus it. You
know, I was talking about the secrets

475
00:36:29.800 --> 00:36:31.880
of making AI work, and we
talked about the humans in the loop or

476
00:36:32.039 --> 00:36:37.159
must interventions are necessary? Data at
the right level is absolutely necessary. Most

477
00:36:37.239 --> 00:36:45.159
data is being lost. But I
will I will add the last bit is

478
00:36:45.800 --> 00:36:51.239
how do we bring in non domain
knowledge and how do we narrow the problem

479
00:36:51.320 --> 00:36:55.760
down to a very narrow, very
specific problem. Because trying to solve AI

480
00:36:55.880 --> 00:37:02.119
in general has proven to be an
impossible tasks, a very difficult task for

481
00:37:02.199 --> 00:37:07.920
the past seven decades, and it
will continue to be that, and our

482
00:37:07.960 --> 00:37:14.119
best prescription for fixing it is by
narrowing the problem and resorting to what we

483
00:37:14.239 --> 00:37:19.000
call narrow AI versus general AI.
General AI is still far out of reach.

484
00:37:19.119 --> 00:37:24.039
Narrow AI has worked and has been
working for many years and decades actually

485
00:37:24.239 --> 00:37:29.199
in different domains, and the key
there is domain knowledge. The key there

486
00:37:29.280 --> 00:37:32.639
is narrowing it enough so you know
that not every possibility is here, and

487
00:37:32.679 --> 00:37:36.199
you don't have to worry about every
language, and you don't have to worry

488
00:37:36.239 --> 00:37:40.639
about every variation and every meaning no, restrict yourself to this domain. Here's

489
00:37:40.679 --> 00:37:45.800
what things mean. Try to get
complete knowledge around that domain. And that's

490
00:37:45.800 --> 00:37:50.599
a secret of making it work.
That only comes by figuring out what are

491
00:37:51.119 --> 00:37:53.440
the use cases, what are the
domains of applications, and how do I

492
00:37:53.480 --> 00:37:58.960
come up with a methodology that allows
me to build solutions for each specific use

493
00:37:58.960 --> 00:38:01.840
case. Which is why we kind
of focused on things like AI plus health.

494
00:38:02.199 --> 00:38:07.719
How do we get healthcare applications into
AI? How do we leverage digital

495
00:38:07.760 --> 00:38:12.599
health? You know, we live
in a world today, think about at

496
00:38:12.639 --> 00:38:16.400
the highest level. You go see
your doctor, hopefully once a year,

497
00:38:16.519 --> 00:38:21.679
maybe twice a year, right only
when something is wrong other than the routine

498
00:38:21.760 --> 00:38:24.920
visits. And what happens between these
routine visits is a total mystery to the

499
00:38:24.960 --> 00:38:31.559
provider, to your physician, or
because who knows what happened to Eric during

500
00:38:31.599 --> 00:38:35.920
this one year? Now Eric needs
to fill me in, and who knows

501
00:38:35.920 --> 00:38:38.559
whether Eric can fill me in completely
or skip a whole bunch of important things.

502
00:38:39.039 --> 00:38:42.559
Well, guess what. We live
in a world now where there's a

503
00:38:42.559 --> 00:38:45.639
lot of instruments. We've got these
watches, we've got the mobile phones,

504
00:38:45.679 --> 00:38:50.519
we've got the you know, all
sorts of devices, cameras, other things

505
00:38:50.559 --> 00:38:53.519
in our homes. They can actually
capture a lot of the information that typically

506
00:38:53.559 --> 00:38:58.719
is only captured in a clinic.
So how do we leverage that to close

507
00:38:58.719 --> 00:39:01.239
the loop at the highest level that
we call that digital health. Another form

508
00:39:01.280 --> 00:39:09.239
of digital health is things like oncology, or actually more generally pathology. How

509
00:39:09.599 --> 00:39:14.679
do we bring these digital techniques down
to a pathology lab where somebody is taking

510
00:39:14.679 --> 00:39:19.519
a sample, looking at it in
a microscope and kind of storing it in

511
00:39:19.599 --> 00:39:22.519
this glass slide. Well, guess
what digitization can help a lot. You

512
00:39:22.519 --> 00:39:28.039
can look at it with instruments and
with algorithms instead of just human eyes,

513
00:39:28.239 --> 00:39:31.440
instead of just relying on staff to
do you can strike capturing what's happening in

514
00:39:31.480 --> 00:39:37.079
the heads of these very talented pathologists
who are looking at an image and concluding

515
00:39:37.119 --> 00:39:42.559
a result. This happened in radiology
in a big way. Radiology has been

516
00:39:42.599 --> 00:39:45.719
digitized. It benefits from a lot
of these things. In pathology, it

517
00:39:45.760 --> 00:39:50.239
hasn't happened. Pathology is still practiced
like it was, you know, three

518
00:39:50.280 --> 00:39:52.920
or four decades ago. It can
benefit from a lot of digitization, which

519
00:39:52.960 --> 00:39:57.119
brings in a lot of AI and
a lot of use of these things.

520
00:39:58.039 --> 00:40:02.199
In life sciences, we have many
use cases, whether it's the acceleration of

521
00:40:02.280 --> 00:40:08.000
drug discovery, how do you design
how do you accelerate the design of the

522
00:40:08.039 --> 00:40:15.039
next antibody? Right antibodies are very
effective ways of coming up with cures,

523
00:40:15.239 --> 00:40:21.039
vaccines, ways to mitigate disease spread. But it typically takes about three years

524
00:40:21.039 --> 00:40:24.119
to come up with the right antibody
for something. So it wasn't a viable

525
00:40:24.159 --> 00:40:29.360
solution for COVID back in twenty twenty
or twenty twenty one because it took three

526
00:40:29.400 --> 00:40:32.039
years. But if I can get
that three years down to three months by

527
00:40:32.079 --> 00:40:37.000
accelerating it through compute and through modeling, amazing things can happen. Right,

528
00:40:38.079 --> 00:40:45.480
take things like that happen in a
hospital or a physician when you're being taken

529
00:40:45.559 --> 00:40:49.880
care of. In one of the
projects we have is called heart, which

530
00:40:49.960 --> 00:40:54.719
expands to something involving AI. But
what it is one of the applications and

531
00:40:54.880 --> 00:40:59.960
working with a large hospital system,
is how do you determine when somebody should

532
00:41:00.119 --> 00:41:04.559
not be downgraded from critical care?
They're in the ICU and things look good,

533
00:41:04.960 --> 00:41:07.519
we can downgrade them, but the
algorithms can tell you no, no,

534
00:41:07.519 --> 00:41:12.360
no, no, there's a huge
danger here if you downgrade that issues

535
00:41:12.400 --> 00:41:15.519
would recur, which is horrible for
the patient, horrible for the hospital,

536
00:41:15.599 --> 00:41:20.679
horrible for the physician. And you
want to make that decision when you're absolutely

537
00:41:20.719 --> 00:41:23.639
sure that no complications will happen.
Similarly, you don't want to keep them

538
00:41:23.639 --> 00:41:29.119
in ICU too long because that's not
great for the patient and it's a use

539
00:41:29.159 --> 00:41:31.719
of a very limited resource, the
ICU. So a lot of these problems

540
00:41:31.800 --> 00:41:36.920
come up in a lot of places. And there with my colleague Ray Winslow,

541
00:41:36.960 --> 00:41:40.280
who's also a core member of the
faculty in the Experiential AI Institute,

542
00:41:42.639 --> 00:41:46.960
he leverages what's called computational medicine.
How do we model these things? How

543
00:41:47.000 --> 00:41:50.800
do we model processes, how do
we model patients? How do we model

544
00:41:50.800 --> 00:41:53.880
diseases and use those to come up
with predictions that are hard to get otherwise.

545
00:41:54.360 --> 00:42:00.320
So those are applications in healthcare,
in life sciences. And our newest

546
00:42:00.360 --> 00:42:07.079
area is the climate and sustainability.
How do we leverage kind of new data

547
00:42:07.119 --> 00:42:10.800
sources and models, et cetera.
Yeah, And to drill real quick into

548
00:42:10.920 --> 00:42:15.960
this, the ICU example, which
is a fantastic one I'd like to point

549
00:42:15.960 --> 00:42:17.480
out for our audience, And let's
get some feedback on this, se me

550
00:42:17.480 --> 00:42:22.440
an email info at dmradio dot biz
or info inside Analysis dot com both come

551
00:42:22.480 --> 00:42:25.159
to me. You know, at
any point in time in a workflow,

552
00:42:25.360 --> 00:42:30.840
you can insert an AI suggestion,
for example, or you can have it

553
00:42:30.880 --> 00:42:35.639
sort of waiting to make a suggestion. So anytime the clinician is there and

554
00:42:35.679 --> 00:42:38.760
they're typing information, they're doing stuff
as you go to the screen where you're

555
00:42:38.800 --> 00:42:45.159
going to think about maybe downgrading this
person from ICU, that's an excellent point

556
00:42:45.239 --> 00:42:49.400
at which to offer an opportunity for
the AI to give a suggestion. Right,

557
00:42:49.400 --> 00:42:52.480
So it'll come in and say,
I see you've just launched this screen,

558
00:42:52.840 --> 00:42:55.159
maybe you should reconsider because of X
y Z. And I think that's

559
00:42:55.159 --> 00:42:59.960
really where a lot of this AI
value is going to manifest is in suggestions

560
00:43:00.039 --> 00:43:04.519
at certain points in a workflow.
Right, absolutely, Eric, And you've

561
00:43:04.599 --> 00:43:08.719
touched on a very important area where
my colleague Professor Ray Winslow, who I

562
00:43:08.800 --> 00:43:15.519
mentioned working also with Professor Gene Tunnik, who runs our AI plus health area

563
00:43:15.559 --> 00:43:20.079
for the institute. They like to
call it, how do I make AI

564
00:43:20.400 --> 00:43:23.800
a valued member of the medical team. That word you use, the recommendation

565
00:43:24.000 --> 00:43:29.159
is key, right, AI doesn't
know everything. It may sometimes it's right,

566
00:43:29.199 --> 00:43:32.119
sometimes it's wrong. The feedback from
the physicians or the care team and

567
00:43:32.239 --> 00:43:38.280
nurse could be anybody, would be
very beneficial to both improve the AI and

568
00:43:38.360 --> 00:43:43.960
to allow that AI to contribute to
the proper care of that patient. So

569
00:43:44.320 --> 00:43:46.239
I love that analogy, which is, you know, AI needs to be

570
00:43:46.400 --> 00:43:52.119
just another member of the healthcare team, making its own recommendations and learning from

571
00:43:52.159 --> 00:43:58.360
corrections, interventions or getting it right
and wrong. That's right in learning.

572
00:43:58.360 --> 00:44:00.880
And here's the beautiful thing. You
know, doctor I had mentioned you have

573
00:44:00.920 --> 00:44:06.000
to capture all this information. When
you correct the engine or when you accept

574
00:44:06.000 --> 00:44:08.400
the recommendation, all of that should
be captured as well, because then over

575
00:44:08.480 --> 00:44:14.119
time you can see, well,
doctor Jones never accepts the recommendation of the

576
00:44:14.159 --> 00:44:17.440
AI engine, and doctor Phillips always
accepts the recommendation of the AI engine.

577
00:44:19.119 --> 00:44:22.880
Somewhere in between is probably where we
want to be. But the point is,

578
00:44:22.679 --> 00:44:27.519
as human beings, we are actually
very predictable. And you will know

579
00:44:27.599 --> 00:44:30.880
that if you start looking at your
device that giving you feedback on how often

580
00:44:30.920 --> 00:44:34.920
you are on this app or that
app or all these different things, you

581
00:44:34.960 --> 00:44:38.119
will see a reflection of your behavior
and you'll realize why I am pretty darn

582
00:44:38.159 --> 00:44:42.800
predictable every day at this time I
do X. But hey, hang out

583
00:44:42.800 --> 00:44:45.079
for one second. We'll be right
back after this break. You are listening

584
00:44:45.119 --> 00:44:52.760
to Inside Analysis. Okay, folks, time for the podcast bonus segment here

585
00:44:52.800 --> 00:44:57.559
on Inside Analysis is always so much
fun talking to doctor Fayad. And you

586
00:44:57.639 --> 00:45:00.840
made a point there after in the
break of out the importance of trust and

587
00:45:00.960 --> 00:45:05.280
building trust in the system, for
the system, for the doctors, for

588
00:45:05.360 --> 00:45:07.480
the team. Go ahead with that. Yeah, So, I mean one

589
00:45:07.519 --> 00:45:14.239
of the critical pieces of making the
AI a valued member of the medical team

590
00:45:14.639 --> 00:45:19.280
or the care team is that ability
to kind of make suggestions and get feedback

591
00:45:19.679 --> 00:45:23.400
those suggestions of recommendations. They get
value, you know, they get evaluated

592
00:45:23.480 --> 00:45:28.559
by a physician or a nurse or
whatever. They may get corrected every once

593
00:45:28.599 --> 00:45:30.760
in a while, which gives them
a chance to contribute to the system and

594
00:45:30.800 --> 00:45:34.880
it's evolution and they feel that their
baby. But at the same time they

595
00:45:34.920 --> 00:45:37.960
get kind of this incremental exposure to
hey, this thing is right. It

596
00:45:38.079 --> 00:45:43.480
caught something I missed. So it's
useful and by the way, applies equally

597
00:45:43.519 --> 00:45:46.280
well whether you're dealing with a medical
team or a factory, a team of

598
00:45:46.320 --> 00:45:52.199
engineers in a factory, a team
of operators in a monitoring center. Every

599
00:45:52.480 --> 00:45:59.280
application where the AI can contribute,
you need to gain the trust of the

600
00:45:59.400 --> 00:46:02.239
users and domain experts, because that's
the only chance you're going to have of

601
00:46:02.280 --> 00:46:07.039
getting acceptance, getting important data,
getting feedback, and keeping the system healthy.

602
00:46:07.679 --> 00:46:10.920
Yeah, that's right. And you
know what I've noticed about the AI

603
00:46:10.960 --> 00:46:15.280
that I've seen in practice, which
I see in my email or my text

604
00:46:15.320 --> 00:46:17.320
messages in other places, it's like, oh, you haven't followed up on

605
00:46:17.400 --> 00:46:20.559
this text, you know, do
you think there was three days ago?

606
00:46:20.639 --> 00:46:25.199
Basically, that's a perfect example of
a reminder that is helpful because, let's

607
00:46:25.199 --> 00:46:29.519
face it, even the best doctor
in the world will forget sometimes to do

608
00:46:29.639 --> 00:46:32.639
one little process, one little piece, you know, and depending upon the

609
00:46:32.760 --> 00:46:36.599
use case, that could be a
very serious thing or not so serious thing.

610
00:46:37.000 --> 00:46:38.960
But you know, you forgot to
wash your hands, for example,

611
00:46:39.280 --> 00:46:43.360
you forgot to put the mask on, you forgot to put the scalpel back

612
00:46:43.400 --> 00:46:46.480
where it belonged, these little teeny
tiny things. Because guess what, People

613
00:46:46.639 --> 00:46:52.039
get distracted, even very focused people
can get distracted, and so the AI

614
00:46:52.159 --> 00:46:55.639
is there as your little helper watching
you, making little suggestions along the way,

615
00:46:57.079 --> 00:47:00.000
not wrapping you on the knuckles,
getting you in trouble with your boss

616
00:47:00.079 --> 00:47:02.159
necessarily, but just giving you a
little bit of a hint and a nudge.

617
00:47:02.559 --> 00:47:06.800
And it reminds me one of my
favorite expressions. That's an old cliche,

618
00:47:06.920 --> 00:47:09.360
don't here much anymore. And of
course it's it's posed on man,

619
00:47:09.400 --> 00:47:13.000
but it means men and women.
It said, man doesn't need to be

620
00:47:13.079 --> 00:47:16.280
taught so much as reminded. Right, we already learned. We just have

621
00:47:16.360 --> 00:47:20.599
to be reminded. Final thoughts from
doctor Fayette go ahead, Yeah, and

622
00:47:21.239 --> 00:47:25.119
absolutely, and you actually need both. You know. Learning in the system

623
00:47:25.199 --> 00:47:29.840
is also critical because you know no
system is going to get it right.

624
00:47:29.880 --> 00:47:31.400
So every once in a while it's
going to guess the wrong thing and it

625
00:47:31.440 --> 00:47:35.920
becomes annoying. It's reminding me of
something I already know. Right, well,

626
00:47:36.039 --> 00:47:37.320
I need to give it that feedback, hey, in this context,

627
00:47:37.360 --> 00:47:42.159
don't remind me. And then if
it adapts to that. Now you have

628
00:47:42.239 --> 00:47:45.000
intelligence. Now you have something you
value and it's personalized. So now you

629
00:47:45.039 --> 00:47:49.000
want to keep it because now it's
beginning to get to know you and your

630
00:47:49.039 --> 00:47:53.199
preferences versus me and my preferences which
could be completely different. So bringing machine

631
00:47:53.280 --> 00:47:58.880
learning into the loop and making sure
these systems learn and adapt to the feedback

632
00:47:58.880 --> 00:48:01.920
they're getting is very Yeah, this
is wonderful stuff. Well look this,

633
00:48:01.960 --> 00:48:07.679
gentleman up online and look up Northeastern
University, the Institute for Experiential AI.

634
00:48:07.159 --> 00:48:12.280
That just means humans in the loop. Doctor Fidgen here and you folks have

635
00:48:12.360 --> 00:48:17.639
put together a great program, and
you understand the interplay between training and research

636
00:48:17.760 --> 00:48:22.360
and development and solution designed. All
these things feed each other, and we'll

637
00:48:22.360 --> 00:48:27.000
feed each other over time. So
you know how to train people, know

638
00:48:27.119 --> 00:48:30.840
how to get people excited about stuff
and really nudge them and get them involved.

639
00:48:30.880 --> 00:48:34.239
Because the humans in the loop,
they're the ones who are going to

640
00:48:34.280 --> 00:48:38.679
make all this stuff successful and worthwhile. And it's going to save lives in

641
00:48:38.719 --> 00:48:42.719
the healthcare space. I promise you
it's going to save lives. It's going

642
00:48:42.800 --> 00:48:45.199
to save time and effort. It's
going to do the most important thing I

643
00:48:45.239 --> 00:48:49.840
think for any business that has improved
morale. When MOREL goes up, good

644
00:48:49.840 --> 00:48:52.119
things happen. When MOREL is down, good things do not happen. So

645
00:48:52.480 --> 00:48:55.559
just that's the Dallas Cowboys. Thank
you, Thank you, Eric, and

646
00:48:55.599 --> 00:49:00.400
thank you for getting our mission and
vision for the institute. You got it

647
00:49:00.440 --> 00:49:02.119
perfectly. I love it. Well, we'll talk to these folks again.

648
00:49:02.159 --> 00:49:05.880
I'm sure folks don't touch that boy. Actually, well, we'll touch you

649
00:49:05.920 --> 00:49:14.159
next time. Infodinsid Analysis dot Com. Bye bye Marino, Valley, Corona

650
00:49:14.199 --> 00:49:22.039
and Riverside. Listen to KCAA,
KCAA the station that leaves no listener behind

651
00:49:23.159 --> 00:49:29.239
and now the voices of KCAA was
an exciting announcement. Want to hear NBC

652
00:49:29.360 --> 00:49:32.159
and News or KCAA anywhere you go? Well, now there's an app for

653
00:49:32.199 --> 00:49:37.480
that. KCAA. You is celebrating
twenty five years in our silver anniversary with

654
00:49:37.559 --> 00:49:42.280
a brand new app. The new
KCAA App is now available on your smart

655
00:49:42.280 --> 00:49:45.960
device, cell phone, in your
car, or any place. Just search

656
00:49:45.039 --> 00:49:50.199
KCAA on Google Play or in the
Apple Store. One touch and you can

657
00:49:50.239 --> 00:49:53.480
listen on your car radio, Bluetooth
device, Android Auto or Applecar Play.

658
00:49:53.719 --> 00:49:59.159
Catch the KCAA buzz in your earbuds
or on the streets. Celebrating twenty five

659
00:49:59.280 --> 00:50:02.679
years of talk news and excellence with
our new kc AAP. Just do it

660
00:50:02.800 --> 00:50:12.320
and download it. K c A
celebrating twenty five years t Hebot Club's original

661
00:50:12.400 --> 00:50:15.199
pure Power Drgo super TA helps build
red corpusols in the blood, which carry

662
00:50:15.199 --> 00:50:21.199
oxygen to organs and cells. Our
organs and cells need oxygen to regenerate themselves.

663
00:50:21.239 --> 00:50:23.920
The immune system needs oxygen to develop, and cancer dies in oxygen.

664
00:50:24.079 --> 00:50:28.440
So the T is great for healthy
people because it helps build the immune system,

665
00:50:28.559 --> 00:50:31.719
and it can truly be miraculous for
someone fting a potentially life threatening disease

666
00:50:31.840 --> 00:50:36.480
due to an infection, diabetes,
or cancer. The T is also organic

667
00:50:36.519 --> 00:50:39.800
and naturally caffeine free. A one
pound package of TA is forty nine ninety

668
00:50:39.840 --> 00:50:44.639
five, which includes shipping. To
order, please visit to Hebot club dot

669
00:50:44.679 --> 00:50:49.119
com. T hebow is spelled T
like Tom, a h ee b like

670
00:50:49.199 --> 00:50:52.079
Boyo. They continue with the word
T and then the word club. The

671
00:50:52.159 --> 00:50:57.639
complete website is to Hebot Club dot
com or call us at eight one eight

672
00:50:57.800 --> 00:51:01.840
six one zero eight zero eight eight
Monday through Saturday nine am to five pm

673
00:51:01.920 --> 00:51:07.599
California time. That's eight one eight
six one zero eight zero eight eight to

674
00:51:07.760 --> 00:51:12.880
Hebot club dot com. Hey,
the Big Games this Sunday. Who you're

675
00:51:12.920 --> 00:51:15.639
rooting for? Joe the Browns?
Well keep waiting, No really, Joe,

676
00:51:15.840 --> 00:51:20.360
what you doing for the big Game? Well? Who has an eighty

677
00:51:20.400 --> 00:51:22.320
five inch TV and you type it
to watch it? Oh? I know

678
00:51:22.440 --> 00:51:27.159
that place. It's the place that
you type on and you type of boulevard

679
00:51:27.239 --> 00:51:30.440
across from the golf course. The
Fat Greek. They have an eighty five

680
00:51:30.480 --> 00:51:32.880
inch TV to watch a big game
on. Hey Joe, why does my

681
00:51:34.000 --> 00:51:37.400
dog not like the Big Game?
Probably likes the Puppy Bowl. Oh,

682
00:51:37.599 --> 00:51:42.320
the puppy Bowl because he's a boxer. Soo's your underwear? Ah? The

683
00:51:42.360 --> 00:51:45.360
Fat Greek has the big Game.
Yeah, Fat Greek has the big Game.

684
00:51:45.440 --> 00:51:47.840
The Fat Greek has the Big Game, big stream and big special.

685
00:51:49.039 --> 00:51:51.960
So Mark, what is Sir mix
a Lot's favorite part of the soup?

686
00:51:52.159 --> 00:51:55.000
I mean the big Game? Oh? No, what what? He likes

687
00:51:55.039 --> 00:51:59.599
big punks that he can out lot? He likes big punks and he cannot

688
00:51:59.679 --> 00:52:02.840
like. Oh okay, the Big
Game starts at three thirty pm. But

689
00:52:02.880 --> 00:52:07.480
the Fat Greek has their free game
going, so get there early. Hey

690
00:52:07.559 --> 00:52:10.159
Joe, what are the Kansas City
Chiefs and a Chick fil a manager having

691
00:52:10.199 --> 00:52:15.280
common? What neither one shows up
for work on Sunday? Gotcha that Jogins

692
00:52:15.320 --> 00:52:20.800
is bat team but serious the people. The food is super at the Fat

693
00:52:20.840 --> 00:52:23.599
Greek thirty three to sixty five you
Paga Boulevard or you can go to go

694
00:52:23.800 --> 00:52:28.760
Fat Greek dot com for more information. Nick Chris and their whole family say

695
00:52:28.840 --> 00:52:31.360
they'll be waiting at the Big Game
party at the Fat Greek. It's the

696
00:52:31.400 --> 00:52:36.400
big game. Bring your big appetite
and have big fun. It'll be a

697
00:52:36.559 --> 00:52:39.960
big easy to plan for your game
celebration at the Fat Greek in Ukaipa,

698
00:52:40.039 --> 00:52:47.480
Go Fat Greek. Oh bah,
it's that time of year again. No,

699
00:52:47.719 --> 00:52:52.400
not the holidays. Medicare open enrollment
and if you have questions about Medicare,

700
00:52:52.559 --> 00:52:57.360
you should talk to the local experts. Paul Barrett and associates. All

701
00:52:57.400 --> 00:53:00.239
of his agents are certified with plans
that are accepted by most of the medical

702
00:53:00.239 --> 00:53:05.719
groups in our area. Call nine
oh nine seven nine three oh three eight

703
00:53:05.880 --> 00:53:08.920
five. Their services free and after
forty two years of the business, their

704
00:53:08.920 --> 00:53:14.159
agents are trained to help you pick
the plan that's right for you. Last

705
00:53:14.280 --> 00:53:20.280
July, several GOP senators combine their
fi WoT intellects to charge that inflation was

706
00:53:20.360 --> 00:53:24.960
rising because of quote insane tax and
spending spree of President Biden and the Democrats,

707
00:53:25.519 --> 00:53:30.639
never mind that the insane spending is
for such sensible, productive, and

708
00:53:30.800 --> 00:53:37.880
enormously popular national needs as childcare and
childless benefits. Mitch McConnell's rapidly partisan flock

709
00:53:37.079 --> 00:53:43.480
saw the chance to politicize the public's
legitimate worries about rising prices. You,

710
00:53:43.599 --> 00:53:47.679
poor consumers are made to pay more
for basics they squawked because of socialist Joe's

711
00:53:47.840 --> 00:53:54.519
investment, and grassroots people follow the
ricocheting pinball of the GOP's logic. One

712
00:53:54.760 --> 00:53:59.920
they say that helping hard hit families
induces them to refuse to go to work.

713
00:54:00.400 --> 00:54:05.760
Two, this creates blockages in the
global supply chain. Three this causes

714
00:54:05.800 --> 00:54:10.760
shortages of everything. Four this forces
corporate bosses to raise all prices, which

715
00:54:10.920 --> 00:54:16.480
five slams the middle class and poor. So six lazy workers cause inflation.

716
00:54:17.119 --> 00:54:22.920
Whooh, Rube Goldberg couldn't have dreamed
up a more fantastical diagram to deflect attention

717
00:54:23.079 --> 00:54:28.840
from what's really happening, namely that
instead of an inflation problem, we have

718
00:54:28.920 --> 00:54:32.880
a corporate greed problem. Of course, the greed meisters insist that their pursuit

719
00:54:32.880 --> 00:54:38.039
of excess corporate profits has not driven
any price surges in our economy of free

720
00:54:38.039 --> 00:54:43.519
market competition, they snap, Prices
are established by the law of supply and

721
00:54:43.639 --> 00:54:47.440
demand. It's the magic of the
market place, they explain. But magicians

722
00:54:47.480 --> 00:54:52.360
don't do magic, They perform illusions, and the illusion of free market competition

723
00:54:52.519 --> 00:54:59.480
implodes when it hits the reality that
our economy doesn't remotely resemble a competitive market

724
00:54:59.519 --> 00:55:02.800
place. This is Jim Heiitar saying, for some forty years, corporate directed

725
00:55:02.840 --> 00:55:08.760
government policies have been transforming America into
monopoly nation, letting the few gouge the

726
00:55:08.800 --> 00:55:15.559
manuye. That's where inflation comes from. Lauren sp century twenty one. Masters

727
00:55:15.599 --> 00:55:20.880
reminds listeners that a family pet is
a year round commitment. Proper care includes

728
00:55:20.960 --> 00:55:23.599
regular checkups and vaccinations, dental care, space, and neuters, along with

729
00:55:23.679 --> 00:55:28.559
flea, tick and heartworm preventions.
This reminder is brought to you by laurenesb.

730
00:55:28.760 --> 00:55:32.400
Century twenty one Masters, reminding everybody
pets are people too. Lauren from

731
00:55:32.480 --> 00:55:37.559
century twenty one Masters goes to extra
mile to keep all her clients, having

732
00:55:37.559 --> 00:55:40.000
to put her fifteen years of experience
to work for you and call her today

733
00:55:40.039 --> 00:55:44.639
at nine by one, six two
three, six y nine five five.

734
00:55:44.800 --> 00:55:47.400
That's nine by one six, two, three, sixty nine fifty five.

735
00:55:47.519 --> 00:55:57.519
That's Lauren esb century twenty one Masters
on the Air Because they Care. Redlands

736
00:55:57.599 --> 00:56:01.480
Ranch Market is a unique, full
service international grocery store that specializes in authentic

737
00:56:01.559 --> 00:56:07.119
food items from Mexico, India,
and from many Mediterranean and Asian countries,

738
00:56:07.239 --> 00:56:10.320
including popular items from the US.
They offer fresh baked items from their in

739
00:56:10.440 --> 00:56:15.840
house bakery, housemade tortillas from their
Tortillirea, a delicious array of prepared Mexican

740
00:56:15.880 --> 00:56:20.760
foods, a terrific fresh food and
juice bar, and a large selection of

741
00:56:20.800 --> 00:56:23.920
meats, seafoods and deli sandwiches,
salads, and hellal meats. Their produce

742
00:56:24.000 --> 00:56:29.079
department is stock full with fresh,
local and hard to find international fruits and

743
00:56:29.199 --> 00:56:32.800
vegetables that you cannot find anywhere else. Don't forget to step into the massive

744
00:56:32.920 --> 00:56:38.440
beer cave and experience the largest selection
of domestic, artisan and imported beers in

745
00:56:38.480 --> 00:56:43.199
the ie. They can also cater
your next event with one of the delicious

746
00:56:43.239 --> 00:56:46.719
takeout catering trades of food. Visit
them at Redlands ranch Market dot com.

747
00:56:46.760 --> 00:56:52.880
That's Redlands ranch Market dot com.
Redlands Ranch Market a unique and fun shopping

748
00:56:52.920 --> 00:56:59.280
destination. KCAA Radio has openings for
one hour talk shows. If you want

749
00:56:59.280 --> 00:57:02.440
to host a radio show, now
is the time. Make KCA your flangship

750
00:57:02.440 --> 00:57:07.239
station. Our rates are affordable and
our services are second to none. We

751
00:57:07.360 --> 00:57:10.920
broadcast to a population of five million
people plus. We stream and podcast on

752
00:57:12.039 --> 00:57:15.639
all major online audio and video systems. If you've been thinking about broadcasting a

753
00:57:15.679 --> 00:57:22.119
weekly radio program on real radio plus
the Internet, contact our CEO at two

754
00:57:22.119 --> 00:57:25.880
eight one five nine nine ninety eight
hundred. Two eight one five nine nine

755
00:57:27.000 --> 00:57:30.360
ninety eight hundred. You can skype
your show from your home to our Redlands,

756
00:57:30.400 --> 00:57:34.599
California studio, where our live producers
and engineers are ready to work with

757
00:57:34.639 --> 00:57:38.880
you personally. A radio program on
KCIA is the perfect work from home advocation

758
00:57:39.000 --> 00:57:44.760
in these stressful times. Just type
KCA radio dot com into your browser to

759
00:57:44.840 --> 00:57:47.320
learn more about hosting a show on
the best station in the nation, or

760
00:57:47.400 --> 00:57:52.239
call our CEO for details too.
Eight one five nine nine ninety eight hundred

761
00:57:54.840 --> 00:57:59.599
NBC News Radio, I'm Chris Graggio. Two people are injured and the shooter

762
00:57:59.679 --> 00:58:02.960
is dead following a shooting at Jolo
Stein's Lakewood Church in Houston. The woman

763
00:58:04.119 --> 00:58:07.239
entered the church this afternoon, wearing
a trench coat and backpack, armed with

764
00:58:07.280 --> 00:58:10.519
what authorities called a long rifle,
and then began shooting. Two off duty

765
00:58:10.559 --> 00:58:15.119
officers engaged the suspects, striking her. She reportedly entered with a five year

766
00:58:15.119 --> 00:58:19.800
old child who was hit during the
incident and is in critical condition. Another

767
00:58:19.840 --> 00:58:22.119
man was hitting the leg and is
receiving treatment at a local hospital. It's

768
00:58:22.159 --> 00:58:25.880
currently unknown what the motive was or
how the child was related to the shooter.

769
00:58:27.559 --> 00:58:30.559
Memphis police say a suspect is in
custody after multiple shootings across the city

770
00:58:30.679 --> 00:58:35.519
that left one person dead and multiple
people injured. Authorities say the shootings took

771
00:58:35.519 --> 00:58:38.320
place across three different locations in the
city, with a carjacking being connected to

772
00:58:38.400 --> 00:58:43.360
all three crime scenes. The suspect
was caught, with authorities saying he was

773
00:58:43.400 --> 00:58:46.480
out on one hundred thousand dollars bond
for attempt at first degree murder and attempted

774
00:58:46.559 --> 00:58:52.639
aggravated robbery when he committed the crimes. At least four people were reportedly shot.

775
00:58:52.920 --> 00:58:57.320
A former GOP presidential candidate says if
the twenty twenty four election comes down

776
00:58:57.519 --> 00:59:00.760
to President Biden versus former President Donald
trum Up, then the American people are

777
00:59:00.800 --> 00:59:07.960
left with two really, really bad
choices. Both of these candidates are people

778
00:59:07.000 --> 00:59:13.920
who are being questioned for their competence
and being questioned for their character, and

779
00:59:13.960 --> 00:59:16.519
that's a problem for the American people. Speaking on NBC's Meet the Press,

780
00:59:16.599 --> 00:59:21.599
Chris Christie said he knew the GOP
race was over in the first debate when

781
00:59:21.639 --> 00:59:24.440
six of the eight candidates on the
stage said they would support Trump even if

782
00:59:24.440 --> 00:59:29.760
he's a convicted felon. The former
New Jersey governor and US attorney talked about

783
00:59:29.760 --> 00:59:34.559
the Justice Department report that said there
will not be any charges against Biden related

784
00:59:34.559 --> 00:59:38.960
to classified documents, saying comments about
the president's memory were appropriate because they explained

785
00:59:38.960 --> 00:59:44.079
why he wasn't being charged. When
asked, Christy said he has not looked

786
00:59:44.079 --> 00:59:47.119
into running as a third party candidate
and hasn't had any talks with the No

787
00:59:47.280 --> 00:59:54.599
Label's movement. I'm Chris Garagio,
NBC News Radio, NBC News on CACAA

788
00:59:54.719 --> 01:00:00.719
Lomalinde sponsored by Teamsters Local nineteen thirty
two. Protecting the Future Share of Working

789
01:00:00.760 --> 01:00:02.519
Families Teamsters nineteen thirty two, dot
org