The AI Daily Brief: Artificial Intelligence News and Analysis - The AI Wearable Wars

Episode Date: October 3, 2023

Tab and Rewind Pendant both announced AI wearables this week, while Humane premiered on a fashion runway. In this episode, NLW looks at the emerging wearable wars, and asks whether these are even use cases that people want? Before that on the Brief: Jamie Dimon says a 3.5-day workweek is possible and Satya Nadella testifies that AI can't stop Google's search dominance. TAKE OUR SURVEY ON EDUCATIONAL AND LEARNING RESOURCE CONTENT: https://bit.ly/aibreakdownsurvey 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 a set of new AI wearables that promised to personalize help by listening to everything that people say. The question is not only which of these products will succeed, but whether people actually want them at all. Before that on the brief, could we be headed for a three and a half day work week? According to one of Finance's biggest minds, it's entirely possible. 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 channel, our Discord, and our newsletter. Hello, friends. Just a quick note before we start the brief. Yesterday, I mentioned that each month I was making available a very small number of educational

slash advisory slash consulting sessions. This is a high-impact short format that's designed to help give you more tools to figure out how to apply AI to your business or to your career or whatever it is that you need AI for. Great response yesterday, so I just have one or at most two slots left. So if you are interested in learning more, send me a note at NLW at breakdown.network. Otherwise, there will be more opportunities opened up at the beginning of next month. With that, let's get to today's brief. Welcome back to the AI breakdown brief.

All the AI headline news you need in around five minutes. Now, when it comes to artificial intelligence, there is obviously a lot of anxiety around what it's going to mean for the future of jobs. Many people are concerned that their job skills might be totally automated away, forcing a pretty fundamental rethink of their careers. But the flip side and proponents of the technology, argued that the increases in human productivity could lead to a flourishing of human society in which people just don't have to work as much as they once did.

Count among that camp, apparently, J.P. Morgan's CEO, Jamie Diamond. In a recent interview with Bloomberg, he said that even though AI is likely to, yes, eliminate some jobs, there are going to be amazing benefits as well. Diamond said, your children are going to live to 100 and not have cancer because of technology. And literally, they'll probably be working three and a half days a week. Now, putting his money where his mouth is, J.P. Morgan has apparently advertised for some 3,500 related roles earlier this year around artificial intelligence and had an entire section in his shareholder letter this year dedicated to

AI. He claimed that the company already had more than 300 use cases in production. Now, in terms of what he's concerned about, Diamond said, quote, the biggest negative in my view is AI being used by bad people to do bad things. Now, of course, if we take this optimistic look at AI increasing productivity, a reasonable skepticism is the fact that with previous productivity enhances, we didn't necessarily get dramatic decreases in how much people worked. Instead, we saw people just fill in that time, and if anything, the boundaries between work and not work have gotten bluerier. Point being that yes, AI might increase productivity, but ultimately, what we do with those gains is going to have to come from a societal conversation around an updated social contract. Just because we can work three and a half

day work weeks doesn't mean that that will be the norm, unless society comes together on mass and decides that's going to be the norm. Now, growing experiments with a four-day workweek currently suggests that maybe it's less unrealistic than it seems, but the point I guess is that I don't think it's just a question about AI alone. Moving to our next topic, AI came up as part of a set of conversations around a Google antitrust hearing. Yesterday, Microsoft CEO Satya Nandela was on the witness stand in that antitrust case against Google, and basically Nadella reinforced the Justice Department's argument that a big part of Google's dominance with search was through agreements with companies like Apple that made Google search engine the default on the Safari

web browser. He used the phrase vicious cycle to describe the fact that Google takes advantage of its roughly 90% market share to allow it to improve search results more quickly and more efficiently than anyone else, which of course reinforces its monopoly in the space. Indeed, he argued that Silicon Valley sees internet search as an area that is just completely unwinnable, whereas in just about every other field, investors and entrepreneurs think that they can take a crack at even the biggest giants. That's one where that's just not the case. And most relevantly, this show, Nadella said that there were serious limits to how much the rise of AI could actually reshape that market. He said, quote, the distribution advantage Google has today doesn't go away.

In fact, if anything, I worry a lot that, even in spite of my enthusiasm that there is a new angle with AI, this vicious cycle that I'm trapped in could become even more vicious because the defaults get reinforced. Indeed, he even suggested that he might have been overly enthusiastic about ChatGBT's potential to break the cycle with Google. He said, yeah, I mean, look, that's called exuberance of someone who has like 3% share, that maybe I'll have 3.5% share. Now, as the Wall Street Journal pointed out, despite adding AI features to Bing, Microsoft's share of the search market has basically not budged. So three interesting takeaways here. One, the Microsoft CEO walking back some of his enthusiasm around how much AI could disrupt existing monopolies. Two, some new

arguments in the question of how much existing distribution modes are going to influence who wins in the AI space. And three, this is an interesting window into the state of competition. among the big tech giants. Speaking of which, this is something that I think we'll get into more later this week, but the information wrote a really interesting piece called meta-open AI square off over open-source AI. The new reporting basically says here that in a lot of these meetings that have recently happened between the big tech companies and policy makers, one of the central points of contention and one of the biggest arguments among those representing the big tech space is around the question of open source models.

You might see a chart floating around the internet that's called friends and foes of open source AI, using a red light, yellow light, green light, heuristic to explain where different companies are. Anthropic and OpenAI, both being red lights when it comes to open source AI, Microsoft and Google being somewhere in the middle, with meta and Databricks being very pro open source AI. As the information argues, the Lama 2 launch has forced the industry and regulators to take positions. BetaWorks founder John Borthwick basically said that until Lama 2, many people felt that open source LLMs were two to four years behind closed models, but the most recent releases

show that that just wasn't the case. Now, in many ways, all of this reverts back to questions of licensure and where policymakers are going to draw the line on too powerful or not, but I'm presenting it here as another example of how this competitive accelerationist conversation is evolving and where it's likely to lead in terms of policy. Meanwhile, companies and other industries remain focused on rejiggering themselves for the AI era. One example of that, Visa yesterday launched a $100 million generative AI Ventures initiative. It's basically a dedicated AI fund that will sit under Visa Venture. Now, Visa tends to be a very tech-forward company when it comes to the financial sector, having been some of the most active in the crypto and Web3 space, and so I don't think it's at all surprising to see this sort of dedicated focus to generative AI either. Speaking of venture funding, last week we told you about OpenAI being in conversations to raise a little bit of money at an $80 billion to $90 billion valuation.

That investment round, if it came to pass, would primarily involve employees being able to sell shares that they already owned. Well, now another reported fundraising in the works comes from Character AI, and they are out seeking a $5 billion. billion-dollar-plus valuation as well. Now, one of the reasons that they may be fundraising now is sort of a strike while the iron's hot idea, given that meta just announced a set of AI personalities that don't seem very dissimilar from Character AI's main offering. Ritespayments.com, character A.O. offers a service that enables users to create chatbots capable of engaging in text conversations. For example, users have already sent millions of messages to characters like Super Mario based on the Nintendo 64 character and received AI-generated responses. The company's previous funding round valued it at a billion

and this updated round could value it at up to $10 billion, although it seems like $5 billion to $6 billion is more likely. I keep seeing statistics that show a ton of growth and usage of character AI, but I'm still waiting for a better explanation of who's actually using it. As I've said before on the show, I think it's entirely possible that I am just out of tune with an entire generational cohort of users, but it would seem to me that if character AI is getting the usage that it seems like it is, it's going to be largely driven by younger users. Then again, that is where Silicon Valley has typically looked for new social starting, to have their first inroads.

Anyways, friends, an exciting world as always. So many more things brewing every day in artificial intelligence. Up next, the main AI breakdown. One of the big themes that is emerging this fall in terms of AI trends is the introduction of new AI powered hardware. Indeed, we discussed this yesterday in the context of the AI phone wars, with Sam Altman and Johnny Ive talking about a potential device and talking about raising a billion dollars. to fund that company. But today we're talking about something a little bit different. And that is

a set of new products which are designed from the ground up with artificial intelligence in mind. These are not products that have real clear equivalence in the past because what they do wasn't possible in the past. Now, anytime you have a new set of products like that, one of the big questions is whether people are actually going to want it. As humans, we tend to be extremely bad at predicting what technology is actually going to stick in the future, which is why venture capitalists can still make so much money. And adding a little bit of intrigue to this, there is some drama between the companies. So let's start with a new product called Tab. On October 1st, the founder of the company behind Tab, Avi Schiffman tweeted an eight-minute demo of the product with the caption,

I just built the world's most personal wearable AI. You can talk to Tab about anything in your life. Our computers are now our creative partners. Let's listen to Avi for just a quick excerpt of that demo. with all kinds of ideas on like how can you truly give your personal context to such a tool. And I've come up with this wearable that you can see hanging around my neck right now, where throughout my days it ingests the context of my daily life through listening in on all of my conversations. I talk with my friends like all the time about every little tiny detail of what I'm working on, everything from the user experience to just decisions about it, concerns that I have to get brought up. And I would love to be able to seamlessly have this just with my AI.

I'll show you a quick little demo. I think the greatest way to show you one would just be that ask tab. Hey, Tab, I'm on stage right now. What would be a cool way to demo tab? A little Apple transcription right there. As you can see, Tabs, just all kinds of ideas. It understands what Newton is. It understands the demographics of the audience here.

It knows who I am, knows who Tom is. It comes with all kinds of ideas here. For example, why am I asked about the dinner conversations that I had today? I think it's a great idea. So, Tab, what did I discuss during dinner today? Let's see what it comes up with. And I think one thing that I found that is amazing with talking to AI like this is you're computing the way you think.

Usually when you're trying to talk to someone or you're just thinking, you don't prompt engineer your thoughts. You don't explain and write this context or like these random paragraphs of specific features. You just want to talk to your AI. Okay, so there are a couple really interesting things about this that I want to point out. The first is that one of the opinions that I feel like I have that is different for many in the AI space is a skepticism of how personal assistants, as they currently are imagined, are likely to be adopted by the general populace. In short, my thesis, I just think that in general, people are going to find it more annoying to have a personal assistant order them food or order them a maneuver than

just doing it themselves with the processes they know. I think that there are very likely high-powered business-type use cases and power users for whom that sort of automation is going to be essential, but I think it's likely to be the people who already have some amount of that automation in the form of personal assistants, who then want to translate it to an AI assistant. Now, as I said right at the beginning, people tend to be very bad at predicting which technology they will use in the future, and I'm fully open to the fact that I might be wrong here. What I think is interesting, however, about this form factor, about this approach to a personal assistant type of AI, is that it doesn't ask anything of the user in terms of changing how they interact with the world. This is something that

just sits and listens as you interact the same way that you would on any other day, whether you were wearing it or not, but then has the ability to use its contextual knowledge of the wearer and what has happened to them recently to make helpful suggestions if prompted to do so. There are a lot of questions for me, as we will get into in a little bit. But in general, I think the technology that doesn't require behavior change has a better chance of being adopted than technology that does require some big behavior change. The second thing that I want to point out about Tab is that it exists. This was a demo of a working product. Now, that doesn't mean it's a production ready product. That doesn't mean that it's worked out all the kinks. That doesn't mean that Avi and the tab team are convinced

that it's ready for prime time. It's just that it exists and it was able to be demoed, which as we will see is relevant for the story. So that post went up, as I said, on October 1st. Then yesterday on October 2nd, another very similar seeming product was also announced. Dan Suroker, the founder of Rewind, announced the rewind pendant, which he described as a wearable that captures what you say and here in the real world. Now, the Rewind AI service already did this for how you interact with computers. Rewind is a product that keeps track of all the websites that you've been to, all the documents that you've created, and gives you the ability to go interact with that via artificial intelligence. The Rewind pendant then, Dan said, is Rewind powered by truly everything

you've seen, said, and heard. Summarize and ask any question using AI, private by design. The announcement also came with a landing page where you could reserve your spot, where you could pre-order for $59, and where there was more information about their approach and the use cases that they saw. They write, we started as an app for Mac, iPhone, and soon Windows, but there is so much more to life than what we experience on our devices. That's why the next step towards our vision of giving human superpowers is the Rewind Pendent. Rewindent is a wearable that captures what you say in here in the real world, and then transcribes and stores it entirely locally on your phone. With Pendant, Rewind is a personalized AI powered by truly everything you've seen, said,

or heard. What about the use cases they give? Forget what. what your spouse just asks you to pick up at a grocery store, add an all-day conference and want to share summaries with your coworkers, be more present in your day-to-day and bookmark moments for later, automatic to-do list generation whenever you verbally commit to doing something for someone else, have an exciting idea while going for a walk or driving and want to remember it for later, insights into your life, what are you doing when your voice sounds the most excited? When are you the most grumpy? What are the most common filler words? And then one that cuts right to the guts for me, capture the incredible things your kids say so you can replay them later and reminisce.

Now, a couple points from the FAQ that are also relevant. What stage of development is this project? Very, very early. Hardware is hard. If you're excited to contribute to this project, we're hiring. When can I expect delivery? Unfortunately, we can't commit to a delivery date at this time. Rewind Pendent is only available for pre-order,

and the path forward very much depends on the volume of pre-orders we receive. So basically, this suggests that the pendant does not have the same sort of working demo as the tab. Now, there is nothing all that different or unexpected about two companies in Silicon Valley working at the same product at the same time. Indeed, as you might feel after this video, this seems like a category of AI-powered personal hardware that we're likely to get hundreds and hundreds of more competitors in rather than it being some big innovation. It's a field that's much more likely to win based on quality execution than it is on having

a novel idea. However, what added a little bit of drama to the whole scenario is that in the wake of the rewind pendants page showing up and Dan's video being posted, Avi actually posted an ability to order one of the first 100 units, which would be shipping in the winter and spring of next year. The initial run would be limited to 100, and each device would cost $600. Many of Avi's friends and compatriots retweeted the post. Thomas Schult says AI wearables are coming out left and right. You only need to order one.

$600, only 100 units unlock infinite memory. Now, in what some saw as fairly bad form, as a response to Avi's pre-order tweet, Dan from Rewind dropped his post about the Rewind pendant. A few people called Dan out on this. Siki Chen wrote, you launch a marketing page a day after Avi's working demo video goes viral and hit his reply on your thread and post on his thread. I'm a paying rewind customer, but this ain't it. Others like Matt Sherman

responded directly to Dan's post on Avi's thread, writing vibes are off. I'm getting a tab. When someone accused Avi of riding Dan's announcement, he said, write his announcement? I literally posted a day before him with a live demo and audience. This guy just forced his team to make some crappy render overnight and make some lame video and called it a day. Now, ultimately, I'm I'm far less interested in the drama between tab and rewind, and much more interested in the fact that this seems to have heralded a new startup category within the AI space that I think we're likely to see a lot more of in the months to come. X user Jonevy says, Avi Schiffman posts one product launch video and suddenly 246,753 companies

are doing the same thing. We just talked yesterday about the Humane Pin as an example of this, which we know has a mini projector in camera, although we don't know if it does the same sort of listening in synthesis. Of course, then there's also the AI Phone Wars. A poster on Twitter, Arcene Lupin, dumps would appear to be a leaked video for the Pixel 8, which seems to really focus on the AI features integrated into particularly the phone part of that service. Then there is, of course, this Johnny I of OpenAI collaboration. And even non-hardware devices are certainly moving towards this customized model.

9 to 5 Google, Google, a couple days ago, wrote a post called Google Bard Ready's memory to adapt to important details about you. And it's effectively exactly what it sounds like. An update to the service that gives Google's Bard AI the ability to run. remember certain key details about the person and contextualize answers as such. But of course, I started this video from the premise that although this might be a new trend, it's a question of whether people actually are going to want it. Investor Zach Kukoff posted an image of the tab and of the humane pin and writes, guy who hates party's voice, I guess this means I'm never going to a party again. Sam Lesson said, I had one of these a decade ago called the narrative clip. I get it intellectually,

but it still isn't practically a thing anytime soon. Beware the pivot to hardware. Perhaps more saliently, Will Minitas tweets, I don't think it will ever be normal slash okay to record 100% of your conversations to play back later with perfect recall. I don't think anyone wants this at all. Yet every single AI hardware product is chasing this. He continues, there's probably a relevant Mark Fisher essay, but this definitely seems hauntology core. All of technological progress is being directed towards building perfect records of the past, perfect recall, bringing the dead back to life, character, and writing in their style, open AI. Now, Avi actually responded to that and said that's why Tab isn't a recording device.

No transcripts, the nuance and combo gets removed, only the entities and facts are stored. Not what you literally said. He then goes on to say, Rewind is just making a mess of everything. Now, many responded to Avi that this alone probably suggested that they needed a landing page to explain more about that, because that's obviously a very significantly different design choice. It also helps explain why there might be so much contentiousness with this Rewind pendant product. It's one thing for two people to go after the same market.

It's another thing for two companies to have fundamentally different belief systems. about how something should be designed that are predicated on questions of ethic and societal norms, where one party thinks that the way that the other party is doing it is not just incorrect from a business perspective, but wrong or even dangerous in a broader sense. But what about what the market thinks? Well, for one, it only took about six hours to sell all the 100 tab pre-orders. After that, Avi opened up to wait list for batch two. The rundown AI newsletter today included a survey about would you buy a rewind pendant, and there the responses were definitely more mixed. 40.2% said, yes, they would, they're excited for this technology.

22.2% said, no, I don't need it, but then 37.6% said this should be illegal. And that alone probably gives you a sense of why the stakes are a little bit heightened around this particular product. And one more small piece of evidence over the divisiveness of this comes from Justin Mickelay, who tweeted a poll, if you could live your life over and have private audio and even video recording of everything that's happened to you, everyone you've ever met, every conversation you've ever had, and every thought you've ever verbalized, along with a current-day chat GPT interface to that information, would you want that? When I found this poll, it was exactly 50-50. I retweeted it to my followers who I think were going to be much more likely to be in the hell no, I'd go insane category,

but the results are still pretty split. 52.1% said hell no, I'd go insane, but 47.9% said, yes, I'd be superhuman. I think that this particular category of AI product is going to be as debated as anything that has come so far. I think that there are huge questions for how people want to interact with the world, with how they want other people to interact with them, and I think in many ways it's a perfect encapsulation of the fact that when it comes to AI products, the questions range from business to ethics to societal and beyond. And so, friends, this is one that I will be watching closely. I would love to invite you now to come talk about this in the AI breakdown Discord.

You can find a link at bit.ly slash AI breakdown. I want to know if you're a strong hell yes or an absolutely not or somewhere in between. In any case, as the discussion evolves, I'm sure you will hear about it here. But until next time, peace.