Senators Debut a New US AI Roadmap

Today on the AI Daily Brief, Chuck Schumer has released an AI roadmap and here's what people think about it. Before that in the headlines, Stability is apparently looking for a sale. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Join the conversation in discord with a link in the show Notes. Welcome back to the AI Daily Brief Headline edition. All the AI headline news you need in around five minutes. Stability AI has had a rough go of it for it feels like a year or more. There have been leadership challenges, culminating with founder and CEO iMadmostock leaving recently. And beyond all of that, Stability is just finding the difficulty in the mismatch between how much it costs to compete in AI and how much money there is to be made in AI. Currently, the Information reports that the company is now facing a cash crunch and because of it, discussing a sale, they write. The four year old startup raised at least 101 million from marquee investors and fetched a $1 billion valuation in 2022 on early fanfare for Stable Diffusion, an open source AI image generator it helped fund. The company also forged business ties with Amazon Web Services and other established tech companies, but making money from its free software proved difficult, especially as rivals such as OpenAI and Midjourney emerged. They continue the potential sale of Stability, which is based in London and has around 180 employees, which would add to a growing list of AI startups that quickly attracted large checks at steep valuations both before and after OpenAI's release of ChatGPT, which catalyzed an AI startup boom. Many of These startups, like OpenAI rival Cohere, have raised capital at high prices despite not generating much revenue. But the high costs of using AI models, as well as rising competition from Google Meta platforms and dozens of AI startups have pushed some of them to head for the exits. Now the big example they point to is of course Inflection, which recently saw most of its team and technology head to Microsoft in something that was an acquisition in all but named. The Information points out that while Inflection was able to use a $650 million licensing fee from Microsoft to give investors a modest return quote. In Stability's case, it's not clear how much potential buyers would be willing to pay in terms of what a deal might look like. They write, any deal would likely focus on absorbing Stability's AI researchers. Company executives have told potential buyers the company has between 70 and 90 researchers and engineers. However, that number is likely quickly dropping, so it's worth discussing a couple different dimensions of this Anytime there's a story such as this one, there's going to be a tendency to try to connect it to a larger trend, and in some cases that will be valid. So the obvious trend here is startups that raised a ton of money at high valuations that haven't been able to convert that to revenue. This is certainly a feature of the current AI startup landscape. It is particularly a feature when it comes to companies that were actually trying to create frontier or foundation models. In other words, the companies that were trying to compete in the most expensive part of the market. Reality has shown that the sheer cost to compete in that space is incredibly high, and the revenue dynamics are really difficult in part because of the unexpected relevance of big tech in this area. One concern I have with some of this type of reporting, which by the way, the information tends not to be guilty of, is trying to make it seem like somehow VCs were just in some stupid frenzy and not really considering the reality. In some cases, of course that's going to be true. There are always follow on investors who aren't really leading the package who are making fairly unconsidered decisions. But that's the case pretty much anytime. I think it's equally important to acknowledge that no one really knew exactly how this is going to play out. A couple years ago when this funding boom started, most people didn't think that the big tech players would be as relevant as they are because that's not the way that it's been in the past. When it comes to emerging technology areas, there are some very specific reasons that make AI different on this front. The obvious one is the cost of compute and access to compute, which we've talked about and other people have talked about extensively. But but another is one that seems obvious in retrospect but perhaps wasn't going in, which is that big enterprise buyers are currently so much more likely to want to work with a provider that they already have trust in when it comes to highly sensitive data that's being fed into AI systems. In other words, the fact that AI is best when it takes advantage of your company's data means that companies that might have in another area trusted or been willing to try out some new startup are more inclined to go with a partner they already know. Now, holding aside all of those trend lines, there's also just the specific challenges of decisions that this particular company has made. It's very easy to armchair quarterback in retrospect, and so by no means am I doing that here, but there is plenty of discussion to be had around strategic decisions around revenue and monetization, assessments of the future availability of capital, which are less about trend lines and more about specific decisions that specific people made. In any case, for all the people involved, I hope there is some good outcome. If nothing else, I'm quite sure that the market will have no problem absorbing this talent into a more productive area. Now, speaking of talent, Anthropic has a major announcement as Mike Krager, one of the founders of Instagram, has joined as their Chief Product Officer. He writes, Anthropic's research continues to be at the forefront of AI when paired with thoughtful product development, it has tons of potential to positively impact how people and companies get their work done. As a two time entrepreneur, I'm particularly excited about how Claude, along with the right scaffolding and product features, can empower more people to innovate at a faster pace and at a lower cost. So this is more to me than just the story of a big old hire for Anthropic. I think that this one actually does represent a trend that we're going to see more of, which is an increased focus on the productization of AI and packaging AI in consumer grade products that actually solve people's problems. Basically there has been so much focus, understandably on competing for the state of the art when it comes to the underlying models that there has perhaps been somewhat less focused on on how you make products that people actually want to use. I think we're going to see a lot more of this type of effort and Anthropic has definitely made a big move in that space. Lastly on the Headlines edition, Microsoft has apparently asked hundreds of China based AI staff to consider relocating outside of China. According to inside sources, about 700 to 800 staff engineers, predominantly of Chinese nationality, were recently offered the ability to transfer to the us, Ireland, Australia or New Zealand. Now apparently Microsoft told them that this was about helping bolster Microsoft's global ambitions in cloud computing and meeting its needs for AI engineers in a variety of places. And certainly there is some truth to that. Every week Microsoft announces another multi billion dollar investment in another country as it builds out its global technology infrastructure. But as the Wall Street Journal points out, there is also a geopolitical element of this. Microsoft, for example, was recently brought in to become a minority owner of G42 in the UAE in in a deal that seemed to have been facilitated by the Department of Commerce. On this theme of whether something is a trend or an individual story, boy howdy is this one of the most significant trends that we're seeing right now. And if you are interested in that trend and the geopolitical tension between China and the U.S. when it comes to AI, well, stick around for the main part of the episode. Today's podcast is brought to you by Plum. Are you a lean product team trying to rapidly develop and deploy AI features that deliver real value to your users? Plum empowers you to build complex AI pipelines, transform data and leverage validated JSON schema to create reliable, high quality AI features accessible as API endpoints, all in an intuitive low code interface. Go from idea to MVP in hours, not days. Get your AI powered product in front of customers as soon as possible with plum. Check out useplum.com that's plum with a B for early access to the future of AI app development. As a listener of this show, I have a strong feeling you like to stay up to date on all things artificial intelligence, including its impact on the workforce. Which is why I highly recommend checking out Managing the Future of Work, the chart topping business podcast from Harvard Business School. HBS professors Bill Kerr and Joe Fuller talk to business leaders, technologists and policymakers grappling with the forces like AI globalization and demographic shifts that are reshaping the nature of work. Recent guests include IBM Chro Nicole Lamoureau on how Big Blue is adopting AI, Morningstar CEO Kunal Kapoor on how AI can raise the investment iq, Microsoft Corporate Vice President Jared Spitero on how the tech giant is experimenting its way from AI assistance to autonomous agents and many other prominent movers in business and the workforce ecosystem. So don't miss out. Follow Managing the Future of Work on Apple Podcasts, Spotify or wherever you're listening now. Hello friends. Quick note before we get back to the show. I'm so excited to share that superintelligent is now live. Superintelligent is a platform for fast, fun and super practical useful AI learning. We have something like 300 video tutorials, adding 30 to 50 each week covering every topic in AI you can imagine, from LLMs to image generators to case studies, use cases. Basically everything that tells you how to use AI and what to use it on. In short, fast four to seven minute tutorial videos which are paired with step by step instructions that help you actually use these tools as well. It's $20 a month for unlimited access and I would love to see you there. Check it out at Be Super AI. That's BeSuper AI. Welcome back to the AI Daily Brief. When the EU passed its AI act, one of the biggest criticisms of it, particularly from those who were based in Europe or wish to be based in Europe, was that the EU seemed so intent on being the leader in AI regulation that they forgot what it would take to be a leader in AI, period. In other words, while they might be setting the tone for global regulation of artificial intelligence, they weren't setting up the European Union to be competitive in this increasingly important space. Meanwhile, over in the US we have over the last year seen a number of different legislative proposals around AI, many amounting mostly to an attempt by those introducing the legislation to put their stamp on the discourse to come. Meanwhile, Senator Chuck Schumer has been coordinating an overarching process which culminated this week in the release of an AI roadmap that's meant to provide a high level overview that can turn into lots of individual pieces of legislation. The banner headline of this, and the thing that stands out incredibly when you first read it, is that this is a roadmap that is intent on the US winning AI, not AI regulation. It's not that it ignores entirely many of the risk issues and the areas of concern that other pieces of legislation have wanted to focus on, but at core it is very concerned with American competitiveness. It feels pretty clear that that's the overarching consideration. Indeed, the Reuters headline really sums it up. The way the entire roadmap was summarized was US lawmakers seek 32 billion to keep American AI ahead of China Chuck Schumer was tweeting up and down X yesterday about this. In one tweet, he wrote, we faced a choice watch as AI reshape the world or make a novel effort to promote innovation and put in place guardrails before it was too late. So we formed a bipartisan working group, kicked off AI Insight Forums, and now released our AI Policy Roadmap. In another tweet, he writes, the US must pursue and perfect transformational and sustainable innovation to do so well, our AI policy roadmap recommends a $32 billion surge in emergency funding and promotes innovation, safety and accountability to cement America's dominance in AI. Schumer hammered these points in comments to the press on Wednesday. For example, at one point he said, if China is going to invest 50 billion and we're going to invest nothing, they'll inevitably get ahead of us. So that's why these investments are so important. Republican Senator Mike Rounds, who worked on this, said, this is a time in which the dollars related to this particular investment will pay dividends to the taxpayers of this country long term. China now spends probably 10 times more than we do on AI development. They are in a hurry. Before we get into all the commentary, let's look briefly at the report and roadmap, which was called Driving US Innovation in Artificial Intelligence A Roadmap for AI Policy in the United States Senate. The very first substantive part of this platform is titled Supporting US Innovation in AI. They write, the AI Working Group encourages the Executive branch and the Senate Appropriations Committee to continue assessing how to handle ongoing needs for federal investments in AI during the regular order, budget and appropriations process, with the goal of reaching as soon as possible the spending level proposed by the National Security Commission on Artificial Intelligence in their final report, at least 32 billion per year for non defense AI spending. From there they get into a bunch of specifics, but that's what they chose to lead with, and that's why press is leading with this as well. It's clear that they want to fund everything funding for AI testing and evaluation, infrastructure funding, a series of AI Grand Challenge programs, funding R and D activities, et cetera, et cetera, et cetera. The next section is about AI in the workforce, and once again this section acknowledges that while AI could be extraordinarily disruptive in the short, short term, there are ways to make it work for a broad section of society. For example, they write, the AI Working Group encourages development of legislation related to training, retraining and upskilling the private sector workforce to successfully participate in an AI enabled economy. Such legislation might include incentives for businesses to develop strategies that integrate new technologies and reskilled employees into the workplace, and incentives for both blue and white collar employees to obtain retraining from community colleges and universities. It's not until we get to the third section, High Impact Uses of AI, that we start to get into any risks. And even here they're very light on details. Indeed, it seems pretty clear that this group is much more convinced of risks that are here and now versus the more farther out AI X risk type scenarios and thus focuses their recommendations on those areas. It's things like dealing with child porn issues that come up from AI. Another category that they're clearly focused on from a risk standpoint is election integrity. In fact, this was the one risk side thing that Schumer decided to talk about yesterday, writing as we launch our AI Policy roadmap. I'm proud to support the Senate Rules legislation to protect our elections. The 2024 elections will be the first ever held in the age of AI. We know AI has the potential to discredit our system of elections and we must act now. In other comments he said, if deepfakes are everywhere and no one believes the results of the elections, woe is our democracy. This is so damn serious. There is also a large discussion in the roadmap of the AI implications for national security, but I want to move into the reactions and the response Fast Company wrote a piece called the New AI Roadmap Gets Panned by Watchdogs. It featured quotes from Professor Alondra Nelson, who called it striking for its lack of vision, and a Brown professor who said he felt like he was part of a dancing monkey show and were the Monkees. Engadget published a piece called Chuck Schumer Is Dropping the Ball on Regulating AI and I'm not totally sure when Engadget decided to be against Big Tech, but the piece is an aggressive condemnation that goes so far as to speculate that the fact that one of Chuck Schumer's daughters works for Amazon and the other has worked for Meta might be coloring his opinions. Now, I don't usually get into this level of commentary, but I would suggest if you are trying to make a legitimate critique, saying that the place that someone's kids work is the reason that someone would make policy is a pretty lightweight argument. But they are far from alone in making this type of argument. The Open Markets Institute and many others feel that Big Tech is not held accountable with this roadmap. The Open Markets Institute says, we call on Congress to ignore this AI roadmap and most of its recommendations. Senate leadership needs to use anti monopoly law to restructure dominant tech corporations and regulate their behavior. Their roadmap doesn't do that, the AI Now Institute writes. Senator Schumer's follow up to his insight forums landed proposing a $32 billion investment in the sector in a roadmap to regulation. The process has been but a stalling tactic with industry in the driver's seat. Interestingly, and perhaps politically, SAG AFTRA called the roadmap a quote, important step forward in ensuring the responsible development and deployment of artificial intelligence. But what about from the other side? What are the folks who are more on the accelerationist and pro innovation side of the conversation? Brian Chao, who's the director of the alliance for the Future, which was created specifically as a counterweight to the AI safety, folks, writes, the overwhelming emphasis on research grants and the benefits of AI is very encouraging. It shows that the committee understands AI is not just chatbots making explicit references to uses in science, medicine and agriculture. This represents a broader vibe shift in DC away from pessimism and towards strategic optimism. Speaking of which, the National Security section is encouraging speeding up hiring, addressing energy demands, defining AGI and space debris. I'll take it. In general, the language about AGI is encouraging. The report has very little language about specific doomsday threats, indicating that the committee is unconvinced by doomers. Now Brian concludes by pointing to one section that he said reads like there's a doomer funded staff member that needs to be placated, but also that the leadership is really not convinced. Adam Thierer, an innovation policy analyst at rsi, writes the positives Overall, the document is big on aspirational statements, but short on finer details. That's fine, actually. The key here is that there is no call for a broad based new AI regulatory agency or general purpose licensing schemes and no existential risk lunacy in here. It's actually got a nicely balanced section on AI risks and an insistence that committees ensure these requirements align with any potential risk regime and do not inhibit innovation. And no strike against open source AI, which is crucial. So what to make of all this? It is important to remember that this is basically a framework of a framework. It's meant to provide high level guidance, but the real policy will be made in specific committees that have different jurisdictions. That means that if you are someone who finds something frustrating about it, there is still plenty of room to get your opinion in and shape what actually comes out. It certainly makes clear that at least for this bipartisan group of senators, the threat of Chinese leadership in the AI space is a much bigger threat than the risk of runaway AI. One thing that isn't really here is the balance between closed source and open source, and I continue to think those conversations are certainly legitimate areas for people to push. Overall, it definitely shows that pro innovation competitiveness is headlining the agenda when it comes to USAI policy right now, and that's the place we're going to be starting from now. There will be much more to dig into on this, so keep an eye out for that. But for now that is going to do it for today's AI Daily Brief. Until next time, peace.