5 Things to Watch at Microsoft Build

0:00Today on the AI Daily Brief, we're talking about five things that I'm watching for with Microsoft Build this week. Before that in the headlines, Snapchat is going to spend up to$1.5 billion annually on AI infrastructure. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, check out the Discord link in the show notes.

0:24Welcome back to the AI Daily Brief Headlines Edition, all the daily AI headlines you need in around five minutes. Today, we kick off with a story around how one social technology player plans to stay relevant in the era of artificial intelligence. And that is, of course, by spending a bunch of money on artificial intelligence. The headline is that Snapchat is aiming to spend around a billion and a half dollars per year on AI. This came out of a Bloomberg interview published with CEO Evan Spiegel, where he said, There was a recognition that we'd fallen behind the curve on the machine learning side, which to some degree was reflected in the business performance.

0:58We needed to improve there and bring together some of our most senior machine learning folks to just talk about what it would look like for us to get to state of the art and really invest. Now obviously there is a big buzzy headline with this$1.5 billion, but what's really interesting about this story is the way that artificial intelligence is changing the fundamental business model of some of these companies. From payments.com, the report noted that Snapchat has been dealing with major shifts in both how people use social media and the types of ads marketers want to buy on them. There was a time when the platform revolved around messaging, with advertisers focused more on telling brand stories instead of trying to drive immediate purchases.

1:31Now the company is spending money on machine learning, AI, and augmented reality features that tap into Snap's ads business and user feeds. So what are the types of things that Snapchat might do with AI? Well, one influencer characterized it like this. Sarah Safari, a fitness influencer with more than 2 million followers on Instagram, said, AI can streamline what used to be a tedious influencer discovery and relationship building process by creating instant filters, screening for influencer fraud, engagement metrics, and the forecasting of campaign results based on past data. This means that campaigns can be better targeted, evaluated, and eventually run at scale.

2:01I continue to think that marketing, writ large, is one of the areas that will adopt AI the fastest and the most aggressively, in part because it is the most receptive to the types of cost efficiencies and performance increases that AI can bring, in part because marketing has gotten used to quickly adapting to new types of technologies over the past decade. Speaking of platforms where marketers spend lots of their money, even as they fight the battle in the US to keep TikTok unbanned, ByteDance is doing well in native China. Bloomberg again reports that ByteDance-owned Dubow, which is an AI chatbot released about a year ago last August, has surpassed Baidu's Ernie bot in overall downloads and now has more monthly users on iOS in China.

2:39One thing that's worth noting, this is just on iOS devices. Bloomberg says while the research does not include Android mobile stores, it illustrates the wider trend in the Chinese market. One interesting difference between the Chinese chatbot market and ChatGPT and Claude, etc. in the US is that Ernie is the only one of the top five chatbots that actually offers a paid subscription tier. And according to data analytics firm Sensor Tower, they say that Ernie has brought in less than a half million dollars from in-app purchases and subscribers since it launched in March of last year, which seems incredibly low.

3:08But apparently advertising is the much bigger market, and Baidu says Ernie will bring in several billion yuan of additional revenue through advertising and cloud services this year. Moving from China over to the UK now, chair of the UK AI Safety Institute Ian Hogarth reflected on one year in that role and made an announcement that they're opening up a new office in San Francisco. Ian writes, It's been one year since I was appointed chair of the UK AI Safety Institute. In this time, we've built one of the largest safety evaluation teams globally and are already conducting pre-deployment testing.

3:37This is our fourth progress report. Over the last year, we've gone from building a startup inside government to shipping product. We've built one of the largest safety evaluation teams globally with a team of over 30 technical researchers and counting. And we're not just scaling in the UK. We're now opening an AISI office in San Francisco to cement our transatlantic partnership with the U.S. and to work with the best and brightest talent on both sides of the Atlantic. A press release from the U.K. government reads, The San Francisco expansion marks a pivotal step that will allow the U.K. to tap into the wealth of tech talent available in the Bay Area, engage with the world's largest AI labs headquartered in both London and San Francisco, and cement relationships with the United States to advance AI safety for the public interest.

4:13Secretary of State for Science and Technology Michelle Donilon said, This expansion represents British leadership in AI in action. It's a pivotal moment in the UK's ability to study both the risks and potential of AI from a global lens, strengthening our partnership with the US and paving the way for other countries to tap into our expertise as we continue to lead the world on AI safety. Speaking of AI safety, Jan Leakey from OpenAI kicked up a dust storm at the end of last week when he resigned as head of alignment and the super alignment lead and took some shots at OpenAI on the way out. He wrote, Stepping away from this job has been one of the hardest things I have ever done because we urgently need to figure out how to steer and control AI systems much smarter than us.

4:47I joined because I thought OpenAI would be the best place in the world to do this research. However, I have been disagreeing with OpenAI leadership about the company's core priorities for quite some time, until we finally reached a breaking point. I believe much more of our bandwidth should be getting ready for the next generation of models on security, monitoring, preparedness, safety, adversarial robustness, super alignment, confidentiality, societal impact, and related topics. These problems are quite hard to get right, and I am concerned we aren't on a trajectory to get there. Over the past few months, my team has been sailing against the wind.

5:13Sometimes we were struggling for compute, and it was getting harder and harder to get this crucial research done. Building smarter than human machines is an inherently dangerous endeavor. OpenAI is shouldering an enormous responsibility on behalf of all of humanity. But over the past year, safety culture and processes have taken a backseat to shiny products. We are long overdue in getting incredibly serious about the implications of AGI. We must prioritize preparing for them as best we can. Only then can we ensure AGI benefits all of humanity. OpenAI must become a safety-first AGI company. To all OpenAI employees, I want to say, learn to feel the AGI.

5:43Act with the gravitas appropriate for what you're building. I believe you can ship the cultural change that's needed. I'm counting on you. Now, this is deserving of a much bigger discussion. It feels to me like we're starting to see a bit more of a cleave where the AI safety side of the space feels greater and greater concern every day, while at the same time, people on the other side of that are moving away from even giving it lip service. It's a really interesting tension that I think could have pretty dramatic impacts on how the field of AI evolves. But for now, that is going to do it for the AI Daily Brief Headlines edition.

6:10Stick around for the main episode. Before we get back to the show, I want to tell you a little bit about Notion. Now you've probably heard about Notion, the AI-powered workspace tool where any team can turn ideas into action, but what you may not have had is some reference point for just how much people who use Notion use Notion. I'm recording this ad at 2.31pm and have interacted with more than 20 different Notion documents. I used one for taking voice notes while I was on a walk this morning, another for keeping my to-do list for the week, another for sharing product priorities with my team, another for reviewing contracts with a different part of my team, another for brainstorming copy for social posts with yet a different part of my team, which by the way used the new Notion AI tools to actually brainstorm those ideas, and so on and so forth you get the idea.

6:56Notion combines your notes and documents into one space that's simple and beautifully designed, with, as I mentioned, the power of AI built right inside. It's a place where any team can write, plan, organize, and rediscover the joy of play. And it's for everybody, whether you're a Fortune 500 company or a freelance designer starting a new startup or a student juggling classes and clubs. Try Notion for free when you go to Notion.com slash AI Breakdown. That's all lowercase letters, Notion.com slash AI Breakdown, and start turning ideas into action. When you use our link, you're supporting our show.

7:25Notion.com slash AI Breakdown. Hello, friends. Before we get back to the episode, I want to tell you about something special I'm doing on Super Intelligent this June. Super is, of course, our platform for AI learning. And I've heard from a lot of you that you really want something for a true AI beginner, someone who's really just getting their feet wet with these tools. So what I'm going to do is put together basically a course that sits on top of and uses super intelligent tutorials and lessons, but where I hand guide you through around 10 different lessons and how to's that I think once you complete them, we'll have you ahead of 80 % of the other people who are just starting to use AI right now.

7:59If you are interested in this learning experience, go to besuper.ai and sign up using code June. You'll get 25 % off your first month, and I'll automatically add you to that AI for Beginners group. That's bsuper.ai, discount code June. See you there. Welcome back to the AI Daily Brief. We are in the season of big announcements, annual developer conferences, and of course that means we are getting to learn a lot about the various big tech and big lab strategy when it comes to AI in the immediate future. Last week was a pretty interesting one to kick this season off. On On Monday, OpenAI tried to scoot in ahead of Google, announcing GPT-4.0, and clearly putting emphasis on some new modalities of interacting with AI as the future they're driving towards.

8:42Now, you could tell that OpenAI was sort of rushing to get this announcement out because it is very unlike them to announce something and not make it available, and while we did get an update to ChatGPT with the GPT-4.0 model, we have not yet gotten the updated voice tools, which so much of the presentation was about. The next day on Tuesday, we also got Google's version of an AI assistant in Project Astra, but we also got about a thousand other announcements. And overall, it's clear how these events showed the different strategies of these companies coming to light. OpenAI continues to try to be at the very edge of the state of the art, pushing consumer experiences in the direction that they think AI is driving us, while Google, while not surrendering the state of the art, seems to be ultimately more concerned with integrating AI across its suite of products and taking advantage of the existing installs they have as a competitive advantage.

9:28So what can we expect from Microsoft? Well, the theme of the event is how will AI shape your future? So no big suspense on what the main topic is going to be. But I put together five things that I am watching for at this event that I think will tell us a lot about Microsoft's strategy going forward. First question, what's the centerpiece they try to get attention around? Will it be the rumored AI Explorer, which is theoretically deeply integrated into Windows? Based on the buzz that we've seen in advance, it seems like Microsoft really wants to make sure that Windows is the first AI-enabled operating system.

9:58The big question at this event will be how much this is buried in the rest of the announcements versus presented right up front as the big thing. As you'll see throughout this episode, Microsoft has a lot of different choices around where they could put the emphasis. They could put it here in the operating system. They could put it in hardware. They could put it on search. Alternatively, they could focus on co-pilot integrations across their enterprise suite. Now, they may do a little bit of what Google did, where they just pummel us with so many different things that it's not exactly clear where there is singular emphasis.

10:25and instead they might just try to show that AI is for Microsoft everywhere and let us sort it out, but where they decide to put that emphasis is ultimately going to be one of the most instructive things that could come out of this event. Related, do we see a new even more AI-integrated Bing search experience? Where this question is coming from is a couple parts. First of all, Google is pretty fundamentally changing its search experience with its new AI overviews. This is the update for what they had previously called the search generative experience, and basically now Google is just normalizing AI overviews as a part of every search.

10:57This represents the biggest shift to the experience of Google search since it was launched. And if you've spent any time around people in the SEO industry or any of the other areas that touch Google search, you can see what a big deal it is as they try to figure out exactly what it's going to mean. To some, it represents a fundamental break in the architecture of search on the internet that could have pretty dramatic impacts on how people get to different websites. Now, at the same time, before OpenAI's event, we had gotten a bunch of rumors that they might be launching a search experience as well.

11:25And while they didn't exactly, at least they didn't in name, it seems pretty clear to me that the input field on ChatGPT is just going to become the starting point for a ChatGPT search experience. In other words, search is just going to become an integrated feature of ChatGPT. And even if they don't explicitly announce a competitor to Perplexity and Google Search, that's effectively what they're building. The more they can get people to shift their behavior over to asking questions of a chatbot like ChatGPT, the more natural it will be that people go there to get answers to the types of questions they might have used search for before.

11:55The question is, where does that leave Bing? Is Bing's strategy just to be the engine that helps power ChatGPT? Or are they trying something else? I'm not sure how much we'll learn about that this week, but it seems like a pretty relevant question. A third thing I'm watching for, sort of related to what we were just discussing, is how much Microsoft touts their OpenAI partnership at this event. There have been tons of discussions about Microsoft potentially moving away from or at least hedging their relationship with OpenAI, and the question is whether we'll see more evidence of that. So where have these ideas that Microsoft might be hedging come from?

12:25Well, first of all, we're seeing them work on new models that are in-house and not connected to OpenAI. We also had the husk of Inflection come over and join Microsoft, with the Inflection co-founder Mustafa Suleiman joining to lead Microsoft's AI efforts overall. And we've also seen them partner with companies like Mistral and Databricks in ways that some have taken as a sign that they may be concerned about their relationship with OpenAI. Now, my take has been that especially in the wake of Sam Altman's firing and rehiring, their hand is pretty well forced in this. The deal that Microsoft has with OpenAI allows them to commercialize OpenAI's technology right up to the point where OpenAI achieves AGI.

13:00What achieving AGI means, though, is determined by OpenAI's board. after we saw what happened last November, with the board acting in a way that had to look pretty capricious from Microsoft's point of view, are you really going to just leave your long-term business strategy when it comes to AI to them? And this is why I don't believe that anyone is lying or putting on a front when they say that they hope their relationship with OpenAI remains as good as it does now forever. I think that is genuinely what they're hoping for. However, that doesn't mean that they can just sit pollyannishly and hope for the best.

13:28There's simply too much at stake. Anyways, it'll be interesting to see whether they use this as a chance to start to carve out some narrative space away from OpenAI, or whether the connection and partnership is discussed as full-throatedly as ever. Somewhat related, I'm watching to see if they announce their own major LLM initiative that competes with GPT. When news about Microsoft's in-house model MAI-1 broke, the company's CTO said that they always are making new models. Some of them are internal, some of them are external, and then it's not a big deal. That'll be put to the test a little bit this week in terms of where they decide to put their emphasis.

13:58If they don't really discuss it or kind of breeze over it, it makes that idea that this is just something they're always going to be working on in the background feel a little bit more legit. Finally, a fifth question is how much emphasis Microsoft puts on AI hardware versus AI software. The other big contender, of course, for what Microsoft chooses to focus on at Build is its vision for AI PCs. And part of the reason that they might want to put their emphasis there is that according to sources, it seems like they believe that their new laptops powered with ARM chips can actually beat Apple's M3 powered MacBook Air when it comes to CPU and AI-accelerated tasks.

14:31Writes The Verge, After years of failed promises from Qualcomm, Microsoft believes the upcoming Snapdragon X Elite processors will finally offer the performance it has been looking for to push Windows on ARM much more aggressively. Microsoft is so confident in these new Qualcomm chips that it's planning a number of demos that will show how these processors will be faster than an M3 MacBook Air for CPU tasks, AI acceleration, and even app emulation. Part of what makes this interesting to watch is that while all of the big computer makers are very stoked on this idea of the AI PC era, it's not at all clear that consumers are bought in on this yet.

15:00And for that reason, I think it would be interesting if Microsoft does decide to put a big chunk of its narrative chips in that area. Anyways, it's gonna be interesting no matter what. The event kicks off tomorrow, and I'll make sure to keep you posted around what we find out. For now though, that is gonna do it for today's AI Daily Brief. Until next time, peace.

15:25Thank you.