The AI Daily Brief: Artificial Intelligence News and Analysis - AI is Personal: On Microsoft's Copilot Super Bowl Ad

Episode Date: February 7, 2024

Microsoft has released a 60-second Super Bowl commercial focused on their AI Copilot. NLW explores how much it says about the way the company is thinking about AI (and how central it is to Microsoft's future). Also on the show: can Nvidia use AI to speed up its chip design process? 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, we're looking at Microsoft's just-released Super Bowl ad focused on copilot. Before that on the brief, how NVIDIA is using AI to improve how it builds AI chips. 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 YouTube, our newsletter, and our Discord. Welcome back to the AI breakdown brief, all the AI headline news you need in around five minutes. One thing that has been a clear phenomenon for the last year is that access to computing power is constraining the development of advanced artificial intelligence. This has, of course, driven companies like Nvidia and AMD to record market caps, and it has also apparently

driven Nvidia to try to use AI to design AI chips faster. Invidia has an AI system called Chip Nemo that's built on top of Metaslamma 2 and trained with Nvidia's own data. Chip Nemo has a chatbot feature that's able to respond to chip design questions such as GPU architecture, and this matters because an advanced Nvidia chip apparently takes close to 1,000 people to build. Now, according to the Wall Street Journal, since Chip Nemo was put into the process last October, it has apparently increased productivity in ways like summarizing notes across 100 different teams and training junior engineers. Now, for now, it seems like chip Nemo is more about improving the human processes surrounding chip design and production,

rather than AI actually improving AI itself, but that hasn't stopped some folks from worrying that this is the beginning of the self-recursive AI development era in which AI is used to improve other AI, which is where some people are concerned that we might lose control. For now, though, we are dealing with simple market forces. Everyone wants chips. They're hard and expensive to produce, and Nvidia's trying to make that a little bit cheaper, faster, and easier. Now, one of the companies that is Nvidia's biggest customers is, of course, meta. The company has announced a new policy around watermarking AI photos on Instagram threads and Facebook. Now, there are a couple parts to this. One is applying a watermark to images that are created with its own AI generator,

and then expanding that to AI generated photos made with tools ranging from Dali to Mid Journey to Microsoft and Google, but they are also demanding that users who update realistic AI photos and videos have to disclose it or else face pretty severe punishment. Now, this is very clearly part and parcel of concerns around the 2024 elections. Facebook has been a very hot topic when it comes to election integrity all the way since 2016, and I'm sure they are trying to get out ahead of anticipated issues this year. There are also reasons to be concerned. Deepfakes are getting more sophisticated.

You might have heard about this AI heist that recently happened, where a Hong Kong multinational was tricked out of $25 million when a number of different team members, including the CFO, showed up on a team video chat and started instructing one of the employees to begin transferring funds. Now, of course, it wasn't actually the CFO and the other employees weren't actually those employees. It was a sophisticated deepfake,

although we don't have information about exactly how it was produced. In other words, whether it was something like real-time face-swapping versus something that was more pre-recorded. In any case, though, when you see companies that are losing tens of millions of dollars because a video is convincing enough for someone to think it's their CFO, it dramatizes just how big an issue deep-fake content is likely to be. Now, meanwhile, over in the U.S. government, a couple interesting things around the AI space. First, a group of 20 state attorneys general led by the Utah State Attorney General have written a letter to Commerce Secretary Gino Raimondo, warning about President Biden's use of the Defense Production Act as a way to justify his executive order on artificial intelligence. The letter reads,

The executive order seeks without congressional authorization to centralize government control over an emerging technology being developed by the private sector. In doing so, the executive order opens the door to using the federal government's control over AI for political ends, such as censoring responses in the name of combating disinformation. Now, whether this letter actually goes anywhere is probably pretty dubious, but it shows how this issue is likely to get politicized in the months to come. Meanwhile, even as politics comes to AI, the U.S. government is retrofitting itself around using AI as well. The Department of Homeland Security has announced that it's trying to hire 50 AI experts this year to help it with everything from halting child abuse to countering fentanyl production,

to assessing damage from natural disasters. Said Secretary Alejandro Mayorkas, government needs the support and expertise of our country's foremost AI experts to help ensure our continued ability to harness this technology responsibly, safeguard against its malicious use, and advance our critical homeland security mission. Now, finally today, two little bits of news from Microsoft, who are the focus of our main episode as well with their new just-released Super Bowl ad about co-pilot. One, in a speech in Mumbai, Microsoft CEO Satya Nadella has argued that AI is being adopted and is diffusing across the world at a rate higher than any advance in the computing era of the last 70 years. He also said that,

Microsoft intends to help train 2 million people in India on the use of AI and encourage national governments like the Indian government to start taking a leading role because, as he put it, this new capability AI is going to have an impact on GDP. Now, a couple of days ago in a very different type of story, Microsoft announced a partnership with next generation news organization semaphore. One of the things that I think is fascinating right now is that even as legal battles are being fought between, for example, news organizations and AI labs like Microsoft and OpenAI, those same types of parties are also finding common ground in partnership. Roeders reports that through the collaboration with Semaphore,

quote, Microsoft will help the organization to identify and refine the procedures and policies to use AI responsibly in news gathering and business practices. For example, they'll be launching a breaking news feed called signals, which, quote, journalists can use with the help of tools from OpenAI and Microsoft to provide readers with analysis and insights on breaking news stories. Now, one of the things that you've heard from people like Sam Altman is that their interest in these types of partnerships with news organizations isn't about getting access to training data. Altman's argument has in general been that any single source of data,

even like the New York Times' past body of articles, is ultimately fairly insignificant relative to the corpus of everything that's out there. Instead, what he has argued is that they're interested in helping news organizations become better. This type of partnership is certainly evidence that they are at least thinking in those terms. However, that is going to do it for today's AI breakdown brief. Next up, the main AI breakdown. There is no bigger event in the world for advertising than the Super Bowl. You might even say that the Super Bowl is the Super Bowl of advertising. What makes it special is that it is the only time in the calendar year

that Americans actually want to be advertised to. Meaningful percentages of the people who watch the Super Bowl say they're there more for the ads than for the game. What that means for big brands and companies is that it's a chance for them to make a very strong point about what they are going to be focused on that year. A Super Bowl ad is never just a one-off ad. It is almost always the beginning of a campaign.

Today, Microsoft released a 60-second ad, which, by the way, costs on the order of $11 to $12 million just for that ad space to say nothing of the production. And the focus of Microsoft's ad is AI, specifically copilot. Now, as the Verge writes, this in many ways is fully the end of the transition away from the Bing brand

and toward the co-pilot brand. Now, in that, there's nothing new. That's been a shift that's been happening for some time. As the verge points out, their plan to reignite interest in Bing kind of just didn't work. They write, a year ago today, Microsoft unveiled its ambitious plan for an AI-powered Bing search engine. It was the biggest launch in the history of Bing, helped push AI usage even further into the mainstream, and spurred a wave of dreams and panic about what AI could impact next. One year later, Bing seems to have fallen out of the conversation. Google is still at over 91% of traditional search market share,

and ChatGPT has rocketed to 100 million weekly users, all while Bing grew by less than half a percent in the search market globally. So what has changed is that Microsoft stopped thinking about generative AI as an opportunity just for their search engine, but instead as an opportunity to change and update every piece of software they put out. Of course, as you'll know from the show, all of those AI features are called co-pilot. And the emphasis, and the story that Microsoft is telling about artificial intelligence, is one in which every person has an AI companion that can help them do more, be more productive,

accomplish more, and generally build a different type of future. So as the Verge points out, this is definitely a departure from Microsoft's traditional advertising, whereas the focus has generally in the past been on the PC or specific devices, this ad is trying to tell a story about the world. It shows a really different sensibility at Microsoft about how Copilot fits into their future. In many ways, Copilot is becoming the leading consumer-facing brand inside the Microsoft's universe. Because it touches everything, it is in some ways bigger than Office or 365, even if it doesn't have the sort of name recognition that those do yet. The other reason that co-pilot, I think, is becoming Microsoft's main consumer-facing brand,

is that it tells a story not about them per se, but about their customers. And that's why I actually think this ad is tremendously well done. What this ad says about AI is that it's a tool for making you more you, for doing more of the things that you want to do, for achieving more of the things that you want to achieve. It isn't about being replaced or obsoleted. It isn't even about the world. It's about how you operate within it. On that journey, AI is your co-pilot. AI gives you superpowers. Now, obviously, part of the attempt here for Microsoft is to not just get copilot to be well-known, but to own and be at the very center of the idea of an AI assistant. That's obviously very different than how people, for example, imagine chat GPT right now. It's certainly different than how people

think of Google search. But it will certainly be something that you see more and more as other companies bring to market similar types of assistant experiences. Now, in interviews around the ad, executive vice president and consumer chief marketing officer at Microsoft use of Medi has shared a little bit more about the future of co-pilot as well. He said, for example, over time, I think you'll see us continue to add more and more of those things. So the notion of a personal co-pilot that is yours, we want to get to one idea, and we want that to unlock everything you've got with your IDs, with your personal IDs, and work IDs. Where we go from there, we'll see. There are lots of extensions that are coming to copilot, whether those are GPs or plugins, or the ability to do

custom co-pilots themselves. Medi even suggested that copilot is going to be at the center of a changing Windows brand. The Verge writes, Medi didn't want to talk about the specifics of how Microsoft will overhaul Windows for AI, but he did drop some red-cour. terms of what to expect. He said, The unique thing of co-pilot inside Windows is that it can be aware of the context you're in. It can understand the pages so it can do more rich things. He also suggested that co-pilot inside Windows will fundamentally change how people interact with the computer. He said, I think it's something like 20% of Windows users use 10% of the

features. Once you can say, hey, turn my PC into dark mode, configure that printer for me, help me get the following going, we can turn everyone into a power user of Windows. Sounds cliche, but what I think it will unlock in terms of people's ability to to use computers to do amazing things will be quite profound. Anyway, I think it's a great ad, and I think it says a lot about where Microsoft is that this is what they chose to focus on. Now, that said, early reports suggest

that there aren't going to be a ton of AI ads in this year's big game. People are speculating that there will be a, quote, notable shift away from technology, like AI in crypto. I'm not so sure. I wouldn't be surprised to see a Google AI ad show up, even if I do agree that crypto is unlikely to make much of an appearance. Anyway, interesting stuff.

Appreciate you watching or listening as always. And until next time, peace.