WEBVTT 1 00:00:05.160 --> 00:00:07.599 This week's episode of the trip Cast is sponsored by 2 00:00:07.599 --> 00:00:19.239 the Texas Tribune Membership program as we celebrate Amplify Austin. Hello, 3 00:00:19.280 --> 00:00:22.120 and welcome to the Texas Tribune Trip Cast for March fourth, 4 00:00:22.120 --> 00:00:25.320 twenty twenty five. I'm Matthew Watkins, editor in Chief of 5 00:00:25.359 --> 00:00:29.199 the Tribune. My two co hosts, Eleanor Klibanoff and James 6 00:00:29.199 --> 00:00:33.840 Berragon are nursing their oscar parties hangover, aren't able to 7 00:00:33.880 --> 00:00:36.880 make it today, so instead I am joined by Cherry Greenberg, 8 00:00:36.920 --> 00:00:39.520 a former state representative and professor of practice at the 9 00:00:39.640 --> 00:00:42.399 LBJ School of Government at the UT Austin. 10 00:00:42.799 --> 00:00:43.920 Hello, Sherry, thanks for being. 11 00:00:43.840 --> 00:00:45.159 Here, my pleasure. Hello. 12 00:00:45.479 --> 00:00:48.159 So today we are going to be talking about a 13 00:00:48.240 --> 00:00:51.280 subject that you have been studying for several years now, 14 00:00:51.280 --> 00:00:54.200 many years even before it became sort of the buzzword 15 00:00:54.280 --> 00:00:57.719 in the Texas capital and beyond the idea of artificial 16 00:00:57.759 --> 00:01:01.359 intelligence and regulating it at this date level, When the 17 00:01:01.399 --> 00:01:04.400 idea of regulating businesses comes up in the legislature, there's 18 00:01:04.519 --> 00:01:08.439 usually a pretty strong consensus among Republicans to sort of 19 00:01:08.599 --> 00:01:10.480 stay out of the way in the name of growth 20 00:01:10.519 --> 00:01:14.040 and free enterprise. But the conversation this year around regulating 21 00:01:14.079 --> 00:01:16.840 AI is turning out to be a little bit more complex. 22 00:01:17.040 --> 00:01:20.799 You have powerful conservatives such as Senate Business and Commerce 23 00:01:20.879 --> 00:01:24.959 Chairman Charles Schwartner and the Chair of the House Doge Committee, 24 00:01:25.319 --> 00:01:30.920 Giovanni Copriglioni, comparing carrying bills that would impose strong guardrails 25 00:01:30.959 --> 00:01:34.079 and rules for the technology, but they're finding some opposition 26 00:01:34.120 --> 00:01:37.079 from members of their own party. Senator Lewis Cole course 27 00:01:37.200 --> 00:01:41.079 was questioning a bill by Schwartner earlier last week. Tech 28 00:01:41.959 --> 00:01:45.200 impresario Joe Lonsdale has really come out against some of 29 00:01:45.239 --> 00:01:49.439 the legislation, despite being pretty usually aligned with Republicans in 30 00:01:49.480 --> 00:01:51.760 the state. It's a fight that sort of avoids that 31 00:01:51.840 --> 00:01:55.439 partisan brack breakdown. But I think there's some fear kind 32 00:01:55.439 --> 00:01:58.239 of driving both sides. On one hand, you have lawmakers 33 00:01:58.280 --> 00:02:01.560 worried about the dangers of AI on people's privacy, their freedom, 34 00:02:01.840 --> 00:02:04.680 and the possibility that really important decisions might be made 35 00:02:04.719 --> 00:02:07.239 by computers not people. On the other side, you have 36 00:02:07.280 --> 00:02:10.120 folks who are worried about going too far and restricting 37 00:02:10.199 --> 00:02:13.240 AI that would leave Texas behind other states in the 38 00:02:13.280 --> 00:02:17.199 technology and maybe even pulling the United States with them 39 00:02:17.520 --> 00:02:22.199 behind adversaries like China. So Sherry, You, like I said, 40 00:02:22.199 --> 00:02:24.199 have been watching this for a very long time. A 41 00:02:24.240 --> 00:02:26.759 lot of the focus so far earlier on this session 42 00:02:27.159 --> 00:02:30.960 is on a major bill called TREGA, the Texas Responsible 43 00:02:31.039 --> 00:02:34.479 AI Governance Act. Can we start by just having you 44 00:02:34.599 --> 00:02:36.400 explain what's in that bill? 45 00:02:36.840 --> 00:02:40.439 Sure? And for background, which you really alluded to, is 46 00:02:40.439 --> 00:02:43.280 what I call the promise in peril of AI. Right, 47 00:02:43.520 --> 00:02:47.800 there's great promise with AI with artificial intelligence and healthcare 48 00:02:47.919 --> 00:02:51.280 detecting diseases that we wouldn't be able to, but there's 49 00:02:51.360 --> 00:02:53.639 peril too. So if we can use healthcare and as 50 00:02:53.639 --> 00:02:58.719 an example, big privacy issues, and we also talk about 51 00:02:59.000 --> 00:03:08.000 balancing innovation with rules and policies and regulations that provide transparency, 52 00:03:08.199 --> 00:03:12.680 accountability and guard against invasion and privacy. And so that's 53 00:03:12.719 --> 00:03:16.639 really what we're dealing with here. And with artificial intelligence, 54 00:03:17.159 --> 00:03:20.199 we do find that there are a lot of common 55 00:03:20.319 --> 00:03:25.879 concerns across political parties and sometimes you find alliances that 56 00:03:25.919 --> 00:03:28.319 you wouldn't elsewhere, and you also alluded to that. So 57 00:03:28.400 --> 00:03:31.360 with TREGA, this is a bill that was filed its 58 00:03:31.400 --> 00:03:36.639 House Bill seventeen oh nine by Representative Giovanni Capriconi, and 59 00:03:36.840 --> 00:03:39.759 there was a lot of discussion about this bill over 60 00:03:39.800 --> 00:03:43.439 the interim. He had a very large stakeholder group that 61 00:03:43.879 --> 00:03:48.800 many of us participated in meetings and whatnot, and I 62 00:03:48.840 --> 00:03:51.319 think he knew that this is a bill that would 63 00:03:51.960 --> 00:03:55.719 have a lot of discussion once it was filed. So 64 00:03:56.120 --> 00:04:02.759 TREGA is really a builds based on risk mitigation, and 65 00:04:02.879 --> 00:04:05.800 this is what the EU bill really is based on. 66 00:04:05.960 --> 00:04:09.879 And also other states such as Colorado which we can 67 00:04:09.919 --> 00:04:14.639 look to, and it's mitigating against risk, which is also 68 00:04:14.879 --> 00:04:21.480 what this the Nationaltitude for Standards has really set out 69 00:04:21.519 --> 00:04:24.839 in its framework, which is part of the National Science Foundation. 70 00:04:25.000 --> 00:04:30.079 So TREGA would create obligations for developers and deployers of 71 00:04:30.240 --> 00:04:34.000 any AI system that is a contribution, you know, decisions 72 00:04:34.040 --> 00:04:39.000 relating to employment, finance, healthcare. Texas would also be the 73 00:04:39.040 --> 00:04:43.079 first state to require high risk AI system distributors to 74 00:04:43.160 --> 00:04:47.360 take reasonable care. The bill would ban AI systems that 75 00:04:47.439 --> 00:04:53.600 pose unacceptable risk. Private litigans would have limited right of action. 76 00:04:54.199 --> 00:04:57.600 The law would require generative AI to developers to keep 77 00:04:57.639 --> 00:05:01.720 detailed records of data sets used to It would exempt 78 00:05:01.720 --> 00:05:04.439 many small businesses. So you can see it's very much 79 00:05:04.519 --> 00:05:10.519 a risk based system. It is also a build that 80 00:05:10.639 --> 00:05:14.560 very much focuses on what we call algorithmic discrimination, and 81 00:05:14.600 --> 00:05:17.800 in the bill, that means any condition in which an 82 00:05:17.920 --> 00:05:23.680 artificial intelligence system, when deployed, creates an on lawful discrimination 83 00:05:23.839 --> 00:05:27.720 of a protected classification in violation of the laws of 84 00:05:27.759 --> 00:05:30.279 this state or federal law. And then it further goes 85 00:05:30.360 --> 00:05:34.240 on because you're wondering, what's an artificial intelligence system? Right? 86 00:05:34.319 --> 00:05:39.279 That means the use of machine learning and related technologies 87 00:05:39.560 --> 00:05:43.680 that use data to train statistical models for the purpose 88 00:05:43.759 --> 00:05:48.839 of enabling computer systems to perform tasks normally associated with 89 00:05:48.920 --> 00:05:51.240 human intelligence. And this is really where we get into 90 00:05:51.279 --> 00:05:56.040 the AI, the artificial intelligence realm or perception such as 91 00:05:56.079 --> 00:06:01.160 computer vision, speech, or natural language processing and content generation. 92 00:06:01.800 --> 00:06:04.399 Okay, so there's a few things that are sort of 93 00:06:05.120 --> 00:06:08.079 key phrases in this bill, one of which you already 94 00:06:08.079 --> 00:06:11.040 mentioned high risk AI systems. What do we mean by 95 00:06:11.040 --> 00:06:13.160 a high risk system in this case. 96 00:06:13.120 --> 00:06:17.040 So a high risk system in this bill, and again 97 00:06:17.079 --> 00:06:21.639 this bill is really based on risk mitigation. High risk 98 00:06:21.759 --> 00:06:26.040 artificial intelligence system means an artificial intelligence system that is 99 00:06:26.560 --> 00:06:31.399 a substantial factor in a consequential decision. The term, though, 100 00:06:31.480 --> 00:06:34.959 does not include an artificial system. If the artificial intelligence 101 00:06:34.959 --> 00:06:38.879 system is intended to detect decision making patterns or deviation 102 00:06:39.439 --> 00:06:43.279 from prior decision making patterns, and is not intended to 103 00:06:43.279 --> 00:06:48.360 replace or influence a previously completed human assessment without sufficient 104 00:06:48.439 --> 00:06:54.319 human review. So the following technologies unless the technologies, when deployed, 105 00:06:54.399 --> 00:06:58.600 make or a substantial factor in making a consequential decision. 106 00:06:58.639 --> 00:07:04.680 And a malware anti virus calculation, cybersecurity database is data storage, firewall, 107 00:07:05.639 --> 00:07:08.079 internal domain. The list goes on and on. 108 00:07:08.279 --> 00:07:10.800 So those are things that are not included in the building. 109 00:07:10.920 --> 00:07:14.680 But what we're really talking about here are basically using 110 00:07:14.800 --> 00:07:19.000 AI using computers to make kind of important, impactful decisions 111 00:07:19.000 --> 00:07:21.480 about customers or people in this state. 112 00:07:21.639 --> 00:07:23.240 That's exactly right, okay, And. 113 00:07:23.560 --> 00:07:25.519 I think the other interesting thing is the idea of 114 00:07:25.600 --> 00:07:32.240 using reasonable care. Right, this this idea that if essentially 115 00:07:32.360 --> 00:07:35.399 it's up to the developers, the people you know, like 116 00:07:35.480 --> 00:07:39.000 open ai or or you know, Twitters developing some some 117 00:07:39.120 --> 00:07:42.560 AI systems and things like that, or even the distributors 118 00:07:42.639 --> 00:07:45.600 or the employer, the deployers, the people who are you know, 119 00:07:46.279 --> 00:07:51.040 utilizing an AI system to evaluate, you know, whatever it 120 00:07:51.079 --> 00:07:54.240 is they might do, make decisions about healthcare, health insurance 121 00:07:54.240 --> 00:07:56.560 and things like that. What it's basically saying is that 122 00:07:56.600 --> 00:08:00.800 they need to take steps that are done reasonably to 123 00:08:00.879 --> 00:08:03.720 protect people from that kind of discrimination in the system. 124 00:08:03.800 --> 00:08:04.240 Is that right? 125 00:08:04.319 --> 00:08:08.439 That is exactly right? And there is a kind of 126 00:08:09.319 --> 00:08:12.279 I guess provision you would say for piloting, you know, 127 00:08:12.360 --> 00:08:17.680 with a sandbox where there can be some innovation taking place. 128 00:08:18.079 --> 00:08:21.480 Okay, okay, very good. And so I mean what's interesting here. 129 00:08:21.519 --> 00:08:23.560 This is the bill that's gotten a lot of attention. 130 00:08:23.800 --> 00:08:27.120 It'll be interesting to see whether it advances in its 131 00:08:27.160 --> 00:08:29.040 current form, and we can talk about that a little 132 00:08:29.079 --> 00:08:31.839 bit more. But what's really struck me about this is 133 00:08:32.039 --> 00:08:34.559 there was a lot of conversation, you know, when when 134 00:08:34.639 --> 00:08:37.559 chat GPT was really out there, a lot of conversation 135 00:08:37.639 --> 00:08:41.000 about the dangers of AI. A lot of those felt 136 00:08:41.000 --> 00:08:44.879 like sort of conversations around existential dangers for humanity. 137 00:08:45.000 --> 00:08:46.840 Right, are the right going to take over? 138 00:08:47.039 --> 00:08:49.720 Yes, sentient and they're not cynient. 139 00:08:49.440 --> 00:08:53.120 Exactly exactly, And that's not really what this is protecting about. 140 00:08:53.200 --> 00:08:57.360 This is more about privacy concerns. Are AI you know, 141 00:08:57.440 --> 00:08:59.799 basically things without a conscience is going to be making 142 00:09:00.080 --> 00:09:03.879 you know, life or death or you know, major financial 143 00:09:03.919 --> 00:09:06.159 decisions for the lives of people in Texas. 144 00:09:06.240 --> 00:09:06.879 Is that is that right? 145 00:09:07.080 --> 00:09:11.440 Yes, this is not about whether AI is, as we say, sentient, right, 146 00:09:11.919 --> 00:09:16.240 It is about looking at those areas that people are 147 00:09:16.279 --> 00:09:20.360 most concerned about when they talk about AI, protecting their privacy, 148 00:09:20.759 --> 00:09:25.600 having transparency, having accountability, knowing when you know AI is 149 00:09:25.639 --> 00:09:31.000 being used. The bill, you know, requires the as we said, 150 00:09:31.120 --> 00:09:36.080 an unacceptable risk. It prohibits systems that without express consent, 151 00:09:36.279 --> 00:09:42.120 identify emotions, capture biometic intent, you know, those attributes that 152 00:09:42.200 --> 00:09:48.440 were really sensitive about as people in this atmosphere. It 153 00:09:48.519 --> 00:09:53.519 also would require generative AI developers to keep detailed records 154 00:09:53.519 --> 00:09:56.679 of data sets that they used to train. So the bill, 155 00:09:57.039 --> 00:10:02.639 again going back to the unist frame work, really looks 156 00:10:02.720 --> 00:10:08.679 at risk and risk management and some of those perils 157 00:10:08.679 --> 00:10:14.759 that we talk about. The risk management framework looks at governmap, 158 00:10:15.000 --> 00:10:20.000 measure and manage and that is a framework that has 159 00:10:20.039 --> 00:10:23.919 been used by NIS, really by the EU, by Colorado 160 00:10:24.399 --> 00:10:27.440 and other states. Now that's not to say that people 161 00:10:27.879 --> 00:10:32.519 everyone is happy about it, right, and really with AI, 162 00:10:32.840 --> 00:10:36.399 we can go back to nineteen fifty with Alan Turing r. 163 00:10:37.000 --> 00:10:39.759 AI has been with us a long time and in 164 00:10:39.799 --> 00:10:44.000 our daily lives. But until chat GPT appeared on the scene, 165 00:10:44.039 --> 00:10:48.080 people weren't really cognizant about it. Once people became cognizant, 166 00:10:48.080 --> 00:10:51.480 then it's oh, what does this mean for my privacy? 167 00:10:51.639 --> 00:10:55.679 My data? Is there transparency? Do I know when AI 168 00:10:55.799 --> 00:11:00.000 is being used? However, there is this, as I said, 169 00:11:00.120 --> 00:11:04.679 promise in peril or this also this issue of how 170 00:11:04.679 --> 00:11:08.480 do we have innovation right without too much risk? As 171 00:11:08.519 --> 00:11:13.759 you mentioned, people are concerned if this stifles innovation, where 172 00:11:13.799 --> 00:11:17.039 will that mean for us as a state? We know 173 00:11:17.279 --> 00:11:19.720 that we're getting ready to have some big investments in 174 00:11:19.759 --> 00:11:22.519 this state right in AI. What would that mean for 175 00:11:22.639 --> 00:11:25.559 the country you mentioned China, And of course we've heard 176 00:11:25.559 --> 00:11:30.159 a lot about China and AI recently. But on the 177 00:11:30.200 --> 00:11:33.200 other hand, people really are concerned about I would say, 178 00:11:33.240 --> 00:11:38.720 their privacy and about this algorithmic discrimination in decision making 179 00:11:39.039 --> 00:11:42.279 and also other areas. There are other bills, of course, 180 00:11:42.320 --> 00:11:46.679 have been filed dealing with using AI for mental health, 181 00:11:47.240 --> 00:11:51.919 or using AI and making decisions with insurance and healthcare. 182 00:11:52.200 --> 00:11:54.200 So there are other concerns that people have. 183 00:11:54.639 --> 00:11:58.720 Yeah, it seems like the would you say, the broad 184 00:11:59.080 --> 00:12:01.519 kind of fee of a lot of these bills is 185 00:12:01.559 --> 00:12:06.440 sort of designed from a kind of consumer protection standpoint. 186 00:12:06.799 --> 00:12:09.559 Yes, I would say it is designed from a consumer 187 00:12:09.600 --> 00:12:15.600 protection and transparency standpoint. The bills, you know, they may say, well, 188 00:12:15.840 --> 00:12:18.600 you can use it for mental health, but there has 189 00:12:18.679 --> 00:12:22.080 to be a provider involved. Keeping the human in the loop. 190 00:12:22.159 --> 00:12:26.399 I would say, you know, when you're looking at health insurance, 191 00:12:26.440 --> 00:12:29.360 you know you have to disclose perhaps that you're using 192 00:12:29.399 --> 00:12:31.879 eye and you have to if it's for specialty care, 193 00:12:32.000 --> 00:12:36.840 specialist for instance, has to review it. So consumer protection, 194 00:12:38.399 --> 00:12:44.159 transparency and accountability and kind of I would say, keeping 195 00:12:44.200 --> 00:12:45.559 that human in the loop. 196 00:12:46.320 --> 00:12:49.360 It's interesting because in some ways this feels like a 197 00:12:49.519 --> 00:12:54.480 very familiar kind of conversation between innovation and industry and 198 00:12:54.960 --> 00:12:58.080 creation and all those kinds of things, and you know, 199 00:12:58.200 --> 00:13:03.240 the consumer protection, the fears about the negative impacts of 200 00:13:03.240 --> 00:13:06.360 this technology, and I just I find it a little 201 00:13:06.399 --> 00:13:09.240 bit striking that we're seeing, you know, at least a 202 00:13:10.200 --> 00:13:14.720 significant segment of the Republican Party sort of going in 203 00:13:14.799 --> 00:13:18.200 the direction of consumer protection this time around. What do 204 00:13:18.240 --> 00:13:22.000 you think it is that's different about AI that's setting 205 00:13:22.080 --> 00:13:22.840 up this conversation. 206 00:13:23.080 --> 00:13:27.440 Well, certainly, I think that there are a lot of 207 00:13:27.679 --> 00:13:33.480 legislators who have followed the path that for instance, social 208 00:13:33.519 --> 00:13:39.159 media took and lack of privacy with data and they're 209 00:13:39.159 --> 00:13:42.159 feeling that this is a big concern of their constituents. 210 00:13:42.240 --> 00:13:45.879 And seeing now with AI, you know, your data, your 211 00:13:45.919 --> 00:13:48.559 personal data is being used to train it. For instance, 212 00:13:49.200 --> 00:13:52.720 your personal data may be available. You may not know 213 00:13:52.919 --> 00:13:55.159 when it's being used. It may you know, is it 214 00:13:55.279 --> 00:13:57.840 making decisions without a human and a loop. And I 215 00:13:57.879 --> 00:14:01.559 think that those are issues that you know trands, they're 216 00:14:01.679 --> 00:14:07.600 they're beyond party boundaries, and that many constituents across the 217 00:14:07.639 --> 00:14:15.120 board are very concerned with these issues and concerned with 218 00:14:15.879 --> 00:14:19.399 the I would say, the path for instance, that data 219 00:14:19.799 --> 00:14:22.720 and social media has taken with you know, lack of 220 00:14:22.759 --> 00:14:25.639 privacy and now seeing AI. 221 00:14:26.399 --> 00:14:29.759 Yeah, I mean among the things in this this trega bill, 222 00:14:30.200 --> 00:14:35.879 it bans deep fakes right, and certain use of biomarkers 223 00:14:36.000 --> 00:14:38.720 like facial recognition that you know, they don't want people 224 00:14:38.720 --> 00:14:41.039 pulling a bunch of pictures from Facebook to be able 225 00:14:41.080 --> 00:14:44.240 to identify your face, you know, wherever you are walking 226 00:14:44.279 --> 00:14:47.879 across the street. Another thing it bans is social scoring, 227 00:14:48.200 --> 00:14:51.399 which I'm quoting from the MIT Technology Review. The definition 228 00:14:51.399 --> 00:14:55.679 of that a practice in which authoritarian governments, specifically China, 229 00:14:56.080 --> 00:15:00.200 rank people's trustworthiness and punish them for undesirable behaviors such 230 00:15:00.200 --> 00:15:03.159 as stealing or not paying back loans. Right essentially seen 231 00:15:03.240 --> 00:15:06.559 as a you know, sort of way to rank citizens 232 00:15:06.639 --> 00:15:10.360 or or identify citizens as types of risk and that 233 00:15:10.519 --> 00:15:13.679 kind of I mean, I think, you know, another big 234 00:15:13.759 --> 00:15:16.279 part of the Republican Party is freedom right. And there 235 00:15:16.279 --> 00:15:19.080 were a little bit about surveillance and just you know, 236 00:15:19.200 --> 00:15:22.320 whether it's big companies or the government itself sort of 237 00:15:22.720 --> 00:15:25.840 you know, reaching into the personal lives in a way 238 00:15:25.879 --> 00:15:27.480 that feels uncomfortable to that's right. 239 00:15:27.519 --> 00:15:32.399 It's that privacy issue. They're very concerned about you not 240 00:15:32.559 --> 00:15:34.480 knowing your data is being used, you not knowing that 241 00:15:34.519 --> 00:15:37.960 your face is being used, to the biometric markers, your 242 00:15:38.080 --> 00:15:40.559 voice being taken. You know, there was a case in 243 00:15:41.399 --> 00:15:46.559 Arizona over a year ago, a really horrible situation, and 244 00:15:46.679 --> 00:15:51.519 there was legislation in the Arizona legislature where there was 245 00:15:51.600 --> 00:15:56.000 a deep fake of a young woman's voice and it 246 00:15:56.039 --> 00:15:58.600 was it was her voice, and they called her mother 247 00:15:59.080 --> 00:16:01.759 and her younger was there too, and said that she 248 00:16:01.840 --> 00:16:05.600 had been kidnapped asking for ransom. Well, in fact, she 249 00:16:05.720 --> 00:16:09.120 had not been kidnapped, and that was not her voice, 250 00:16:09.440 --> 00:16:12.440 but it was a replica, let's say, of her voice 251 00:16:12.480 --> 00:16:16.000 a deep fake. So there are a lot of concerns 252 00:16:16.039 --> 00:16:20.639 I think about criminal elements, about privacy, and as you said, 253 00:16:20.720 --> 00:16:24.759 about freedom, my freedom to walk down the street without 254 00:16:24.960 --> 00:16:30.679 being ranked as a threat to society, or my freedom 255 00:16:30.759 --> 00:16:35.360 to control how my image and likeness and voices being used. 256 00:16:35.720 --> 00:16:36.120 All right. 257 00:16:36.200 --> 00:16:40.039 Today's trip cast is presented by the Texas Tribune Membership program. 258 00:16:40.320 --> 00:16:43.759 The Texas Tribune's in depth coverage of Texas politics relies 259 00:16:43.799 --> 00:16:47.000 on donations from people like you. Right now, your donation 260 00:16:47.120 --> 00:16:49.879 to support this reporting goes twice as far thanks to 261 00:16:49.960 --> 00:16:53.600 a fifteen thousand dollars match for Amplify Austin Day. Support 262 00:16:53.679 --> 00:16:58.440 independent journalism by making a donation online today. Okay, So 263 00:16:58.600 --> 00:17:01.639 I want to ask you a little bit about just 264 00:17:02.399 --> 00:17:05.240 the state's ability to succeed in this process. 265 00:17:05.400 --> 00:17:05.559 Right. 266 00:17:05.839 --> 00:17:09.480 We saw a you know, after this bill came out. 267 00:17:09.559 --> 00:17:11.839 Joe Lonsdale, who I mentioned sort of in the opening, 268 00:17:12.200 --> 00:17:15.680 a venture capitalist, a very influential one both politically and 269 00:17:15.920 --> 00:17:19.359 in the tech world, came out very strongly against this 270 00:17:19.480 --> 00:17:23.359 trega bill. He says he did kill innovation, kill jobs, 271 00:17:23.720 --> 00:17:27.039 let China win the AI race. He essentially argued that 272 00:17:27.079 --> 00:17:30.440 it would create a new AI governing agency in Texas 273 00:17:30.559 --> 00:17:33.960 and see the regulation of AI to you know, bureaucrats, right, 274 00:17:34.000 --> 00:17:37.440 the dread of bureaucrats tell me a little bit about 275 00:17:37.519 --> 00:17:39.200 the industry response. 276 00:17:39.359 --> 00:17:44.000 Sure, sure, and it's not you know, you mentioned his statements, 277 00:17:44.000 --> 00:17:45.799 but he's not the only one. There were articles, I 278 00:17:45.880 --> 00:17:50.279 think in Forbes there was an opinion piece, and I 279 00:17:50.319 --> 00:17:52.319 believe the Dallas Morning News last week. 280 00:17:52.920 --> 00:17:57.960 If you google TREGA, all the top links are industry 281 00:17:58.000 --> 00:18:01.039 aligned groups with articles and other things kind of speaking 282 00:18:01.039 --> 00:18:01.519 out against this. 283 00:18:01.759 --> 00:18:05.200 Right, you know, there was another article, everything is bigger 284 00:18:05.200 --> 00:18:09.640 in Texas. TREGA is no exception. But will it, you know, 285 00:18:09.720 --> 00:18:14.960 provide accountability? So and they're saying no, it wouldn't. And 286 00:18:15.960 --> 00:18:19.759 so any time that you file a significant bill, Having 287 00:18:19.799 --> 00:18:22.680 spent ten years in the legislature myself, I can tell 288 00:18:22.720 --> 00:18:25.640 you that you're going it's a starting point, and you're 289 00:18:25.640 --> 00:18:28.799 going to expect that there will be changes, that there 290 00:18:28.880 --> 00:18:33.559 will be amendments. And with a bill like this, even 291 00:18:33.680 --> 00:18:37.039 though there was a stakeholder group and a lot of input, 292 00:18:37.200 --> 00:18:40.759 I think the bill that's filed, I think it's safe 293 00:18:40.799 --> 00:18:42.799 to say probably is not going to be the final 294 00:18:42.880 --> 00:18:48.960 piece of legislation, and so we take it as a 295 00:18:49.000 --> 00:18:52.400 starting point. And certainly there is this concern about innovation. 296 00:18:53.440 --> 00:18:57.160 You know, we're in another industrial revolution, and with every 297 00:18:57.160 --> 00:18:59.720 industrial revolution, whether you say we're in the fourth or 298 00:18:59.759 --> 00:19:03.519 fifth or whatever, you know, we have seen concerns about 299 00:19:04.559 --> 00:19:07.440 the you know, the perils of the new technology, but 300 00:19:07.480 --> 00:19:12.200 also the promise. And there are concerns, you know, with 301 00:19:12.400 --> 00:19:15.880 AI about work and future work, for instance, will people 302 00:19:15.920 --> 00:19:18.039 be out of jobs or will there be new jobs? 303 00:19:18.039 --> 00:19:19.839 And we can look back at history and see this, 304 00:19:19.960 --> 00:19:23.200 So I think there there is a real issue here 305 00:19:23.559 --> 00:19:29.599 with balancing innovation and protecting people's freedom and privacy and 306 00:19:29.640 --> 00:19:34.640 having that transparency and that when you file a bill 307 00:19:34.839 --> 00:19:37.240 like this, you know you can expect that there will 308 00:19:37.279 --> 00:19:37.599 be some. 309 00:19:37.640 --> 00:19:40.079 Amendments, right right, absolutely, you know. 310 00:19:40.240 --> 00:19:43.079 I Lonsdale, after he put out a little sort of 311 00:19:43.240 --> 00:19:46.440 a tweet threadgras this, he came back not too long 312 00:19:46.519 --> 00:19:50.200 later said he met with quote relevant leaders in Texas 313 00:19:50.400 --> 00:19:53.880 and that they agreed, again I'm quoting him, conservatives must 314 00:19:53.880 --> 00:19:57.119 block the creation of AI regulatory agencies. 315 00:19:57.720 --> 00:20:00.000 So I think, you know, you're exactly right. 316 00:20:00.079 --> 00:20:02.799 It sounds like there is there's going to be some 317 00:20:02.880 --> 00:20:06.200 resistance to some of these ideas. Right here, I want 318 00:20:06.240 --> 00:20:09.279 to talk a little bit about enforcement here. So the 319 00:20:09.319 --> 00:20:14.279 bill as it's currently laid out, would place enforcement in 320 00:20:14.319 --> 00:20:18.000 the hands of the Attorney generals. That's correct, organizations that 321 00:20:18.559 --> 00:20:21.279 violate the rules and that would be subject to fines 322 00:20:21.319 --> 00:20:25.480 in the like ten thousand dollars range per instance. But 323 00:20:25.559 --> 00:20:27.359 of course we know that a lot of these things 324 00:20:27.400 --> 00:20:30.319 are done on a mass scale, so those ten thousand 325 00:20:30.319 --> 00:20:32.880 dollars finds can can really add up. I mean, you 326 00:20:32.920 --> 00:20:38.440 may remember the Attorney General's lawsuit against Facebook for use 327 00:20:38.480 --> 00:20:42.319 of facial recognition. Again, it was one of those you know, 328 00:20:42.400 --> 00:20:45.519 individual finds. Those individual finds ended up adding up to 329 00:20:45.559 --> 00:20:48.559 one point four billion dollars worth of settlements. So you 330 00:20:48.599 --> 00:20:52.000 could really see, you know, some aggressive actions related to this. 331 00:20:52.720 --> 00:20:54.200 But I mean, one of the things I wonder about 332 00:20:54.240 --> 00:20:58.000 is I read this bill is Texas state government isn't 333 00:20:58.039 --> 00:21:01.440 exactly known for being a tech know, logically advanced, and 334 00:21:01.440 --> 00:21:03.279 how it goes about things. I mean, you know, the 335 00:21:03.359 --> 00:21:06.160 computer systems, all those types of things. 336 00:21:06.279 --> 00:21:10.400 Well, there is another bill that calls for training of 337 00:21:10.519 --> 00:21:14.680 certain employees on cybersecurity and artificial intelligence. 338 00:21:14.880 --> 00:21:15.440 Well there you go. 339 00:21:15.519 --> 00:21:20.440 I mean, what does Texas have the capacity to regulate this. 340 00:21:20.640 --> 00:21:23.200 Do you think if even if a bill like this 341 00:21:23.240 --> 00:21:23.640 were to. 342 00:21:23.599 --> 00:21:30.319 Be passed, you know, the Attorney General's Office, as you said, 343 00:21:30.440 --> 00:21:35.079 would be the enforcement mechanism, not the Department of Information Resources, 344 00:21:35.519 --> 00:21:38.119 does it have the capacity? I think that depends on 345 00:21:38.279 --> 00:21:42.559 the who are the employees and what training do they have? 346 00:21:43.039 --> 00:21:46.960 And as I said, I think that is incredibly important 347 00:21:47.920 --> 00:21:51.240 having of course the systems in place, but also the 348 00:21:51.319 --> 00:21:53.240 training of state employees. 349 00:21:53.880 --> 00:21:59.839 Okay, and so the other thing here to be thinking 350 00:21:59.839 --> 00:22:02.960 of is just again that debate. I mean, this is 351 00:22:03.039 --> 00:22:05.119 going on in a lot of different ways. I think 352 00:22:05.160 --> 00:22:07.640 one conversation is do you do this in a piecemeal approach? 353 00:22:07.759 --> 00:22:12.880 Right with individual you know, here's an AI healthcare bill, 354 00:22:13.119 --> 00:22:15.799 as an AI education bill and all those types of things. 355 00:22:15.799 --> 00:22:17.920 There are a lot of different measures about that. But 356 00:22:18.000 --> 00:22:21.079 even when you get into the individual cases, it can 357 00:22:21.359 --> 00:22:22.920 it can get a little bit dicey when you go 358 00:22:22.920 --> 00:22:25.599 into the details. There was a conversation in the Senate 359 00:22:25.640 --> 00:22:29.480 Business and Commerce Committee last week Charles Schwartner finding a 360 00:22:29.480 --> 00:22:32.000 bill that would ban health insurers from using AI to 361 00:22:32.079 --> 00:22:36.759 deny treatment or downcoading patient's treatment. He argued pretty passionately 362 00:22:36.839 --> 00:22:40.599 against the idea I'm quoting here that insurers are quote 363 00:22:40.720 --> 00:22:43.759 deploying AI that is focused on their bottom line in 364 00:22:43.759 --> 00:22:47.359 my opinion, over you know, patient well being basically. 365 00:22:47.680 --> 00:22:49.039 But los culd course. 366 00:22:49.000 --> 00:22:53.119 Another very conservative senator came right back at it and said, 367 00:22:53.200 --> 00:22:57.039 you know, this bill could cost billions of dollars to 368 00:22:57.160 --> 00:23:00.000 the state, right that we have you know, Medicaid programs 369 00:23:00.039 --> 00:23:02.559 and things like that that might be able to use 370 00:23:02.599 --> 00:23:06.119 AI to root out fraud or identify in efficiencies and 371 00:23:06.160 --> 00:23:09.920 things like that. And I understand that there's fears, but 372 00:23:09.960 --> 00:23:12.319 there's also real opportunities here for. 373 00:23:12.279 --> 00:23:15.240 It to get back to the promise and the peril again, right, 374 00:23:15.440 --> 00:23:18.440 And this has been a national issue. You've probably seen 375 00:23:18.480 --> 00:23:25.079 some headlines nationally about insurers using AI and were they 376 00:23:25.119 --> 00:23:30.759 relying solely on an algorithm to deny coverage and insurance, 377 00:23:30.960 --> 00:23:34.680 particularly for specialty care. And there have been several bills filed, 378 00:23:34.920 --> 00:23:38.519 as I mentioned, dealing with healthcare, with mental health, and 379 00:23:38.599 --> 00:23:41.200 with health insurance. And I think the question will be 380 00:23:43.680 --> 00:23:47.200 do you have a bill that denies the use of 381 00:23:47.240 --> 00:23:50.400 AI completely or do you have a bill that says 382 00:23:50.440 --> 00:23:52.440 you can use AI. But as I said, you have 383 00:23:52.519 --> 00:23:54.519 to have a human in the loop, a human has 384 00:23:54.599 --> 00:23:58.680 to review it. And you know there are other bills 385 00:23:58.680 --> 00:24:02.599 that have been filed that say, well, you can use AI. However, 386 00:24:02.599 --> 00:24:05.880 if it's specialty care, the specialist, for instance, has to 387 00:24:05.960 --> 00:24:09.000 review it, so you keep that human in the loop. 388 00:24:09.480 --> 00:24:11.680 And a lot of what we're seeing around the country 389 00:24:11.680 --> 00:24:15.920 and around the world is that discussion of whether it's 390 00:24:16.119 --> 00:24:20.000 in domestic violence cases in Spain where you had a 391 00:24:20.039 --> 00:24:26.880 situation where a particularly police department was relying solely on 392 00:24:27.079 --> 00:24:31.920 AI predictive analysis to say whether an abuse or would 393 00:24:31.920 --> 00:24:34.359 abuse again, and there was not a human in the loop, 394 00:24:34.880 --> 00:24:38.759 and a woman ended up dying. And so a lot 395 00:24:38.759 --> 00:24:42.799 of the conversation now has turned to using it but 396 00:24:43.079 --> 00:24:47.920 ensuring that a human makes the final decision and is reviewing. 397 00:24:48.039 --> 00:24:51.960 So I think it's this discussion of where are you 398 00:24:52.039 --> 00:24:54.519 on this continuum. Do you not use it at all, 399 00:24:55.440 --> 00:24:59.279 do you use it and you solely rely on the AI, 400 00:25:00.160 --> 00:25:03.400 or are you somewhere in the middle in this continuum 401 00:25:03.400 --> 00:25:07.400