The AI Daily Brief: Artificial Intelligence News and Analysis - AI Startups Are Having More Trouble Fundraising. Here's What It Means.

Episode Date: December 12, 2023

AI startups have run counter to the trend of decreased VC all year, but even they're now feeling the pinch as well. Here's why it matters more broadly than just for AI startup founders. Before that on the Brief, tension around the EU AI Act, and a CEO gets fired over an AI kerfuffle. Interested in the January AI Education Beta program? Learn more and sign up for the waitlist here - https://bit.ly/aibeta 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://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI breakdown, startup fundraising for AI companies isn't as easy as it once was. Before that on the brief, tension around Europe's AI Act, America's first AI robocaller, and a CEO of a major publisher gets fired after an AI experiment gone wrong. 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. Welcome back to the AI Breakdown Brief. All the AI headline news you need in our around five minutes. We're kicking off today with the story of an AI wrought firing and job disruption, although not the one that had been prophesized. So one of the things that doesn't have a great

track record so far is AI being integrated into publishers' workflows. We've seen a lot of experiments over the course of 2023, where publications start experimenting with AI writing articles, and it just seems like every headline about it relates to how poorly it's gone for one reason or another. Sometimes the articles are wrong and have misinformation. In fact, that's kind of the most common problem. But then this Sports Illustrated thing took it to a whole different level. Basically what happened is that in November, Futuresim reported that Sports Illustrated, which is of course an absolute institution in the sports publishing world, had been publishing articles that were written by writers who didn't actually exist in the world. The authors had AI generated profile pictures,

And as you might imagine, the magazine's actual staff were none too happy about this. Now, part of what seems so weird to me about this is that Sports Illustrated was claiming that the articles hadn't been written by AI and that the fake names and AI generated photos were just meant to protect the privacy of the writers or something. It was all very weird. Now, subsequent to that, there has been a lot of shake-up at the company. Sports Illustrated is owned by a publisher called The Arena Group, and the Arena Group has now fired its CEO, Ross Levinson.

As Futurson points out, they were not discreet about this news, publishing a press release that was sent out to business wires as a for example. Arena Group majority investor and founder of Five Hour Energy, Menosj Barga, will now be taking over temporary control as interim CEO, and apparently he is trying to shake things up. Apparently, in a meeting last week following the termination of a separate set of executives, he was lecturing the staff to, quote, stop doing dumb stuff, and said, quote,

the amount of useless stuff you guys do is staggering. Now, it is worth noting that although they are being loud about Levinson's departure, they didn't say that it was specifically having to do with this whole AI scandal. The party line remains that the group that was at fault was actually a contractor called Ad Von Commerce, and that it was them that had posted those particular articles. In any case, it's another great reminder of just how tense and weird the transition to an AI integrated world is going to be, especially in the world of publishing, and I'm sure it's not the last time that we'll see a story just like this one.

Next up, we move over to the world of policy. Of course, as I mentioned yesterday, the EU-AI Act has advanced past its next level of political agreement, and now the technical experts get together to start to hash out exactly what the boundaries of the law will be and how it's supposed to be implemented, which of course brings with it a whole new dimension and opportunity for challenge and disagreement. Speaking of which, not everyone is happy about the deal that has been struck. The Financial Times reports that French President Emmanuel Macron is unhappy about the state of the act, and is arguing that it's going to cripple European tech companies as they try to compete with rivals from the U.S., the UK, and China. On Monday, Macron said,

we can decide to regulate much faster and much stronger than our major competitors, but we will regulate things that we no longer produce or invent. This is never a good idea. Now, Macron's specific frustration is around the rules for foundation models, and it seems that he specifically has Mistrel in mind. Mistral, of course, is the Paris-based startup that just raised money at a $2 billion valuation, and which has stolen a lot of the momentum in the open-source, space from the biggest player in that area, which is, of course, meta.

Writes the F.T. Macron's comments may presage a new battle over the final terms of the AI Act, which still needs to be ratified by member states over the coming weeks. France, alongside Germany and Italy, are in early discussions about seeking alterations or preventing the law from being passed. Said a person with knowledge of the talks, the stakes are high and the French will try to block this. Now, again, at issue is the fact that at the end of this process, which has been ongoing for

years, the AI Act tried to fold in these foundation models, which is good. caused a lot of the chagrin. Said Cecilia Bonefield-Dahl, Director General for Digital Europe, the last-minute attempt to regulate foundation models has turned this on its head. The new requirements on top of other sweeping new laws like the Data Act will take a lot of resources for companies to comply with, resources that will be spent on lawyers instead of hiring AI engineers. Meanwhile, over in the U.S., as election season begins to heat up heading into 2024's presidential election, one campaign in Pennsylvania has deployed what they're calling the first AI-powered robocallor. Now, robocallers are, of course, pieces of software that call vast lists of donors

and try to remind them to vote or give them information about specific candidates. But this new AI-powered version being deployed by Democrat Shemaine Daniels is a fish of a different color indeed. The AI campaign volunteer called Ashley, which was built by a London-based company called Syvox, actually analyzes data about the voter that it's calling, and then uses that to hold a conversation where the responses are neither can nor pre-recorded. writes Reuters. Over the weekend, Ashley called thousands of Pennsylvania voters on behalf of Daniels. Like a season campaign volunteer, Ashley analyzes voter profiles to tailor conversations around their key issues. Unlike a human, Ashley always shows up for the job, has a perfect recall of all of Daniels's positions, and does not feel dejected when she's hung up on.

Said Ilya Musikansky, the CEO of Cy Vox, this is going to scale fast. We intend to be making tens of thousands of calls a day by the end of the year and into six digits pretty soon. This is coming for the 2024 election, and it's coming in a very big way. Now, one interesting note, given our later discussion of the changing nature of startup funding in this space, the company behind Ashley decided not to take venture capital because they don't want to have any tension around how fast they decide to grow and any ethical considerations that come up as they deploy this technology in a very challenging area. The company has also set up a committee that's empowered to force the CEO to publicly disclose anything of concern about the company, and they decided to give Ashley a robotic-sounding

voice and disclose she's in AI, rather than try to trick people into thinking that she is a human. Now, of course, it's possible that this type of use of AI won't actually be allowed by the time the election cycle is finished, but as for now, we are in weird gray area, and again, there's going to be a lot more like Ashley coming soon. Moving over to markets for just a moment, in another sign of investor excitement around AI, Kathy Wood's arc is loading back up on Microsoft and meta after a multi-month absence from both of those stocks. For the first time in five months, Kathy Wood has bought MetaStock, and for the first time in nine months, the company is loaded up on Microsoft. One of the fascinating sub-stories of 2023 will be the extent to which enthusiasm around

AI boasted the market in the face of what seems like a never-ending supply of negative news and other parts of the market. Lastly, today, a nice little interesting story that actually serves in some ways as counter to what's going to be a part of the main episode. A startup from two ex-googlers who wrote a famed research paper while they were at the company has come out of stealth and announce that they've raised $57 million. The company is called Essential AI and says that they're basically trying to build the enterprise brain. From Bloomberg, the company will use AI for corporate functions such as data analysis and promises to automate monotonous tasks. Now, of course, there are numerous companies going after the same space, but with $57 million and the type of talent

that this company has, it seems like investors think that they have some particular insights, which might give them a leg up. However, as you will see if you stick around, it seems like increasingly this sort of mega fundraising round is harder and harder for AI startups to come by. Anyways, friends, that is where we will wrap the brief. Up next, the main AI breakdown. Hey, guys, before we get into the main part of the episode, I wanted to mention just briefly that we are now in the midst, we're actually just closing out the first week of the AI breakdown AI education

and learning beta. This is a community of learners where each day I'm dropping in tutorials, case studies, challenges, and a community of people are discussing them, going out and doing those challenges, in other words, learning AI by doing, and getting a chance to ask questions and talk with people who are experiencing similar problems,

taking advantage of similar opportunities, and generally adapting to this new AI-powered world. I'm incredibly encouraged by how it's going so far, and in about a week I'll be opening up registration for next month's second beta test for January. For now, I wanted to let you guys know that that was coming, and if you are interested in getting on the wait list for that, go to bit.ly-slash-a-i-beta. You'll see the short write-up that I did of December's beta,

plus a link to a form where you can sign up for the waitlist. I'd love to have you participate in January. So again, that's bit.ly slash AI beta. And now let's get to the main episode. Welcome back to the AI breakdown. Today we're starting something that is informal, but is probably a series that I'm going to do throughout the rest of this month, where we look at big trends heading into 2024. Now, in some ways, this whole way of thinking kind of kicked off and was sort of inevitable coming off of the one-year anniversary of Chad GPT, of course November 30th, and naturally this is finding its way into end-of-year coverage, and I think it's a good time to reflect on what has been, as well as what's coming, and one of

the things that you are starting to see is a growing conversation around the funding dynamics around artificial intelligence startups. Now, for those of you who are sitting there saying, this doesn't really matter for me, I'm not running an AI startup, so why should I care? I think that actually the funding dynamics of the startup industry are sort of a proxy for a number of other trends, as you will see as we dig into it. But before we get into it, but before we get into the latest reporting and especially recent data from Pitchbook, let's put this in the larger context of how startups have been shifting in general. For a long period of time between sort of the end of the global financial crisis and just about two years ago now, we were in an extremely

low interest rate environment. This was of course because of the massive stimulus that was needed to right-side the economy coming off of the global financial crisis, which sort of just became a more permanent policy over the subsequent decade. Now, during this time, there was an ever-present shift of more and more investors moving farther out on the risk spectrum to find yield. Basically, in a world where you couldn't get 5% or even 3% off of bonds, you just had to go do more risky stuff. What that meant, practically, was more capital than ever coming into private equity and yes, venture capital.

This led to lots of different things, including higher valuations for startups on average, as well as just more startups getting funded. Another impact was that startups were able to stay private for longer, because more and more often you were seeing Series E, Series F, Series G, type, rounds. Now, of course, this was nothing but supercharged during COVID when there was massive both fiscal and monetary stimulus at the same time. Depending on your perspective, this was either a golden age for startup fundraising or a misallocation healthcape, but in either case, there was no way it was going to last forever. Now, by the time inflation started creeping up at the end of 2021,

there had been people screaming for a while that the Fed really was behind the eight ball and that they really needed to ship their policy in order to not see just runaway inflation. Of course, that runaway inflation is exactly what happened in 2022, and the Federal Reserve shifted from this incredibly accommodative policy from the zero interest rate era into the fastest rate hiking cycle in 40 years. Just as startups and crypto and the things that were on the farthest end of the risk spectrum benefited the most, when stimulus and liquidity were forthcoming, the withdrawal of that liquidity had a more deleterious impact on those sectors than on anyone else. Obviously, we saw the impact of that in many different sectors. Crypto was decimated, although of course that had a lot to

do with fraud and own goals as well, but in general, startups have had a very rocky run of it. Now, one exception to this seemed to be AI startups. Even as towards the end of 2022, other sub-sectors of technology had started to internalize the lesson that for the first time in more than a decade, companies were going to have to think about things like profitability and move away from the growth at any cost sort of mindset that had dominated for the last 10 to 12 years, AI startups remained the exception. In the wake of and around ChatGPT's launch, there was so much capital pouring in and around AI companies, almost as a sort of last gasp of the Zerp-era-VC mindset. Now, already by the middle of the year, this had started to shift. AI engineer Sam Hogan wrote a

very long and frequently shared thread back in July where he wrote, six months ago, it looked like AI and LLMs were going to bring a much-needed revival to the venture startup ecosystem after a tough few years. With companies like Jasper starting to slow down, it's looking like this may not be the case. Sam goes on to articulate from there a couple of different theses that VCs had that seem not to have borne out. One was around big enterprise-focused companies whose product was ultimately a, quote, generic thin wrapper around OpenAI. Sam writes their U.X and brand are good but not great, and competition from companies building differentiated products specifically for high-value niches are making it very hard to grow with such a generic product. So one category that wasn't doing so well were

these enterprise wrappers. The other category of losers, as Sam put it, were the quote, BC-backed teams building at the application layer that raised 250K to 25 million in December to March on the back of the chatbot craze with the expectation that they would be able to sell to later stage in enterprise companies. These startups typically have products that are more focused in something very generic like Jasper, but still don't have a real technology moat. The products are easy to copy. Now from there, Sam goes into a whole argument for why, surprisingly, enterprise companies had actually been spinning up their own tools in a way that we hadn't seen with previous technology movements. But all in all, the TLDR was that already, even just a few months after that big funding

bonanza, which again happened in the wake of ChatGBT, BT, there were already big questions around how sustainable it was likely to be. All of that said, AI still continued to be a funding bright spot relative to other tech sectors. On Bloomberg headline from October reads, AI funding soars to 17.9 billion while rest of tech slumps. The article reads, Multi-billion-dollar investments in artificial intelligence startups have become almost commonplace in Silicon Valley, with dollars raised for AI companies outpacing funding totals in every other category of tech, and reaching 17.9 billion in the third quarter. According to Pitchbook data compiled for Bloomberg, the value of funding for AI companies climbed 27% globally in the third quarter compared to the year before. That's even as overall deals for startups fell 31% from a year earlier to hit 73 billion worldwide.

In other words, in the third quarter, AI company funding was up 27% while overall startup funding was down 31%. That's a 58% gap between how well AI startups were doing relative to a year before, as opposed to startups in general from a year before. Putting a fine point on this, Bloomberg writes, the opposing trend lines highlight a divide between AI startups and the rest of the industry. Rising interest rates in a post-pandemic slump have hammered VC funding, making AI one of the venture capital world's lone bright spots.

Now, at the same time, those statistics were sort of still obscuring the larger point, which could be seen in Pitchbook's assessment of its own data, which they summed up as generating less momentum, generative AI deal count dips in Q3. Pitchbook writes, it seems unthinkable, but momentum for generative AI appears to be slowing after captivating VCs in the broader public, leading to a surge in VC funding. While investment remains high by historic standards, excitement appears to be waning in the sector as large tech companies make waves and investors realize that many generative AI applications may not be ready for prime time.

The number of generative AI deals fell to 101 rounds in Q3, a 29% decline from Q2. Deal value was also on a downward track, but ended higher at $6.1 billion thanks to the Blockbuster deal for up to $4 billion that Amazon inked with Anthropic. Said Brian Offutt, a partner at Index Ventures, Momentum is definitely waning as the market comes back to Earth. We are living in the messy middle of AI. Again from Pitchbook, Pete Flint, a general partner at NFX, said there's a realization now among founders and investors that some early companies were interesting experiments, but not great

businesses. He said, the retention and monetization are just not there. Still the bigger piece comes in the next paragraph, which says, Flint added that big tech is looming large over the generative AI space, contributing to the slowdown by potentially scaring away startups and investors. He said, it's clear that established digital incumbents and challengers are not asleep at the wheel. Indeed, there's a bunch of ways that this reality is showing up in the market. First of all, as TechCrunch points out, mega deals could be inflating overall AI funding figures. And this one points out that those big numbers are incredibly top-heavy with the $10 billion investment from Microsoft into Open AI, the $4 billion into Anthropic from Amazon, even though $1.3 billion into inflection.

All of these things give the appearance of an overall healthy sector when really it's a very small number of firms that are leading the way. What's more, a huge amount of capital is coming from the big tech incumbents, which is a very different phenomenon. and something that we've discussed is very different from previous tech movements. Now, part of that is just that it costs more to compete in this space than almost any other technology sector. There's only so much compute to go around, and in many ways a lot of the big labs are frankly outstripping the capacity of the venture capital industry to service their needs. But beyond that, there is, just as that VC put it a moment ago, a messy middle of AI.

One of the things that we talk about a lot on this show is how we are moving into a period where, yes, even as there is excitement around new frontier models and, increased capabilities, a lot of the emphasis is shifting to integration of AI tools into existing workflows and seeing where the value actually arises. AI content creator and trainer Greg Cameron has been looking for stories of AI actually working in the workplace and tweeted this morning, lots of new models, not a lot of new stories of AI creating value. Now, I have a lot of theses about what's going on, and I think it's a lot less bleak than

it seems on the surface. TLDR, I think that the inertia of corporate and enterprise processes is powerful enough that it can slow down even artificial intelligence. What's more, given how quickly AI is rooting itself in people's personal lives and the way that they interact with the internet in an individual capacity, the gap between how an individual is using it and how it's finding its way into the enterprise feels more jarring than it might otherwise. Anyways, that's starting to get into a whole different topic, which again we will probably discuss at some point. But for now, it is leading to this larger sense in a shifting of the AI startup funding environment that is having an impact on how the field is evolving. That information article that we started with, some AI startups find the money is no longer so easy,

tells the story of a number of companies that went out to raise $100 million rounds over the summer and have had to settle for much more tempered investments. Now, that piece makes it seem even clearer that VC consensus is moving to a belief that the big tech incumbents are going to be the big winners from this whole shift. Now, obviously a world in which VCs are retreating because they just assume that MET is going to win looks very different than the previous world's venture capital in the past and has big implications for the ability of startups to compete. There is obviously a self-fulfilling prophecy aspect of this

that's going to be really interesting to keep an eye on. And so for now, that's what we're going to do. This is a story that continues to evolve every day and every week, but one to watch for sure. That, however, we'll do it for today's AI breakdown. Until next time, peace.