The AI Daily Brief: Artificial Intelligence News and Analysis - The Top 5 AI Stories This Week

Episode Date: June 14, 2024

Counting down the top 5 AI stories of the week, from AI candidates running for political office to Apple’s major AI strategy announcement. This episode of the AI Daily Brief covers key developments, including OpenAI’s financial growth, the launch of Luma Labs’ Dream Machine, the ARC Prize competition, and more. Stay updated with the latest in AI technology and how it’s shaping the world. ** Join Superintelligent at https://besuper.ai/ -- Practical, useful, hands on AI education through tutorials and step-by-step how-tos. Use code podcast for 50% off your first month! ** ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://aidailybrief.beehiiv.com/ Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@AIDailyBrief Join the community: bit.ly/aibreakdown

Today on the AI Daily Brief, we're counting down the five most important stories in AI this week. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes. Hello, friends, doing something a little bit different today. I've done this before, but I'm coming back to it. We're doing a countdown episode, sort of a weekly recap, weekly summary. A lot of you liked the last one I did a few weeks ago. I'm not sure yet if this will become a regular thing or if I'll do it just every once in a while,

but I thought it would be a fun change of pace for today. So let me know how you like it, and let's dive in. Welcome back to the AI Daily Brief. Today, as I said, we are counting down the top stories in AI this week, and we're actually going to start off with a product launch and a couple of honorable mentions. The product launch of the week, I have to give to Luma Lab's new Dream Machine video generation.

There was so much hype all over Twitter for this thing. Indeed, if you had just looked at the conversation on X, you would have thought that Luma dropped something that blew SORA out of the water. Now, when Push came to shove and people actually dug in, the quality was not SORA level, at least the SORA examples we've seen. But while it may not have been as visually advanced as that forthcoming, sometime model from OpenAI, it had something very important going for it, which is that it was available.

One of the use cases that emerged that really showed off Dream Machine's capacity was animating different memes. Even Andre Carpathie, one of the founding members of OpenAI, said, wow, the new model from Luma Labs extending images into videos is really something else. I understood intuitively that this would become possible very soon, but it's still something else to see it and think through future iterations of it. The story of the Luma launch, for anyone paying attention out there, is to, one, have a product that people can actually use, and two, lean all the way into the power of social media as a distribution channel to get people excited about their creations.

Next up, we move over to an honorable mention. A super cool thing that was announced this week is called the Ark Prize. It's a $1 million competition to beat the ARC AGI benchmark and open source the solution. It's hosted by a co-founder of Zapier and a very prominent deep learning expert from Google and has a bunch of other great people like Greg Kamrat who's been on this show before as part of their team. Let's watch the quick intro video. Progress towards AGI has stalled. LLMs are trained on an unimaginably vast amount of data.

yet remain unable to adapt to simple problems they haven't been trained on. Frontier AI research has become closed source. We need new ideas to move forward. ArkPrize, a million dollar competition to open source AGI progress and beat the only AI benchmark that measures general intelligence. Visit arcprise.org. So on ArcPrize.org, you can see more about each of the different parts that you just heard in that announcement. They point out that skill is different than intelligence, that general intelligence requires the ability to acquire new skills,

that there is only one formal benchmark for AGI, which is the abstraction and reasoning corpus for artificial general intelligence, and that an open source solution is key. It is a super cool project. Again, the URL is arcprice.org. Go check it out. Get involved. We'll certainly be keeping track of it here on the show. second honorable mention is a story that just launched on Bloomberg, where one of the founders of HuggingFace told a reporter from Bloomberg that he's hearing from about 10 AI startups a week that want to sell themselves to HuggingFace. So why is this notable? I've seen a lot more discussion recently around whether we're moving into a consolidation period where the glut of

AI companies is starting to whittle down, where it's getting a little bit harder for AI companies to raise money, and this would certainly be some indication or evidence of that. Now, this is deserving of an entire conversation, and one that I'll probably have at some point soon, but I do not in any way believe that a consolidation period in startups is anything other than extremely healthy. The sheer tonnage of entrepreneurial excitement around AI has led to just an incredible amount of companies being formed. The surface area of new opportunities is just so dissimilar to things that we've had over the last 10 or 15 years. It's also been the one area of venture capital that has defied the odds and a shift away from ZERP area

investment approaches to operate in the same hypey venture capital way. A recent crunch-based survey found 40% of recent VC deals being in the AI space. Inevitably, with those dynamics, you're going to have more companies formed than can possibly win. And to the extent that you believe that the market should be reallocating talent and energy towards companies that are differentiating and starting to win, consolidation is a good thing, not some scary thing that says that the bubble has popped. But now we move into our top five stories on the week. And the first is kind of a combination of some new entrance to the AI conversation, or at least entrance that are getting louder in a specific context.

The first is Pope Francis, who will be joining the G7 specifically to talk about artificial intelligence. This is the first time the G7 will have a Pope as an invited participant, and interestingly he's choosing to take his time to focus on why the world needs a coordinated AI strategy. Pope Francis has been talking about this for a couple of years now, and of course was also famously deep faked with a puffy jacket last year. And basically his schick is that AI needs to be in the service of people, in the service of the good. Now, what that means practically is, of course, the devil in the details, but it is certainly notable that he's getting more involved in the

conversation. Another notable person who's waiting into the AI conversation is former president and current presidential candidate Donald Trump starting to discuss AI as well. The specific context for the conversation was a podcast he did with Logan Paul. PhD student James Campbell summed it up like this. Trump on AI. Finds it alarming and scary. See how it could lead to accidental nuclear annihilation. Says if it's going to happen, it's going to happen and we can't stop it. Says we must be China. No's training runs or electricity bottlenecked and will fix that as president. Received $12 million from unspecified San Francisco supergeniuses. Used to chat GPT to edit a speech

and joked about firing his speechwriters. Referred to superintelligence as super duper AI, realizes this is going to be insanely powerful and acknowledged risk of taking over the human race. Another commenter pointed out that it was an AI in general that he seemed to indicate was scary, but specific applications of it like deepfakes. Love him or loathe him, the fact that Trump is now talking about AI means that its profile in the public discourse is about to go up. And that's why this has a place on our list. Today's episode is brought to you by Super Intelligent. Regular listeners know that Super is our platform for helping people learn how to actually use AI tools.

These are not long, laborious courses. these are fun, fast tutorials that get you actually using the world's most interesting and useful AI tools within minutes. If you want to build a web application with no code, we've got tutorials for that. If you want your presentations to look better than ever and take you less time than ever, we've got tutorials for that. If you want help brainstorming, writing social media copy, and just generally working smarter, faster, and better, we've got tutorials for that. We've worked really hard to make it so that there is no better place on the internet to learn how to actually put AI to work for you, and I'd love for you to check it out. Go to besuper.a.i and use code podcast for 50% off your first month. Once again, that's besupor.a.i.

Number four story this week in AI is around geopolitics. The U.S. is once again considering adding even more limits on China's access to AI chips. According to Bloomberg, the Biden administration is considering limiting China's ability to use a new architecture called Gate All around or GAA, the promise of which is to make semiconductors more powerful. No one official commented on this story, but according to Bloomberg's sources, a potential draft GAA rule is not yet finalized, and it's quote unclear whether the ban would restrict China's ability to develop its own GAA chips or go further and seek to block overseas companies, particularly U.S. chipmakers, from selling their products to Chinese electronics manufacturers. There are some credible reasons for the U.S. to

continue to be concerned about this. The information recently did an expose called China's Nvidia loophole, how bite dance got the best AI chips despite U.S. restrictions. The loophole is pretty simple. As the information puts it, the U.S. government forbids NVIDIA from selling some of its most advanced AI chips to customers in China, but it doesn't stop Chinese firms from buying or renting NVIDIA's chips if they're used within the U.S. So BytDance, the information points out, which is, of course, the owner of TikTok, has been renting NVIDIA's best chips from Oracle. This is, of course, strongly outside of the intent of those rules,

and so it's not surprising that the U.S. is trying to close these types of loopholes. What's more, one of the areas of loopholes that the U.S. has been trying to close are Chinese access to AI chips via proxies in the Middle East. A UAE minister says that U.S. fears about that are actually quite valid. Omar al-A.A.E. Minister of AI and Digital Economy told Bloomberg, quote, I think concerns about chips coming to the Middle East and going to China are valid concerns for any country that has adversaries. He added, we think the UAE has proven to be a strategic partner with the U.S. Sometimes you are a victim of the neighborhood you're in. The other small geopolitic story is some perhaps surprising comments from Canadian Prime Minister

Justin Trudeau, where he came out certainly more accelerationist than AI safetyist. He basically said that we shouldn't be overly concerned about dystopian sci-fi type scenarios and that the approach to AI concerns of slowing it down is unlikely to work, and instead we should be focused on building AI for good. It strikes me as part of a somewhat broader shift we're seeing in mainstream political discourse around AI, but it's still too early to know how deep that shift is. Number three most important story this week, OpenAI is doing well financially. Or at least, it's continuing to grow.

Last summer, reports were that OpenAI's revenue run rate was around a billion dollars annually. By the end of the year, it was between $1.6 and $2 billion. And now over the last six months, it seems to have doubled to $3.4 billion. Interestingly, if this information is correct, and it comes from Sam Altman talking to staff, then OpenAI seems to be doing a lot better than Microsoft at selling OpenAI software. Only 200 million or so of that 3.4 billion came from their partnership with Microsoft, and Altman said that that 200 million represented around 20% of Microsoft's revenue from OpenAI software, which would of course put Microsoft's revenue from OpenAI software at a billion dollars as compared

to this 3.4 billion for OpenAI itself. I'm not exactly sure what to make of that yet, but it is notable to me. Another OpenAI story, which is just broken and I haven't had a chance to dig into too deeply yet, is that they've appointed a new board member, former NSW. head Paul Nakasone. He's a retired U.S. Army General and former head of the National Security Agency, and boy are many people not happy about this. Edward Snowden writes, they've gone full mask off. Do not ever trust OpenAI or its products, chat GPT, etc. There is only one reason for appointing an NSA director to your board. This is a willful, calculated betrayal of the rights of every person on earth. You have been warned. Mario Nauffel summed up the news, saying that Nakasone's appointment will help improve

AI's role in cybersecurity by detecting and responding to threats quickly, and that his appointment follows concerns over safety culture at OpenAI. But the top comment on that comes from Elon Musk, who said, can't wait for OpenAI to have access to my phone. Dot, dot, dot. It's back to hear more on this controversial appointment in the weeks to come. Bundling in with this story still in this number three slot is that Mistral, another frontier model competitor, has increased its war chest as well, adding over $600 million at evaluation above $6 billion. This one-year-old company is, of course, a slightly different path and is very clearly identifying OpenAI as the challenger that it's trying to beat. One small note that you might have seen, some sources are reporting this as close to

$6 billion. The Financial Times, for example, says the valuation was $5.8, while other companies are reporting it as $6.4 billion. This is just a choice on whether you're reporting pre-money or post-money. In other words, mistral was valued at $5.8 billion, but then added $645 million, in cash, which means that it is now worth $6.45 billion because that cash has a one-to-one value of a dollar for each dollar raised. Conservative European papers tend to focus on the pre-money number, which is lower, while American news outlets are, of course, talking about the biggest number possible. Number two story is the AI candidates for public office.

There are not one but two of these stories floating around the news this week. In Wyoming, an AI avatar named Vic or Virtual Integrated Citizen is running for mayor via as someone named Victor Miller, who says that they will use the AI to make all the decisions if it is elected. Miller said, I realize that this entity is way smarter than me and more importantly, way better than some of the outward-facing public servants I see. Rights wired, Miller added that he'll serve as Vic's meat puppet to sign documents and attend meetings and other in-person events. Over in the UK, an AI candidate is also running for Parliament. The candidate is named AI Steve and is running for Parliament in Brighton. Very similarly, it's represented by a businessman

name Steve Endicott, and it's a whole approach to try to get more direct representation in government. On the one hand, it would be easy to write these things off as novelties, and I'm sure that's how many people will initially treat them. However, I do think there is a trend of disenfranchisement with the existing political process in such a way that it could open up possibilities that people would rather have some AI avatar who can take in a bunch of information and make decisions based on that data rather than personal preference as simply a better approach to leading. We'll see, of course, if any traction these candidates get. But for now, I think it would be a mistake to just write it off as fun clickbait.

Lastly today, our number one story with a bullet this week is, of course, the launch of Apple intelligence, as Apple puts it, AI for the rest of us. After a huge number of questions and a significant amount of waiting, we finally are starting to get the idea of what Apple's AI strategy is going to be. And what clearly stands out is that this company believes that the way that AI will enter most people's lives is through very simple day-to-day sort of uses. It will be simpler, better ways of doing the same things we're already doing, not revolutionary, professional, or creative use cases. It's a fascinating stance, one that makes a ton of sense for Apple, and so far has been very

validated in the market, which temporarily made Apple the most valuable company in the world again after the announcement. Now, the proof will, of course, be in the pudding, and we won't really be able to get our hands on it until the fall with iOS 18, but for now, Apple has taken a strong stance on AI and is officially in the game. That is going to do it for this countdown edition of the AI Daily Brief. Let me know if you like this type of episode and maybe we'll do it more. And until next time, peace.