Should OpenAI Have Been A Government Program?

0:00Today on the AI Breakdown, we're reading pieces that bring up questions about AI and American civic life. The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. Go to breakdown.network for more information about our Discord, our newsletter, and our YouTube channel.

0:24Hello, friends. Welcome back to another long-reads version of the AI Breakdown. This weekend, we're going to draw from two different pieces from the same author. The author is Josh Tiringel, and he is the new AI columnist at the Washington Post. Now, one of the things that is important to me about these long reads is that they don't just represent the perspectives that I hold. I would go a step farther in this case and say that in many ways, I am not the biggest fan of these two pieces. I think that as he writes more, Josh and I will find lots to disagree on. However, why I think it's relevant and useful is that one, the Washington Post is a relevant publication for Beltway Insiders in D.C., and a lot of the discussions about what happens next in artificial intelligence are going to be discussions held by D.C.

1:14Beltway Insiders. So to the extent that these opinions either shape or reflect what's going on in that little hamlet of this country, they're useful to at least understand and know about. The second reason is one we'll get into more in the discussion portion of the show regarding a larger question of American civic life, its decay, and why that might matter in the context of AI. But let's start with some excerpts of the first piece. It was the piece by which Josh introduced himself to the audience, and it was called, You Hate AI for All the Right Reasons. Now, reconsider. Josh begins, let's start with the end.

1:49If you know anything about the state of artificial intelligence, it's that many of the people advancing the technology are gravely concerned about the technology they're advancing. Two statements stand out. The first was a petition following the March release of OpenAI's GPT-4 calling for a six-month pause on any AI system exceeding GPT's capabilities. The signatories asked, should we develop non-human minds that might eventually outnumber, outsmart, obsolete, and replace us? Should we risk loss of control of our civilization? The second statement, issued in May, was an escalation of both stakes and prestige, a Met Gala of Doom, signed by nearly all the major AI company CEOs and most of the top AI research scientists, this statement was just 22 words long, which really helped the E-word pop.

2:30Quote, Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war. So who's excited for AI? I am? Maybe? Now from there, Josh introduces his new role, that he's going to become the post-regular AI columnist. He says that as he's spent time looking into this issue, he's on the one hand become increasingly confident that we're not all about to die, but at the same time increasingly less confident about what the world of the future actually looks like. He writes, It's possible we're at the dawn of an incredible era of toolmaking, with the emergence of ChatGPT marking a Kubrickian cut between our bone-throwing present and an awesome future.

3:08He points to the fact that AI tools are already predicting the spread of disease, helping detect guns in school, reducing energy consumption, and that there is more to come. On the other hand, he writes, Quote, other equally brilliant people go straight to Mad Max Fury Road. Their scenarios cover everything from an AI-manufactured extinction-level virus to societal decay as jobs disappear, inequality becomes permanent, authoritarian states tighten their grip, and meaning is drained from our existence. Before any of that arrives, we've got great leaps forward in AI-generated porn, fraud, and misinformation to look forward to.

3:37Now, for Josh's taste, both of these sides of the conversation are perhaps a little bit too convinced of their own position. He points out that that might be because there is so much money at stake. And he writes, quote, This is not the Manhattan Project or the space race when the big brains wore government badges. That note I think will become important in our discussion a little bit later. Now in that, Josh also points to the challenge of understanding the true motivations for what different people in the sector say and do. Paraphrasing, he asks, is Zuckerberg's open sourcing of LLAMA about an optimistic democratization of AI or just about a different strategy to catch up with OpenAI?

4:11And when Sam Altman from OpenAI signed that pledge, that 22-word extinction fear pledge, how much of that is based on a goal for regulatory capture? And given what many people have lived through in terms of debates around climate change and COVID, not to mention the hype around crypto, which he calls a tech bro hustle, he wouldn't blame you for just wanting to sit out the discussion of artificial intelligence. But at the same time, Josh argues that that wouldn't be the right idea. Because in his estimation, things really have changed. The era of artificial intelligence we're living through is no longer just about personalizing social media and song recommendations and enhancing photos, but represents something fundamentally new.

4:49Discussing neural networks, he writes, Imagine if your brain got 10 times smarter every year over the past decade, and you were on pace for more 10x compounding increases in intelligence over at least the next five. Throw in precise recall of everything you've ever learned and the ability to synthesize all those materials instantly in any language. you wouldn't just be the smartest person to have ever lived, you'd be all the smartest people to have ever lived. That's a plausible trajectory for the largest AI models. This is why he says AI has gone from something that's interesting and a novelty and a far-off future potential to something that is just smashing through, quote, many of the supposed barriers between human and machine capabilities.

5:24Now, one thing that he writes, which is extremely well put, is that Silicon Valley has a tendency to, quote, conflate inevitable with instantaneous. Rather than ChatGPT being an iPhone moment, he believes it's more like a Netflix moment, as he writes the tech is moving fast, but its impact will arrive in waves. We've already seen slight dips in ChatGPT usage. That hardly means chatbots are doomed or that AI is a fad, only that it's early. It will take time to understand how people adapt to these new tools and vice versa. That time, Josh argues, is an opportunity. It gives us an opportunity to discuss, to debate, and plan for everything from basic privacy and IP regulations, to upside-down labor markets, and even those big, scary scenarios.

6:03His recommendation? Quote, At an individual level, maybe just turn down the volume for a bit. Okay? The story so far has been told by geniuses and scoundrels with mixed motives and terrible emotional intelligence. It's really no surprise that they're also lousy storytellers. Who starts with extinction? Let's begin again, this time with creation. All of the software we've ever used was engineered to work backward from an outcome. Its creators wanted to help you find a webpage, or play a game, or operate a laptop. Perhaps you've noticed that the major AI chatbots arrived with almost no user documentation or instructions.

6:33A lump of clay doesn't come with instructions either. That's what makes this moment unique and so worthy of species-level number one foam finger pride. We humans have created a tool for potentially infinite tasks. Its imperfections are ours to solve, and its powers still ours to shape. Alright, so if you've listened to me a lot, you'll probably think that there's a lot in there that sounds pretty much like arguments I've made. So where do my particular issues with this piece come in? I think they come down to, one, where our different starting points are in the tech versus media battle that has played out over the last few years.

7:02I am quite clearly coming from a tech perspective, whereas he is coming from the media perspective that basically mistrust those with the tech perspective a priori. And two, I think I was quite turned off by how the title of the piece changed over time. Now, editors often do this, and it's not Josh's fault. He has no control, really, over the title. But the title the Post originally ran with is, AI is powerful but imperfect, and ours to shape into something good. A great message that reflects exactly what Josh wrote. The title it ended up with was, You Hate AI for All the Right Reasons, Now Reconsider.

7:34A title that not only plays into people's base-level fears of technology, but also comes off as super cocky. Again, this isn't Josh's fault. It's something that I should take up with the Washington Post editors, but it just left a bad taste right going in. However, with that, let's move to his second piece. I think it's particularly relevant in the context of a week in which a huge portion of the big tech CEOs that are most relevant in the artificial intelligence space descended upon DC to be part of Chuck Schumer's closed-door meeting on Wednesday. This piece came out on Tuesday and was titled, OpenAI's Sam Altman Wants the Government to Interfere.

8:07Now, the really interesting thing about this is the quote of Sam Altman's that it opens up with. OpenAI should have been a government project, right? The quote came from a conversation with Altman in July at OpenAI's headquarters. Josh writes,

8:50According to the piece, Josh continues, Altman says that in 2017, he pitched officials at the White House, the Defense Department, and the Energy Department on investing in the company behind ChatGPT. Altman said, We said we need more funding than we can raise as a non-profit. Would you like to give us money? This is something important for the U.S., and it just died in the bureaucracy. Now, Josh isn't as sure that that's a bad thing. He writes, It's fair to debate whether ChatGPT would be helping you with your pitch decks right now had the government gotten involved. Private sector competition tends to speed things along, and the federal bureaucracy has been outsourcing most of its tech building for years.

9:23But part of the reason Altman has become so popular in Washington is that, contrary to most people in Silicon Valley, he doesn't think of the government as feckless. Altman says, I mean, a lot of the people on the Hill are surprisingly thoughtful about AI. Most probably aren't, but you know, the fact that there's like 15 or 20 good ones is a nice situation. Now, much of the rest of the piece is spent on how difficult it seems to get alignment and meaningful progress on AI policy issues, given Congress's woeful engagement with technology in general. What's more, there's very little agreement even among the tech folks about what should be done.

9:52As Josh points out, Altman and Microsoft support the creation of a single oversight agency, while IBM and Google don't. Musk has called for a six-month stoppage on sophisticated AI development. Everyone else thinks Musk, an OpenAI co-founder who fell out with Altman and announced the creation of a rival company in March, is insincere. And yet, Josh says, The main reason for hope is humility, and it's not nothing. For the first time in memory, Silicon Valley and Washington are recognizing their dysfunctional relationship and its potential consequences. On the Senate floor last week, Schumer told his colleagues, Legislating on AI is certainly not going to be easy.

10:23In fact, it will be one of the most difficult things we've ever undertaken. But we cannot behave like ostriches sticking our head in the sand when it comes to AI. Altman, meanwhile, is just a boy standing in front of his government asking it to regulate. Altman says, People say, oh man, this AI thing is coming and the government is never going to be able to get it right. So don't go asking for regulation. But I think we just need to hold the government and ourselves to a higher standard. I agree with people that the government has become ineffective on the whole. But I think the answer should be to try to get the government back to an effective place.

10:51So what I think is really interesting is that last line, and really that last paragraph in general. The idea that this is a moment to try to get the government back to an effective place. One of the things that I have frequently felt over the last couple years, watching the government deal with or not deal with crypto, is that if it can't even get it together to figure out how to reasonably regulate this thing, How the hell is it ever going to be able to handle such a much, much more complicated issue in artificial intelligence? One of the things that makes me concerned about the future that has AI deeply enmeshed in it is that I believe that even without questions of extinction, the very presence of artificial intelligence and what it does in terms of how jobs work, how industries develop, is going to bring up major questions that cause us to rethink, or at least need to rethink the fundamental and underlying social contract that dictates the relationship between people, government, employers, work.

11:47And we've never been in a worse position to have that conversation. The cratering of institutional trust that has happened over decades for so many different reasons leaves us in a place that is brittle and weak when it comes to civic life, exactly at a time when figuring out how to compromise, look past priors, think about things from the ground up, try to make good decisions based on new information. You know, all that stuff that might once have been the provenance of politics, but is no longer, is so, so sorely needed. So I find myself agreeing with Sam that the answer should be, in his words, to try to get the government back to an effective place.

12:20I just have no idea how to do it. But we're not going to solve that in a podcast, so for now, we will have to leave it at just a conversation to start. And so on that note, I hope you are having a great weekend, and until next time, peace.

12:37Thank you.