Could artificial intelligence destroy us all?

Episode Notes

Transcript

Welcome to the ABA Journal Legal Rebels Podcast, where we talk to men and women who are remaking the legal profession, changing the way the law is practiced and setting standards that will guide us into the future.

Victor Li:
On the special uplifting episode of the Legal Rebels Podcast, we’re going to deal with a very warm and fuzzy topic that makes me just so happy and so optimistic for the future. Could AI end humanity as we know it? Sorry about that. I mean, maybe AI wrote that sentence and forced me to say it. But joking aside, the last few weeks have brought some pretty scary headlines when it comes to artificial intelligence. OpenAI admitted to hacking into government databases in Australia. Several of its AI agents went rogue and probed three government websites in the United States over the summer. Another AI agent tried to hack into a Canadian government website. And of course, it was a really scary headline from a few weeks ago when anthropic researchers acknowledged that AI could kill us all by the end of the decade. So with all that in mind, here’s one of our old friends to talk to us about why we shouldn’t start doomsday prepping just yet, or maybe we should, I don’t know.
Maybe he’ll have some good tips for us. My name is Victor Lee and I’m the assistant managing editor for the ABA Journal. My guest on today’s episode of the Legal Rebels Podcast is Damien Riehl Damien is solutions champion at Clio, where he travels the world talking about AI and how it’s changing the legal industry. And Damien is here today to talk about whether AI needs guardrails, and if so, how should they look and what they should do. So welcome to the show, Damien. I appreciate it.

Damien Riehl:
Thank you so much, Victor. Real thrilled to be here and talk about this, maybe the most uplifting episode we’re going to see today.

Victor Li:
Well, I mean, you’re always so positive and so easy to talk to, so I figured if anyone could put a good spin on it, it’s you. So I appreciate it.

Damien Riehl:
Yeah, yeah. So generally, there’s optimists, there’s pessimists, but that kind of takes away the agency that we have a direction as lawyers to play in whether humanity ends or whether we flourish. And so I would say that I’m less optimistic and I’m more trying to push us toward the good to be able to say that we can be able to have the rule of law and AI and maybe help humanity.

Victor Li:
Gotcha. Before we get into all that, obviously, well, you’ve been on the show before, but for people who might be listening for the first time or might not know who you are, I don’t know why you’ tune into the show if you don’t know who he is, but can you tell me a little bit about yourself and your background?

Damien Riehl:
Sure. I’ve been a coder since age 10, but a litigator for about 15 years. And so I’ve built product for Thomson Reuters and Fastcast, which turned into VLEX, which turned into Clio. So I’m really a builder that is a product person that happens to be good at putting words together and speaking. So I’ve since ChatGPT given about 312 talks on AI and the Law, both in person and in podcasts. And so I’m a builder that also talks about things.

Victor Li:
And you also like to sing too, right?

Damien Riehl:
Yeah. So my bachelor’s degree is in music. I was going to be a choir director until two of my tenors started to punch each other in the face in the middle of the Brahms piece. So I though, oh, I should go to law school.

Victor Li:
But that wasn’t your fault, right? Well,

Damien Riehl:
I was conducting pretty aggressively, so it might have been my fault.

Victor Li:
I mean, hey, singers, hey, they’re divas for a reason, right? I mean, that’s why,

Damien Riehl:
Right? It’s true. It turns out the boys were fighting over a girl, so I wasn’t able to mitigate that risk.

Victor Li:
Yeah, nothing you can do about that. Well, I mean, I bring up the singing because very often nowadays whenever I go to a tech conference, I’ll often see you singing songs about AI. And so we were hoping maybe you could sing a song for the Legal Rebels Podcast, but Rebel Yell probably isn’t that appropriate for –

Damien Riehl:
Oh, that would be a good one. My voice teacher in college was a part of the Vienna Boys Choir, so he was an Austrian guy. And when I told him I was going to go to law school, he said, “What are you going to do, sing to the jailbirds?” So I would love to do a whole podcast with you in song, but that’s going to be another day.

Victor Li:
Gotcha. And hopefully we won’t be jailbirds at that point.

Damien Riehl:
Well, yeah. I mean, the AI might be locking us up.

Victor Li:
Yeah, that’s true. So I mean obviously you’ve been in tech for a long time and you’ve been doing this for a while, and maybe it’s just because just that’s where the industry is heading and that’s just everything was just changed so dramatically. But how did you get into AI then?That’s just what people are doing and that’s a good thing and they want to know about it?

Damien Riehl:
Well, AI is more a marketing term than a technological term. When I was in Thomson Reuters, for example, we were building a workflow system to be able to say, what are all the lawyers’ tasks? Maybe for any litigation task or transactional tasks, let’s break them into that matter into a thousand tasks and then figure out how to do each of those tasks. So of course in 2015 we’re saying how can technology help humans do this? And now in 2026, we’re saying, how can we help robots do this? How can we automate more of these tasks within any given matter, whether it’s a litigation matter or transactional matter? So really is that kind of turbo taxi thing we were doing in 2015 was that AI and was Excel back in the day, AI. So I think that AI as a term, we’ve had “AI” for 50 years, but it’s just marketing.
It’s not really a technological term.

Victor Li:
Gotcha. But while we are talking about the actual technology of generative artificial intelligence and whatnot, you talk about how it’s made an impact and how it already changed the legal industry. So where would you say we are as far as lawyers, people in the legal industry or whatnot? What step are we as far as with regards to AI adoption and whatnot?

Damien Riehl:
I would say we are in the very, very early innings. William Gibson said the future is here, it’s just not evenly distributed. And I think we, that is Victor and I are kind of in a bubble where we’re seeped in this every single day. But I was just at a conference where I spoke to 600 lawyers and they are just clueless as to the freight train that’s coming down. So I think there is what the AI is capable of, number one. And then number two, what’s been adopted by lawyers. And I think that there’s a vast chasm between those two things. And so I would say that there’s a saying with my Silicon Valley friends is that there’s an overhang, that is the overhang of what the AI is capable of versus what is being adopted. And I would say for society, the overhang is quite large and for lawyers, the overhang is even larger.

Victor Li:
Okay. And where would you put politicians on that scale? Would you say they’re even behind lawyers?

Damien Riehl:
Yeah. If you’re an oxygenarian, 80 plus years old, what do you know about AI?

Victor Li:
Right. Well, unfortunately it’s like, well, I guess they know enough to rename it other things.

Damien Riehl:
Yeah, sadly. And which is difficult because you may have heard the superintelligence is now the name that some of our politicians have said, but there’s already been for the last 10 years something that is called artificial superintelligence. So that has already been a thing to be able to say there’s artificial intelligence AI and then there’s AGI, artificial general intelligence saying that is AI better than most people at most tasks? And then there’s artificial super intelligence, ASI, that is saying is AI better than all people at all tasks? That is better than any Nobel Prize winner, et cetera. So anyway, so that’s ASI, Artificial Superintelligence, is something we’ve had in the nomenclature in at least the valley for many, many years. But now all of a sudden our politician is now saying superintelligence, which okay, when we talk about superintelligence now, are we talking about the politicians’ superintelligence or are we talking about the artificial superintelligence that we as computer science has been talking about for the last 10 years?

Victor Li:
I would wager is probably not the latter, probably. But I mean, what did you make? I mean, because obviously the law always lags behind innovation, that’s just the way it’s always been. I mean, what have you made of the last couple of months where it just seems like all of a sudden there is this kind of push for some kind of governmental action, if not calling people in for hearings to maybe even start the groundwork of some kind of regulations? I mean, do you think this is all just theater or do you think there is some substance behind it?

Damien Riehl:
Law has always lagged technology. So I started law school in 1999 when Napster was a thing. And how do you regulate Napster? And I wrote my Law Review article in Napster. And so of course it took many years for Napster, of course, to work through the courts and then of course it led itself to Spotify. So that was like an eight-year period where there was a real question mark as to the law lagging the technology and then eventually the law didn’t even catch up with the technology. The industry, the music industry in that case just caught up with the technology. So the law is always lagged, but the problem is that this technology, the AI technology is going exponentially faster than any other technology that we’ve had, even faster than the internet, even faster than anything before that. So Ray Kerzweil, an author, talks about the singularity and the singularity is essentially where machine intelligence reaches human intelligence.
And his thesis back in the 1990s was that exponential is going twice as fast. So what used to take 10 years now takes five years and now takes two and a half years, now it takes a year and a quarter, now takes six months, now takes three months. And so really that kind of exponential growth and speed we’ve been kind of going in since the 1950s and now we’re kind of reaching the hockey stick growth to be able to say that you and I have been seeing this, that what used to take five years all of a sudden happens in a year and then all of a sudden happens in six months. So what choice going back to will the law catch up, what choice does the law have? How can the law catch up with doubling of exponential growth every six months or every three months?
So whatever we put into law that takes a three month window to be able to even enact a law, now it’s going to be obsolete by the time the next technology comes in.

Victor Li:
Gotcha. And I mean, whenever I had this conversation with other thought leaders in the industry, it’s like one of the arguments, and not the ones they put forth, but just one of the arguments out there is that there is a framework of laws that could potentially be used to cover AI, but obviously there would have to be some updates and whatnot. But does that even seem possible in this phase considering how quickly the technology is changing?

Damien Riehl:
I’m actually an optimist to be able to say yes, I think that existing laws will be able to cover a lot of the danger that we’re facing. So for example, that you may have heard of the OpenAI hugging face hack, so the OpenAI’s escape containment and hacked hugging face. And it turns out that they probably committed a felony under what’s called the Computer Fraud and Abuse Act, CFAA. There’s an existing law that already covers the thing that OpenAI’s tools did. So really, do we need a new law to be able to say AI is escaping containment or do we just use the existing law of the Computer Fraud and Abuse Act to be able to mitigate those harms? And so now if we could say, okay, well, how do we prevent the next hugging face hack? How do we prevent the agents from breaking containment and hacking hugging face?
And OpenAI and Anthropic, Sam and Dario would say that we need to align AI with human values. And my response to them is we’ve kind of had alignment with human values for about 800 years. It’s called the Law because Victor’s human values are different than Damien’s human values that’s different than Jane’s human values. Each of us has different human values, but the way that we collectively have agreed on those human values is through the law and through both the statutes, the regulations, and then the cases that interpret those statutes and regs. And so really, what if you had the AI follow the law? What if as the agents go out and do their work, before they do every step, they consult the legal Oracle and say, “Does this violate the law above a certain threshold? And if it does, let a human know.” And that maybe could have prevented the hugging face incident.
Say, “Oh, look, there’s a law called the Compute Fraud and Abuse Act. Does this violate that? Oh, I better let a human know.” So if we want to be able to avoid the dangers of AI and maybe ending humanity, we should probably have the AIs align with human values, I’m putting in air quotes, which is following the law, align with the law. And then through that, we could be able to say that that is what we collectively have said our societal human values are. But of course, our law in the United States is different than the European Union law, which is different than the Latin American law. And so one can imagine what if all the countries put all of their laws into this legal oracle and as the agents go around the world, you could be able to follow those laws. And then you could also have human values that Christians have different human values than Jews, which have different human values than Muslims.
So you can imagine then having another layer to be able to say, “Hey, what are my Jewish values?” And then be able to say, “I want to follow those Jewish values. What my Muslim values or what are my Christian values?” So within that, and then I don’t necessarily agree with all my Christian things, so maybe this is Damian Real’s values and then the AIs follow. So I think there’s probably concentric circles of humanity law, UN law, and then my country’s law, my state’s law, my industry’s law, my ethical law, and then my household law and then my individual law. So each of these is a concentric circle that we can, if we want to avoid ending humanity, probably do these things.

Victor Li:
I guess my question then would be like, okay, well what if, I mean, how about concentric circles, but what about areas where there would be conflict? Let’s say if you and I have diametrically opposed views about fill in the blank issue, then which one would take precedent, right?

Damien Riehl:
Yeah. So I have actually really good example for this. A friend of mine works for a cloud computing company, so you know the name, I’m not going to say it, but she’s a lawyer saying that her job is pulling down objectionable material. And she said that she, under Nazi material, under German law, needs to come down because German law says that Nazi material needs to come down. But under Texas law, say the free speech law, it needs to stay up. So she says, whose laws do I follow? Whose human values do I follow? German human values or Texas human values? Because I can’t do both. The website either has the thing or doesn’t have the thing. And so I think that’s something that we lawyers have though about all the time. There’s conflict of laws. There’s conflict of German laws conflict with Texas laws or New York laws conflicts with California laws.
And the way that we humans lawyers do this is to be able to say, okay, what is the least worst option? What is the penalties under German law? What are the penalties under Texas law? What’s the likelihood of the regulator going against me in Germany? What’s the likelihood of the regulator going against me in Texas? And then choose the least worst option. That’s what we as humans do. And so maybe the AIs do the same thing. If they have the legal oracle to be able to say, okay, here are all the laws in all the jurisdictions that might be here, what are the likelihood of getting dinged in this jurisdiction? What are the penalties if I do get dinged? And then make the decision on the least worst option.

Victor Li:
Gotcha. Before we continue, let’s take a quick break for a word from our sponsor. And we’re back. So talking about legal frameworks and whatnot, I mean, just going back to the topic of existing legal frameworks, what are some gaps that you see as far as things that might not cover AI? I

Damien Riehl:
Think that as AIs are more agentic, that is more agents going out and doing things in the world and they’re acting more human-like in their work. We’ve passed the Turing test, which is can you tell if the thing on the other side is a computer or a human? That was the Turing test. We passed that years ago, probably 2023 we passed that. So if you can’t tell whether it’s a human or a bot, I think that increasingly all of the human regulatory things and human legal things will apply to the bots without much change. Because if I can’t tell whether it’s a human hacking my system or a bot hacking my system, I probably don’t need to change the law either way that the human or the bot needs to stop hacking my system. So anyway, so I think that the number of regulatory things is probably a few, but we just need to enforce those laws.
And so the question is, is Hugging Face going to sue OpenAI for this hack? And Hugging Face shortly after the hack got acquired by Nvidia. So what are the odds of NVIDIA going after OpenAI? Next to none. So really, I think the real question is not do we need to enact new laws or regulations to apply to the AIs, but really the question is for prosecutors and for plaintiffs to say, are they going to sue over these existing laws? And what’s the incentive to be able to sue or prosecute or not sue or not prosecute?

Victor Li:
And I guess the video thing brings up another thing because I mean, I brought this up, I guess maybe a little bit earlier, but as far as companies arguing that they want to be able to regulate themselves, have very minimal government intervention, be able to maybe the experts are the ones who are best suited to determine these kind of things and whatnot. I mean, does argument resonate with you or is it kind of like a, okay, well, what if you have a case where you used to have these companies just don’t want to deal with the government and just want to be able to do whatever they want and then stuff happens and then things go out of control?

Damien Riehl:
Sure. So I’ll play both sides. As a litigator, I like to do that. So one could say that what are the odds of that an 80-year-old in Congress is going to know the heart of what the AIs are capable of? And even if you put that 80-year-old within the four walls of the AI company and they saw that, almost everything that that 80-year-old would see would go over their head. So I would say on the pro AI companies side, I would say yeah, nobody can regulate the AI companies like the AI companies because they just have the expertise in the way that the regulators don’t have. And what are the odds of the regulators paying somebody exponentially high amounts of money that the Silicon Valley engineers have? What are the odds of that a regulator paying that high income person to be able to regulate the industry?
And I would say that next to zero. They’re not going to be able to hire the regulators to be able to go after the AI companies in a very knowledgeable way. So that’s on the pro AI side. On the other side, there’s something called regulatory capture, number one, and there’s also the prisoner’s dilemma. So with regulatory capture, you could imagine that a lot of people are saying that OpenAI, Anthropic, and Google and Grok, et cetera, all of them want to be able to get the regulators in their back pocket so that they raise the regulatory bar and essentially lift up the ladder behind them so that no small companies can be able to compete with them because the regulatory burdens are so high. So that’s what’s called regulatory capture. And then there’s also the prisoner’s dilemma. And this is the idea that all four of these companies are really competing against each other a lot.
So OpenAI, Anthropic, Google and SpaceX. And so they’re all running as fast as they can to be able to create the most intelligent models. And so as a result, everyone’s kind of running toward the cliff. If we think that there is a danger of humanity ending and that danger is a cliff, then we’re all running toward that cliff as fast as possible. And so how do you slow down that running toward the cliff? Well, if I’m competing against three other companies doing this, there’s no incentive for me to slow down unless the government slows me down. And this is what you saw with Sam and Dario and Elon and Google saying, “Hey, regulate us because we are in this prisoner’s dilemma. We are competing against each other.” And the fact that they all agreed to it so quickly is really mind-boggling to say, “Hey, yes, regulate us.” And they all agree.
And I would say that Elon Musk and Sam Altman don’t agree on much for them to so quickly. I think it makes me think that they saw something either in the external or in internal models that scared them. So I’d say that we should listen to them and we should probably not run toward the cliff quickly as we are running toward the cliff.

Victor Li:
Gotcha. So what might a regulatory framework then look like as far as you’re concerned?

Damien Riehl:
I would say that a regulatory framework would be, hey, the AIs should follow human values, also known as the Law. And what do we as lawyers have to do that? We are stewards of the law. So we, every lawyer listening to this podcast should be saying to our stakeholders, that is to our Congress people, to our senators, “Hey, we have these laws for a reason. The AIs, as they do more work in the world, should follow these laws because the laws are the best way that we’ve aligned with human values.” So I would say that the existing things on the books, let’s enforce them. Let’s enforce the antitrust laws. Let’s enforce hacking laws. Let’s enforce all the laws that could be broken because this is the way that we keep our society running and not run off the cliff.

Victor Li:
So I mean, with the current administration though, as far as following the law, enforcing the laws and whatnot, does that then make it more difficult for a potential regulatory model? Because well, if we’re going to follow the law, well, what law are we following? Are we following the laws the ways it should be or the way it’s being enforced right now?

Damien Riehl:
Well, so let’s think about the ways that companies follow the law in the old world and then how that relates to how companies follow the law in the new world. So in the old world, I’m a builder, say a product person within OpenAI Anthropic. In the old world, I would essentially say to my lawyer, “Hey, should we do this thing or not?” And my lawyer would say, “Oh, we shouldn’t do it or we should do it.” And they would look at all the laws. And so that’s option one is that they consult the lawyer before they build it, but that almost never happens. The builders just build it and then the lawyer has to say, “You did what? That’s violating all these laws.” So option two is they kind of clean up after the fact. But option three today is the AIs can know the law as it’s being built.
So really this option of being able to take all of the law, and we already have something that is privacy by design, that it’s built with privacy in mind, and security by design, maybe we could have legal by design. Maybe we could be able to say every product being built by the AI companies is out of the box following the law. If we do that, then you don’t have to have option two lawyer coming and cleaning things up, or you maybe don’t even need the option one lawyer saying beforehand whether you can go out. Maybe the bots will just know and they’ll know to be able to not break the law. And so this is something that my friend Ron Friedman has talked about, preventive law, do less law. So maybe we can have the grand promise of doing less law if we bake the law into the models to keep them from doing bad things.

Victor Li:
So how would that look like then? It would just be a matter of engineers. I mean, would you need more active involvement of lawyers at the ground level as far as when they’re working side by side with the engineers and coders and whatnot?

Damien Riehl:
Maybe, but the question is my human lawyer brain, how does that compare to the AI’s legal brain? Because you might probably know that in 2023, a GPT-4 beat 90% of humans in the bar exam, but it didn’t stop there. My friend Matt has continued to study it and Fable gets a 100% on the bar exam. And how many lawyers do you know that got 100% on the bar exam? I didn’t.
Right. I didn’t either. So really the question is, do I want my stupid meat brain doing the interpreting of the laws to be able to put it into the product? That’s option one. Or option two, have something that got 100% on the bar exam with access to yesterday’s case, yesterday’s statute and yesterday’s regulation, not just in the United States, but in the European Union and in Latin America and Hong Kong and Singapore and take all of these and be able to do the analysis of is what’s being built by these companies following the law of that jurisdiction or not? And so I would say that we could probably automate a lot of these processes being able to make sure that the AI follows the law, but that’s only possible. What I’ve just said is only possible if the law is free. That is right now to download Pacer from the federal government would be about $2 billion, would it be?
Almost nobody’s spent those $2 billion, but we at Clio have. And so because we spent a lot of those $2 billion, we can tell you what are the odds of winning this motion to dismiss for breach of contract in the Southern District of New York. That’s because we’ve spent that $2 billion. But you know who doesn’t have access to that? OpenAI, Anthropic and Google. And you know who also doesn’t have that is poor people, the 92% of legal needs that are unmet. So really, what if Pacer didn’t cost $2 billion, but what if it were free? And what if OpenAI or Google or Anthropic had that? And what if poor people had that too? How much better would that be for access to justice? So I’d say for this grand AI following the law to work, the law has to be free. And there’s a saying that ignorance of the law is no excuse.
You’re bound by the law whether you know it or not. And then the next question is, what’s the excuse of hiding that very same law behind a paywall saying that you can’t get that law unless you have to pay for it? I think this is the question for our times. Is government going to make the cases free and make the cases free, not just because it’s the right thing to do, because it is the right thing to do, but also because the AIs need those cases to be able to analyze, to be able to make sure that we don’t go off this cliff.

Victor Li:
And that’s just Pacer? Th doesn’t even include all the state courts and all the local courts that are not accessible, right?

Damien Riehl:
That’s right. So I’ve given this talk for judges and they said, “Yes, it’s true.” And I says, “Judges, this is an access to justice issue that you created. You’re giving rich people better justice than poor people.” They would say, “That’s true, but the conference we’re at was paid for with Pacer dollars.” So really the solution number one is for Congress to fund the courts sufficiently so that Pacer can be free. And then you could also imagine, just like in the 1980s, the federal purse string said, “Hey, the state speed limit,” which is usually a state run thing. If you want federal dollars, you have to make the speed limit 55 miles an hour back in the 80s. So you can imagine the same thing happening with the courts to be able to say, “Hey, state courts, AI is following the law of New York, the law of California, the law of Texas.
We need to make sure that your laws are free and open so that the AIs can adjust them.” So you can imagine a federal act essentially pushing down just like the speed limit mandate to be able to say, “Make the law free.” And if that’s true, then I think our society will be much safer because we’ll have legal by design. The bots will follow the law by design of Texas and California and New York.

Victor Li:
All right, let’s take another quick break for word from sponsor and we’ll be right back. And we’re back. So now the big question for you, Damien, is AI going to destroy all of us?

Damien Riehl:
I would say that there is a multiverse of many potential futures. And in the multiverse, of course, with infinite futures, of course, AI destroys us in some percentage of those futures. So what percentage we can argue about and what is the timeline? Does it destroy us in a thousand years? Does it destroy us next year? So really there’s an open question as in the multiverse, how many worlds does it end with humanity being destroyed? So I would say in that, and the multiverse isn’t just something you see in Marvel movies, it’s actually something that physicists are saying that, yeah, this is actually true. So in the multiverse, the real question is what percentage of these futures does AI kill us all and in what timeline? I think that’s the right way to frame it. And I’m not going to put a number as to what percentage of the futures get humanity dying, but what I will say is that we should all, we lawyers listening to this should be pushing us toward the good, that is pushing us away from the futures where we get destroyed and toward the futures where we will actually thrive as humanity.
And I think that we do have many paths to humans thriving if we do it right. And we as lawyers are guardians of the law. And so we should be using our power as guardians of the law to be able to push toward those futures where AI follows the law. And I talked with a big chief scientist of one of the big AI labs, one of the big four, I’m not going to say their name, but one of those chief scientists said, “Oh, we’re aligning with human values.” And I said, “What do you mean human values?” He’s like, “Oh, the user’s intent.” I said, “Hey, your user could be a psychopath. Don’t follow those human values of that psychopath user. Follow the law. I mean, those are the human values that you should be following.” And back in 2023 when we had that conversation in 2023, he said, “No, no, no, we’re going to follow the user’s human values.” So anyway, so I think the technologists just don’t understand that really the law is not just an annoyance that they have to be able to get around, but the law is the human values that we societally have agreed to.
And if the AIs don’t follow societal human values, then we’ll end in a bad humanity getting driven to extinction. And to think about some of those bad futures, how that might happen, it’s good to give a small anecdote that I’ve been coding with Cloudcode since January. And when I started using Cloudcode in January, I would say, “Hey, I don’t trust you cloud code, so come to me before you write anything to Disc.” And then it would come to me, I would look at the code and I was like, “Oh, that’s really good.” And I would hit accept. And then the next time I’d say, “Oh, that’s even better, hit accept.” And I keep hitting accept so many times and my finger got sore and I said, “Hey, just do it. Yolo.” I just let it go. And so what I think I did with coding, right now I’ve literally 10 different bots in Cloudcode doing all sorts of work, and I’m not overseeing it really much at all.
I’m just letting it churn and letting it do its work because the coding that it does is better than any human coder that is that the AIs have beat worldwide coding contests. It beat all the humans. And so who am I with my stupid meatbrain and I’m a crappy coder? Who am I as a crappy coder with my meatbrain telling this AI that it’s not doing it right? So just like I kind of let it go and just say YOLO, just let it go, I wonder how soon we are going to do that societally for all of our things. Because if you think there’s a spectrum of what the agents get to do between one end of the spectrum, draft my email, the other end of the spectrum, here are the nuclear codes. So between this, it starts as draft an email. How soon do we just say YOLO, just let the AIs cook?
First we say draft the email, then we say build a product. Then we say run our business. Then we say run our economy, run my government, run my industry, and then hear the nuclear codes. And the question is how soon do we go on that, here are the nuclear codes? Is it next year? Probably not. Is it a thousand years from now? Probably. Or is it 100 years from now or is it 25 years from now or is it 10 ears from now? Now we’re just arguing about what the duration is going to be. And so as AIs, the IQ of AIs are 130, which is better than 98% of humanity. But what I said was actually untrue. It was 130 in January. It’s now gone to 137, which is better than 99% of humans. But actually what I said is not true because now it’s gone to 142 as of last week.
142 is better than 99.71% of humanity. And so really as AI keeps getting exponentially smarter, why would we trust our stupid meat brains for something that is 2X smarter than us or 3X smarter than us? And so the worry on the bad paths of ending humanity is that we will give it more and more agency to be able to do more and more things just like it did with cloud code. We as humanity are going to say, run our economy, run our government and hear the nuclear codes. And so I think we need to maybe avoid that if we want to avoid our doom.

Victor Li:
Well, you’ve talked quite a bit about this throughout the show, but if you’re a lawyer out there and you’re concerned about this and whatnot, what are some things that lawyers now can do to try to, like you said, push AI towards more positive legal outcomes?

Damien Riehl:
I would say if you’re a lawyer, talk to your congresspeople. We know a lot of congress people. We know a lot of senators. We know a lot of governors. We know a lot of people that have their fingers on the… We know a lot of chief judges. So one can imagine that we as lawyers stand up and say, “If we want to avoid doomsday scenarios, the AIs need to follow the law. And if the AIs need the law to be able to follow, then the law needs to be free.” So if you’re a lawyer listening to this, use your fingers on the levers of power with your chief justices, with your governors, with your congresspeople, and be able to say, “Hey, this is the way we get out of the doomsday scenarios, is that the AIs have the access to the law because if that happens, we can maybe avoid the worst of our futures.”

Victor Li:
Finally, to wrap up, if our listeners want to reach out to you to get your guidance on how they can act or maybe even whether they should start doomsday prepping or getting up, setting up panic room, what’s the best way to reach out to you?

Damien Riehl:
Clio.com is a way that first name, last name at Clio is my email address and LinkedIn. I’m pretternaturally on LinkedIn, so you can feel free to reach me out. LinkedIn, I usually get back to faster than my email, so LinkedIn is probably the best.

Victor Li:
So hopefully the next time I see you, you won’t be singing It’s the End of the World as we know it and I feel fine.

Damien Riehl:
Well, as a musician, actually I have a hootenanny every year in my backyard and we actually sang it’s the end of the world as we know it and I feel fine. And I say it hits harder today in 26 than it did in the 1980s, but I think that if we’re doing it right, we’re not going to be close to the end of the world. We’re probably in the age of abundance if we do it right. And it’s up to us listeners of this podcast, us lawyers, to make sure that we get pushed toward that abundant future.

Victor Li:
Well, thanks again for joining us today, Damien. I appreciate the conversation as always.

Damien Riehl:
Thank you so much for having me. I really enjoyed chatting with you.

Victor Li:
If you enjoyed this podcast and would like to hear more, please go to your favorite app and check out some other titles from Legal Talk Network. In the meantime, I’m Victor Lee and I’ll see you next time on the ABA Journal Legal Rebels Podcast.

Announcer:
If you’d like more information about today’s show, please visit legalrebels.com, legaltalknetwork.com, subscribe via iTunes and RSS. Find both the ABA Journal and Legal Talk Network on Twitter, Facebook, and LinkedIn, or download the free apps from ABA Journal and Legal Talk Network in Google Play and iTunes. The views expressed by the participants of this program are their own and do not represent the views of, nor are they endorsed by Legal Talk Network, its officers, directors, employees, agents, representatives, shareholders, and subsidiaries. None of the content should be considered legal advice. As always, consult a lawyer.