So we can start with you, Rob, tell us about what you do what your job description is in the state of play in the House when it comes to AI.
Sure. Good afternoon, everyone. Rob Hicks. I'm the legislative director for Congressman Jay Obernolte. from Southern California. He represents the San Bernardino Greater County area, if anybody knows the southern portion of the state, I've been with the congressman from his first day, this is his second term. So, I started with 2021, been in the House about 10 years myself, previously working for an Indiana member and a Virginia member, who's in the room. I'm from Indiana originally, but none, that's disappointing.
As far as the Congressman's priorities go on AI, he has a master's degree in artificial intelligence, he got it back in the 90s, so he's well ahead of the curve on all this stuff. Quick story about him, he actually was planning on getting a PhD in it, and being in academia, doing research on the topic for his career. At the time, his side hustle in the dorm rooms was video game programming, and that ended up becoming a full time gig for him, so he ended up with just a master's. I think that that experience, for him, in many ways kind of highlights his attitude towards AI. I think, for him, it's anastounding revolution in many ways in the way computing works, becauseit's going from traditional algorithmic computing, to probabilistic computing, in many ways is what a lot of AI is in today's world. I think, for him, it's trying to preserve that entrepreneurialism, and just the spirit of seeing what can happen if you let kids figure out what new technologies can do.
John?
I'm John Beezer. I am a senior advisor, serving on the Consumer Protection Subcommittee staff on the Commerce Committee. My background is primarily in tech startups. I've been here about four years. I happened to meet Senator Cantwell during her five years in the private sector where she got into a dotcom startup, made insane amounts of money, and came back, which was an interesting experience. Again, mostly I've been in tech startups, but advising Cantwell on and off informally, and when she took over as, at the time, Ranking Member of the Congress Committee, I came out to advise full time.
Great. So, John, let's start with you. We saw that last year, Senate Majority Leader Chuck Schumer held a series of AI insight forums, those might or might not pick back up this year, but our understanding is that now the work has really gone to the committee level. Could you tell us what's next in the Senate Commerce Committee, as far as putting legislation together?
Yes, certainly. Full disclosure, I don't actually know. It's been a little crazy. I take the insight forums as a sign that Schumer is a) serious, and b) that he's been very, very careful to bring Republicans along, so it's bipartisan, and there's a serious commitment. When I initially came out here, my top focus was privacy. That has been an ongoing multi-decade frustration for many of us in the room, and all over the Hill. Based on that experience, I can't promise you anything is going to happen, because it just never seems to. But, that being said, I think the level of commitment that Schumer has shown, and that everyone who's gotten involved, is a good sign, and so I think we will be seeing action very soon, probably in a matter of weeks. I've heard rumors about a sort of a position paper that Schumer is putting together, which may kind of be the starting gun. I don't know much about it. I don't even know if that's secret, so please don't tell anyone. So, we're looking for some sign that the game is on, and we'll see. Mostly though, I would say, we're seeing a lot of small bills. There are a few sort of big consequential ones, but also a lot of very focused smaller ones. I think it's been stated publicly that our committee is working on a few, some of the topics we're interested in are deep fakes, regulatory structure that does the job without impeding progress, competitiveness, bias. I'm probably forgetting one, but those are our priorities.
Senator Cantwell had an idea for a GI bill for AI and the workforce, if you recall, could you remind the audience what that was exactly?
Thank you so much. That's the big one that I forgot.
What's next with that?
There's kind of a debate about whether we should look at huge existential risks, or just address what's right in front of us. Personally, I kind of lean towards the existential risks, because I do think they're credible, and I do think we have time to prepare. That being said, I think the most obvious challenge that's in front of us Is workforce impact. There was a comment made by Mustafa Suleiman at Davos this year, that basically AI's main function is replacing labor. I think that's true. There's some arguments that it will create so many new opportunities that there'll be lots of new jobs, and everything will be fine, but I don't think that's guaranteed, and I think we need to be prepared for it. So, I can't say too much about what's in the AI GI Bill. Some pretty obvious things, like training and stuff like that. But, I think we're still trying to get a handle on what's actually going to happen, but I think it's probably the clearest, most obvious thing that we need to be working on right now.
Then Rob over in the House, can you touch on what the state of play is there, and then specifically get into some of these funding issues. We saw that President Biden signed an executive order that tasks all sorts of agencies with all sorts of new responsibilities, and multiple have come out and said that they don't necessarily have the resources or the funding for that. Can you talk about Congress's role in just funding these agencies, and then, specifically, NAIRR. I know your boss is a proponent of the Create AI act?
Sure. The past year has been spent on the House side, especially, in many ways, getting members up to speed, and my boss has, frankly, played a decent role in that. Obviously, when you have a member with his degree and background, I think a lot of people just kind of went, Jay, what am I supposed to think? He's a great educator, and he's incredibly smart, obviously. From our perspective, moving forward, there's a little bit of, the members who are interested, they're going to be part of whatever that process looks like to try to address AI for the near term future. Other members, they'll look at AI from their lens of whatever committee they sit on, and they'll address it when they need to, but they're not concerned with being an expert, which is fine, but members can't be an expert on everything.
Regarding the funding situation surrounding AI, there's obviously going to be a lot of research that is going to need to be done, to some degree. The EO obviously, like you mentioned, stood up the National AI Research Resource, for those who aren't aware. It's effectively leveraging compute power to allow academia, mom, and pop businesses, single entrepreneurs, students, to just, frankly, kind of poke at it, and see what they can get. Because I think, going back to my boss's days in his dorm room, one of the big things that separates AI from previous computing revolutions is that you need a lot, in many ways access to massive amounts, of compute to really see what some of the frontier stuff can do. The federal government, frankly, is one of the largest compute controllers on the planet, with all the labs that DOD controls, and the other science apparatus that we fund.
So, my boss has a bill, it's called the Create AI Act, which is, in many ways, basically just formalizing and making permanent the pilot program that the EO stood up, with the NAIRR. I think, in many ways, the boss is a big believer in trying to provide access to what this technology can be. Obviously, there's a longer term science funding question there. What does this look like? And, how do we ensure the US is a leader in AI technology moving forward? But, that's part of the annual appropriations contest?
What's your read on the prospects of these various agencies and efforts actually getting funded?
I think they certainly will, the question is, to what degree? Obviously, we're still trying to work out funding for FY 24, before we even move on to FY 25 for the coming year. It's a long term paradigm, people need to think about, too, because, in many ways, the federal government never had to do with computing the way it did 30 years ago, and then it's just a paradigm shift of, Oh, yeah, we've got to have cybersecurity, like I said earlier, DOD has entire supercomputers devoted to their own research. That's just a paradigm shift that I think, we'll eventually have to figure out.
We had Congressman Beyer here earlier, noting that there are 191 AI bills circulating in Congress right now, and he sees four to five or maybe even 10 getting done. Could I just get your reaction to that. Do you see that happening? And, how would you prioritize which bills should maybe get done first?
I think my boss would love to see the Create AI move, obviously. Trying to put a number on it, I don't know if it's helpful for the conversation. My boss views, doing AI, it's not one bill, it's multiple bills over multiple years, as the issues arise. For him. it's kind of the opposite of what the EU is done. They've had their AI Act, I don't know if those who are following it or not, they had this whole big package spun up to address AI, and then they had to halt and rewrite it in many ways, because generative AI came online, and that wasn't even addressed in the original form of the bill. Som I think my bosse is concerned that trying to do too much too fast, just opens you up to a) you're locking out potential innovations, and b) you're not able to react fast enough as things develop. He's got such a such a strong tech background that, for him, this is just another stage of computing, so he is definitely one to say, let's wait and see what real issues arise over time.
John, what's your reaction?
I set out a few months ago to create the ultimate AI Bill spreadsheet, and my number is 85. So, my initial reaction is I need better search terms, maybe? 85 that are introduced, and maybe another dozen or so that are being talked about. So, there's 50 or 60 that I haven't heard about?
I just want to sort of second on the Create AI act. That's probably the bill that has got the best prospects. Maybe there's some others I'm not thinking of. But, I think Democrats look at it, and see a way to level the playing field, and I think Republicans look at it and see there's sort of a danger of market concentration, and NAIRR adresses both of those, so I think it's the closest thing to a slam dunk we have. In terms of what we're actually going to do, again, it just seems very dynamic right now, so I don't want to make any definitive predictions, but what I think would ordinarily happen is there would be some kind of Christmas tree bill that you stick all the ornaments on. There may be a few candidates for that, but, for various reasons, I'm not entirely sure we have that yet. So, that's kind of what I'm looking forward, that's what I would expect, but I do think everything about this is unusual, and we could see almost anything happen.
And then Rob, there's also a working group in the works. Can you talk about how that might be different from the AI Caucus? And when we might see that being officially launched?
I would say, we're waiting. It's the Speaker's prerogative at this point. Obviously, it's been something we've been looking for with their blessing. We're hopeful that something may come soon, but, again, that's beyond my pay level. In terms of what that is, versus the AI caucus, the vision my boss has spoken about, it's a little more formalized, it'll almost be a cross section of members from different committees, potentially, where their interests lie, because what we've been saying, AI is not one thing, it's a tool. If, at the end of the day, it doesn't need its own thing, when you talk about legislating on AI, it's what does it mean for each specific issue area.
You mentioned Speaker Johnson, he's new. We're also seeing that there's going to be a shake up on the Energy and Commerce Committee. Speaker Cathy McMorris Rodgers is going to retire. What should the new leader of E&C prioritize when it comes to tech policy, and what direction should the committee take in your view?
Well, I think my boss would certainly say, keep beating the drum on AI. Obviously, there's a lot that E&C has jurisdiction over there. We'd love to see them continue their work in that space. They had subcommittee hearings, earlier last year, on every topic that they had jurisdiction over, so just continue beating the drum.
On the E&C, we've seen that there's been a big focus on the nexus of privacy and AI. How do you see that work continuing? Can we expect the ADPPA to be reintroduced this year?
That is for the committee to announce when they feel ready. My boss has obviously been extremely supportive of privacy efforts. He has said that he does think we need to preempt state activity on privacy, and have a single national data standard. That's obviously something we'd continue to support in whatever form ADPPA may take up.
John, what's your read on regulating privacy this year, and where it might go, especially as it relates to AI efforts?
Well, this is a 25 Minute panel. But, let's go back to 2000, on Senator Cantwell's first campaign, she had two issues. One was universal health care access, and the other was doing something about Internet privacy. People used to laugh at us and say that, you guys are crazy, you're never going to get health care. Only tech audiences laugh at that joke. So, obviously it's very, very important to her. It's been something I've been focused on. I'm definitely paying close attention, but, like everything I seem to say today, I don't have definitive answers, but I can speculate based on what I've seen.
There are three scenarios that could play out in the near future, and I think they're equally.... Do I look like I've been working on privacy for years? Yes.
I think they're probably equally probable, so let's just call them 33% odds. One is, we do nothing, and the logic on that is, that's what we always do. The second one is that we go, Well, you really can't address AI without privacy in some way, so maybe perhaps a limited privacy bill would make sense, or privacy components within an AI bill, and in particular, I think there's things like training models on sensitive personal data, that should just obviously not be legit, not without authorization. Things like that need to be addressed immediately. Then there's also the possibility that maybe, the sort of we meet the moment, that AI is so transformative, and such a big deal, and we're making such a big bill around AI, that maybe we just kind of throw privacy in, and do comprehensive privacy at the same time. I wouldn't rule that out. I wasn't born yesterday, so I don't expect it to happen, but it could. So, we'll see how it plays out.
And, how would you describe the communication and coordination between the Senate and the House on both of these issues, privacy and AI, and how the two might relate at this point?
I've been here for years, we've been having conversations the whole time. I would describe it as, the conversations ratchet up and they ratchet down, but they never actually stop. In particular, I guess it's about three years ago now, it was basically after the Facebook stuff, that was sort of huge and really motivated a lot of people, we made a commitment at the time, Cantwell was Ranking Member, and Wicker was Chair, and we made a commitment to do whatever it would take. We had very, very constructive discussions that that were like twice a day for an hour or two each, on and on and on for weeks, and we did whittle it down to just a couple sticking points, one of which you mentioned, which is preemption, and then private right of action being the other one, and there are kind of a few other minor ones, but we got it down to that. We were even discussing ideas that could have been compromises, and then it's kind of like the tide went out. We just didn't quite get there. But, the bill that we worked on eventually did become ADPPA. That had some things in there that we couldn't agree on, and so we didn't support it, unfortunately. But it could, we're much closer than it looks. We've just been working on this a very long time, and it's very hard, and I think we're very close, so anything could happen.
I would say I'm not a bit involved with a lot of those discussions previously, but I would say, committees are still talking, in as far as I know. From an AI perspective, I've had countless staff meetings, House and Senate side, just staff interested to know more and try and figure out what the technology is about, where there boss should be on it, so I certainly feel like I have a lot of the relationships needed.
In the meantime, we're seeing that states are moving forward with their own AI legislation, and, of course, we have the EU AI act. How does that impact your work on the Hill? Do you feel increased pressure? Is it informing lawmakers? How would you describe that impact?
Sure, I think it certainly gives a sense of urgency to a lot of what we've been thinking about. My boss has said that he would preempt state activity on AI. Part of his rationale for that is that, like I said earlier, AI, as we know it today, is fundamentally probabilistic software, and ultimately, it's based on the data that you're giving it, is how it gives you good or bad results, garbage in, garbage out. One of the things I think it's easy to imagine is, you could see where states balkanizing how data gets handled across the country, where in some capacity they're requiring affirmative user consent to use their data in some way. My boss says these systems are incredibly valuable when they can ingest an enormous amount of data, and if you're turning off the spigot, so to speak, on that, you're limiting the potential for the benefits they're going to bring to society. In many ways, this is an industrial revolution of a new sort that we haven't seen yet, and I think my boss is really concerned that if you're fracturing the ecosystem of where AI can be used, and how you can use the data needed for it, you're limiting the benefits that everybody can eventually have.
Interesting. I'm going to disagree. But, I also want to make it clear that this is a personal opinion. When I say we're close, we are very close, so, if I express a strong opinion on preemption, that is my personal opinion, not the committee. I think it's a good thing that the states are putting their own laws together. I don't agree with all of them, but I like some of them, or some aspects of some of them. This used to be kind of a Republican talking point, but I've always liked the laboratory of democracy concept where, if it's something as difficult and subtle as privacy, maybe we should experiment, maybe various states should try different things, and kind of see what works and what doesn't work, so I'm in favor of that.
Also, a quarter of the states have privacy in their state constitution, and the federal government does have the right to overrule that, but I think that's something you need to think very, very carefully about. If it's something important enough that it's in the state constitution, we need to think twice about overruling that. I would also point out there are several different sort of flavors of preemption. Floor preemption just means don't make any law weaker than ours, go ahead and make it stronger if you want, so there's some flavor of preemption that will do the job, and then...
I'm getting the high sign over there. I could go all night,
I do have one more for you, just one more question. We saw, on social media, how legislation there hasn't moved because of lobbying efforts on the Hill. I wanted to ask you both how you see that playing out in the conversation on legislating on AI, and how can your bosses prevent the same problem from happening, if you agree that it is such a huge issue.
Frankly, it needs to be a collaborative effort. Right now, it's a decent time for industry and associations to be weighing in, because it's green pasture. We don't know, inherently, what the technology is really going to bring to efficiencies within the economy, and I think it's recognizing that. There's a sense that there is a little bit more risk to it, because of how widely it can be applied, whereas the Internet initially was kind of this, here's this quirky little thing kids are playing on, and maybe you'll buy your clothes online someday, that'll never happen. AI, there's a sense that this is a transformative shift, and that is giving a greater sense of gravity to the way people are having those discussions, so that is helping all those conversations we've had thus far.
Yeah, it's definitely a factor. I've heard stories from when the famous 32 words in Section 230 were written, and they amount to, well, we really didn't realize this was such a big deal. Unfortunately, I think we have passed that stage, so people are aware that this is a big deal, and so all interests are fully activated. There's a saying I hear a lot, which is, for every com staffer on the Hill, there's, like, 4000 lobbyists out there, or something ridiculous, so we're outnumbered, we're outgunned. But, we can haul people in front of hearings, we have certain powers ourselves. I don't feel like lobbyists are calling the shots, they definitely have a seat at the table, there are definitely a lot of interests, and we're not trying to put anybody out of business. If we're proposing something that's not going to work for someone ,we want to know about it, but I don't feel like we're being pushed around.
All right. Well, that's all the time we have. Thank you both so much. Thank you.
