Did The AI Bubble Just Pop?, Apple Intelligence Beta Arrives, Friend Dot Com

28:05I'm very, very sorry to say around Sean. Oh no. Siri still sucks. It's terrible. It's so bad. It's so bad. All right, please go tell me what you've learned. Here's the caveat. And I said this on air yesterday, so I'm going to just say it again for our listeners perspectives. This is not the final product. Who knows what they've shipped underneath it. I'm on Apple's iOS 18.1. and that gives me access to the beta of Apple intelligence. And this is not going to really come out till fall, but I can say that they have a long way to go from here to where they're trying to get because if this is the final product, and we know it isn't, but if this is anything close to the final product, be prepared for extreme disappointment with Apple.

28:53And I'll just give you a couple examples of what I try to have Siri do. i asked it to show pictures of my wife and it showed pictures of a celebrity with the same first name okay that's what it did wait that's in apple photos or no it went to the internet so it wouldn't even navigate through the photos on my phone it went to the internet and showed me a picture of this celebrity oh my god and then it would not be it was not able to process process directions it could not answer basic questions the one thing i thought was cool was it was it helped me text somebody like i had it say okay i'm you know i was going to get a haircut yesterday so i had it like pull up my barber's cell phone and i dictated a message to it and hit send but like wait but we could already do that yeah like if that's intelligence i don't really know where we're heading there's also this other cool feature that actually i'm really excited about as a podcaster where you can record calls and it will transcribe for you and so i made a call i hit record the recording worked but the transcription did not work like it's supposed to transcribe the notes and that wasn't working either did that was then did it even have any transcription and the transcription was bad or no it just said this isn't working at this time there was an error so apple intelligence from my first look is unintelligent i i do i often wonder around how a company like apple from a product marketing standpoint thinks about these beta releases because i mean maybe it's just us and our listeners who care and like from a larger market perspective you know no one is really paying attention so it actually does help them work out the kinks and get to a better product.

30:47But I have to imagine like after the presentations and demos that they did, even the first release, they would understand that it has to be pretty good. So that does have me worried a little bit, but. Yeah. And look, don't take my word for it because here's Ethan Malik, the Wharton professor and AI expert. He says on Siri, because it's a small model, it also isn't that smart. In fact, it feels like using the old Siri with minor improvements. If I ask it, I want to go to dinner and a movie tonight and make sure I can get there by six and home at 10, I would love some spicy Latin food and an action movie.

31:24It fails miserably. This is not actually a hard problem for LLMs though. A slightly larger model, Llama 8B, actually does a much better job. And look, this is the, if you want to be hopeful, he says, this is just the start for Apple AI because future update will make it so that Siri on your phone can ask a larger Apple AI in the cloud to help when it can't solve a problem or even pass really hard questions to ChatGPT and it will be able to interact with apps, triggering actions and taking in information from multiple sources, the technology will certainly improve. And I think this is the most important thing he says.

31:59He says, AI carries risks. It is unpredictable. It hallucinates. It has the potential for misuse. It's not always private. So Apple decided to reduce the danger of misuse or error. They have turned Siri into a co-pilot. You can see these sort of co-pilots appearing in many products, very narrow AI systems designed to help with specific tasks. In doing so, they hide the weirder, riskier, and more powerful aspects of large language models. Co-pilots can be helpful, but they are unlikely to lead to leaps in productivity or change the way we work because they are constrained. Power trades off with security.

32:36Power trades off with security. Yep. I think this is important, right? It's basically like, I think what Malik is saying is Apple is effectively not quite ready to be weird enough to reap the full benefit of generative AI. And that's why you're seeing these half measures. And my prediction is if Apple does not get bolder with this technology and leaves it in co-pilot versus pilot mode, to use Malik's analogy, that we are going to see blow back to them. Because remember, that Apple intelligence demo, it blew people away. The stock flew up the next day after it happened. And people were talking about how Apple has just made the masterful Gen AI play.

33:20Ben Thompson had an article about it. If they don't deliver, if their demo and their dream of this technology, which was always going to be good, think about the Vision Pro, doesn't live up to the promise in reality, I think this is going to be a very bad moment for Apple because unlike mixed reality, AI is a category that's moving and moving fast and has real, real application already. And you can't swing and miss there. Wait, did Mollick's post use the phrase like co-pilot versus pilot, that idea? Or did you just come up with that? I just built on what he was doing. He talked about co-piloting.

34:00Yeah, no, no, but I think, ooh, I think this is a, we can run with this one a bit. I think this might require a post here because, because I think that's a really, really good distinction because again, this idea of co-pilot is something that's been introduced in a smart way to kind of remove the, the responsibility from the model. And it's fair. Like it's, especially in a lot of use cases starting, I mean, GitHub branded, it's a coding assistant, co-pilot, and really creating this branding and idea that it's something that's there to help you that doesn't make decisions, but actually just is there to add a little bit of assisting, but is not all knowing and responsible.

34:41But the way Apple demoed intelligence was a pilot, it was like, and the whole idea of agentic AI is that it actually makes decisions and takes actions using different systems based on just a prompt, just initial prompt, it should be smart enough and have that element of reasoning that will know which system to call and what decision to make. And that's a pilot. And that's a much tougher problem to solve. And we talked about this, it always was going to be harder the way they marketed this. And it's even when Apple advertises, if you say like, what's my flight status without specifying the flight, it goes into your email, looks for a flight, then knows to use that information to call flighttracker.com or whatever other API service to actually get that information.

35:34Those are actual decisions being made. And that's the act of a pilot. And if Apple, and it was always going to be a harder problem to solve. So that could be a big worry for them right now. This is unfair. But I am flying to Ireland on Sunday. So let me see if I can get Siri to. Oh, don't even. Don't do it on air. I might just slam my computer shut. All right, try it. Try it. Try it. Hey, Siri. Pull up the flight information for my Sunday flight to Ireland. Oh, okay. it gave me dublin it gave me a google search dublin arrivals and departures live flight information and then find your trip on american airlines i'm flying jet blue i we might have to end this right now and not even get to friend.com two tiger sugars coming up 100 sugar please i I mean, that from like a system standpoint should be the easiest problem to solve.

36:45My, like you were almost like throwing it a layup with how specific that prompt was. Like my flight on Sunday to Dublin. That, that's like an email search. That's not even a general, that doesn't require an LLM almost. And it's just not getting it. Doesn't know what to do. So that's disappointing. But they have until, let's say, August. Or no, it is August. They have until, let's say, October to figure it out. So I hope they're figuring it out. It's two months. Two months. You guys got it. Now, M.G. Siegler has some good context. He says that if you think about the Apple approach, it's intentional.

37:20And Meta is shoving AI into the faces of hundreds of millions, if not billions, of users across a range of products. So is Google and other ways and other products. Microsoft is trying to, if anyone will use their products. and OpenAI is trying like crazy to build up a user base of such sizes, including with the help of Apple, as is every other startup in the space, so much so that EU regulators appear to be actively working this summer so as not to miss an opportunity to mess it all up. And he says, oh, over there, it's Apple just casually laying back on a raft, drink in hand, floating down the AI stream, ready to launch, stuff people will actually use when it's ready and not before.

38:00And that's what Apple does. Yeah, I think, I mean, and that is the product rollout strategy that we have seen. I mean, even that comment around like, I mean, right now, this idea of voice as an agent, and in our document, you had another from Ethan Malik around the idea that like using voice to now become the interface for generative AI is interesting to me. And again, Siri should have this as a possibility because to me, generative AI, even whether using Claude or ChatGPT or any kind of chat interface, when you start talking to it in a conversational manner is when it gets good. For anyone who's like used these tools a lot, like really, and it's funny because even like friends and coworkers, when they see how I'm using chat, sometimes it's almost weird to them and unnatural.

38:51And I'm like, oh, that was pretty good, but you got this wrong. Could you actually try to fix this? Or, oh, great answer. Can we do more of this? Can you dig a little bit more into this? It's weird. But in written form to an AI, people do find that weird. But maybe voice, it just becomes much more natural because that's how we all talk. So maybe that will be a saving grace and maybe Siri will work. Yeah. Well, it has some catching up to do because this is the other thing that Ethan Malik wrote about which is the open ai's chat gpt 4.0 voice edition has come out in limited beta now i haven't gotten access to this but i have seen some incredible demos on the internet including someone asking it to count to like 10 as fast as it could and it goes like one two three four five six hundred nine ten and then it says okay now do it louder and faster and count to 50 and it gets to 30 and it goes 31 32 33 and it like takes a breath it's crazy and this is what mollix has about ChatGPT 4.0, voice edition.

39:54He says, if Siri is about making AI less weird and more predictable, ChatGPT voice is the exact opposite. It does not use a small tailored model, but rather provides access to the full power of the generalist GPT 4.0. While there has been a kind of voice mode available for ChatGPT for months, this is very different. It engages in natural conversations with interruptions and fast flow. And he says,

40:49features are locked behind guardrails but those are the capabilities that it has i can't wait to try this thing and we give open ai a hard time here but i do think that you know talking about them being in the hits business this looks like looks like it's going to be another hit no i i think i mean they had what was interesting about when they launched voice and the whole scarlett Johansson brouhaha was they have had voice available on the chat GPT's iOS app for a long time and it's worked pretty well. And again, the idea before it was more like one prompt of response and a prompt in response, but obviously the demos are showing it's even more human-like and more natural and more back and forth conversational that you could even interrupt it and uh like it'll know that you are interrupting it and take the new prompt as feedback so i think it's yeah i think the whole voice area and they they might get their next hit what it means for their business we'll find out especially if the ai bubble is collapsing i think open ai has got some bigger issues than most but let me ask you this is this potentially their biggest hit of all time no i i don't think open ai will ever have a hit in the way that chat gpt was i think too quickly even i think like everyone i talk to now is switching to clod like in terms of where you started that by the way yeah i think you led the clod revolution you taught me about it i became a clodhead clod and the rest of the world became clodheads clodheads unite.

42:29It's funny. Most people I talk to from a individual consumer standpoint, it's like everyone has a$20 budget that they will allocate to one chat bot at that moment. So it's like, Oh, I'm doing chat GPT plus, uh, I'm going to go Claude pro. Um, maybe I'm a perplexity pro, but everyone seems to just, I allocate that this is my 20 bucks. I'm going to choose one that I find to be the best and go there which is again for these businesses a reminder of how cutthroat and competitive the consumer part of this whole equation is and how difficult it's going to be absolutely so oh yeah let's go let's go speaking of voice uh let's go to friend.com difficulty let's go right to friend.com friend.com how do i describe it let's just let wire describe it It says, wear this AI friend around your neck.

43:26And so friend is a wearable. And Wired says it's a pal, a buddy, but mostly an AI chatbot that lives inside a pendant that's around your neck. Or around like a necklace thing around your neck. It always has an opinion to share about what's going on around it, which it communicates using text messages and push notifications on the phone it's paired to. Always listening is one of the main taglines of the yet unreleased AI device. The friend has an onboard microphone that listens to everything happening around the wearer by default. You can tap and hold it to ask a question, but sometimes it will send messages, commentary about the conversation you just had, for example, unprompted.

44:08It's powered, and here we go, by Anthropic AI's Claw 3.5, which can engage in helpful conversation, offer encouragement, or rib you about being bad at a video game. And this thing became, I don't even know, I couldn't tell this week, was it an object of mockery or was it an object of true fascination? To me, it seems ridiculous to have an AI friend around your neck that's conversing with you. It just screams like I'm a loser. I don't know if you wear it, but I'm curious what you thought about the launch of friend.com. And is this, is this really, do we call the top on the AI moment now, or is this just the beginning of something big?

44:49All right. So friend, well, there's friend.com released this video. And I think the video, sometimes you, I honestly wonder, it was so cringeworthy and bad that it almost, I mean, you had to watch it. it was so fascinating again it's people in social situations turning to look down at this pendant around their neck even one of them is a girl on a date and the guy is uh is she tells the guy they're on a rooftop and she says i never bring anyone up here except for her and looks down at the pendant around her neck and then the guy says like oh like i something around oh i guess i should be grateful or like, I appreciate that.

45:32It's so weird and black mirror-y that I do wonder, are people that out of touch who made this or do they just want to go viral and bring attention? Cause otherwise none of us would have ever had heard of friend.com and it's pure earned media. So from a customer acquisition standpoint, pretty cost-effective. Um, so does it brings attention to it. But my favorite part of this story is to make this more ridiculous, if a AI friend pendant on a necklace around your neck all day long is not weird enough, there was actually two competing like wearable AI startups called Friend. And one of the founders bought friend.com and spent 1.8 out of their$2.5 million seed round on the domain.

46:20And, you know, reportedly to kind of secure any kind of copyright potential issues. If you own friend.com, you're going to have more of a claim. The other founder responded with a rap video, a rap battle video to the tune of They Not Like Us from Kendrick. Of course that was it. Yeah. And it was just, I mean, it looked like an episode of Silicon Valley. It was amazing to watch. It kind of made me want to get back to common sense and AI. And maybe this week, the fallout in the market will help on that. But it was a fun week for AI pins everywhere. Right. And we should note that the one that got the friend.com domain is the one with the AI pendant that we were just talking about.

47:11Clearly, they're good at marketing. I mean, with the video and the name, counts for something. And CEO apparently said, people just don't get consumer. I view this as saving money, much less money needs to be spent on marketing. It's a one-time thing. Okay, I'm just doing the math though, and 2.5 million total raise, 1.8 million on the domain. You're going to get a hardware AI company off the ground for 700 grand. I do not understand how that works. The thing that's tough about this is I want, and we talked about this with Humane as well, like the idea of creative new hardware form factors that leverage AI, I really want to see.

47:54I think that would be cool. I think like hardware innovation is something that we have not seen in a long, long time. And the idea that instead of everything having to live in your phone bundled by Apple, or whatever other big tech company to see cool new little hardware devices come up and just create just pretty singular experiences leveraging AI, I really want to see happen. I'm excited about and I think would be great. It's just sad that everything seems to be just a complete whiff. Right. No. So for the AI friend pendant, okay. So this is one plausible explanation, which is that they basically raised their series A hoping to market and raise their series B and just put all this money into the marketing and they weren't really planning to build a company with that.

48:44It's more of like the proof of concept and see if you can raise again. And this is from someone named Andrew Rhea, who's an entrepreneur. He was with On Deck, one of those guys. He says his take on the friend.com launch. This was a great go big or go home bet. The product vision is bold different. Maybe it doesn't work, but it's certainly controversial and controversy is great for marketing. consumer is so hard you have to do something this bold to have a shot at breaking through the noise the launch was beautifully executed the 1.8 million domain was a great red herring clickbait for people that wanted to hate they will almost certainly raise a large round off of this hype obviously this person has a the founder has a gift for building movements or moments which is rare Okay.

49:30Is that the playbook? Three things on that. First, if that is the case, and I had made that comment a little while ago that earned from an earned media standpoint, like any marketing department would actually call it a success. If this is the future of how business works, that is the most depressing thing in the world for me. But second, it is kind of the Kardashian model, if you think about it, like first, and which a lot of other people first be an influencer. first build the audience and then build the product. So from that standpoint, again, it sadly and depressingly could almost be right and correct.

50:09But then third, the idea of the way he says, like, you know, now they'll definitely raise another round off of this hype. I think that's exactly the kind of thing that pre-Goldman report would have been a definite thing to happen. in a post-Goldman generative AI report world, maybe is going to be a bit tougher. For sure. Okay, so as we ran out this episode, I think it's a good moment to talk about the Olympics. Have you been watching the Olympics, following them at all? Of course. I'm a huge Olympics guy. So we just have a couple minutes left. I think we should just trade some favorite Olympics moments.

50:47Do you have any? Opening ceremonies, their performance by Gojira, the heavy metal band that started with a headless Marie Antoinette. And it was funny because, again, controversy is a good thing. It was considered a bit controversial, but it was the most epic thing I think I've ever seen on TV. They took over Il de la Cite, which is like the prison. It was a castle and then the prison during the French Revolution and just heavy metal just roaring through an entire castle on the river. like uh it was that from a sports aside that was my favorite part of the opening ceremonies yeah did you watch it or i did watch parts of it last week i also for me my favorite moment at this point has been the turkish uh marksman yeah who just kind of casually strolled up to the target blasted his way to a silver medal and then was just like i don't need equipment i'm a natural shooter i thought that was amazing the memes have been incredible too it's funny how the Korean women's shooter with the eyewear just looking badass was great as well.

51:58That event had a couple of memes. Yeah, yeah. But I mean, from the actual sports standpoint, I think swimming is probably my favorite thing to watch at the Olympics. And I like that it feels, I don't know if this is correct, but they're moving more metal events further up. Before it felt like you always had to wait even for swimming and everything else. at least like a week into it. I think they were doing a lot of the qualifiers before the opening ceremony so they could get right into the medals. And I think that was a smart decision. Can I also say there was another great moment where they had this like regular dude dive into the pool and pick up a swim cap that had fallen into the pool where the swimmers are supposed to swim in.

52:43And the memes on that were amazing. This guy is just like in his little speedo. And he's like, I'm just as puzzled as you are that I'm about to go in this pool. And people were like, this is VCs showing founders how to operate. Like, I think, yeah, no, no. I said, I saw, I think a bunch of versions of that one and the shooter memes. Like, I think overall, this is probably the most memed Olympics. I think I can remember at least where, and one thing I remember on social media, I think the 2016 like you couldn't they wouldn't allow video sharing so like and anytime anyone shared a clip of a video oh god you would get you'd get the copyright owner disabled no i mean as a reminder that like to make this stuff relevant like hyper relevant in everyone's mind because i mean my all my feeds are just dominated with this stuff people at work everyone's just talking and sharing the memes.

53:42It's pretty great. Can I share a couple more? Bring the world together for good memes. I agree. I love that the South Sudan basketball, which is apparently funded by Luol Deng, is just going out there and kicking some ass. Oh, I love that story. It's an amazing story. It's like a Sweet 16 style story. It's great. No, they did not have an indoor basketball court in South Sudan when they started this effort. And I mean, before the Olympics, I think it was a friendly or was it a qualifier? I think it was a friendly against the US. They almost beat the US. Yeah, I saw the notification on my phone.

54:17I'm like, what the hell is going on? That is a feel-good story. That's like the Olympics story. I think that could be the probably most heartwarming story from the whole thing. And it's crazy because so apparently South Sudan and Sudan had a civil war before South Sudan separated. And the Olympics played the Sudanese national anthem. and you see the players being like, what the? Oh, wait, I missed that one. I missed that one. Trying to think of, okay, so let me just share then. Finally, my favorite story of the whole Olympics, which is that, did you hear about this? The Italian flag bearer, his name is Gianmarco Tamberi.

54:59He was carrying the flag in the opening ceremony and it popped his wedding ring right off his finger and into the Seine River. Oh, I remember hearing that. Yeah. Explaining that one as you go to the Olympic Village on those cardboard beds is going to be a tough one for my man. Yes. Yeah. So he said that he was trying to carry the flag as high as possible and that his ring will remain forever in the riverbed of the city of love. And then he tells his wife, if you want, we will throw yours into that river so that they will be able to be together forever. And we will have one more excuse to, as you have always asked me, renew our promises and get married again.

55:44That's what they all say, John Marco. It fell in the sand when I was carrying... A car on ton reaction to a beautiful story. You want the trough, man. Hey, it fell in the sand when I was carrying the Olympic River. I'm sorry. When I was in Paris at the Olympic Village with all the thousands of other athletes. I'm sorry. Oh, my God. Well, I will still take it as a heartwarming story into my weekend. Shape up, John Marco. Shape up. Come on. Look at the John Marco is doing all he can, carrying the flag. He posted about it, offered to get married again. But yeah. Come on. It will remain forever in the riverbed of the city of love.

56:25That is a man scrambling right there. Unbelievable. All right, John. You broke my heart about Siri, so I will... You really got me bad. I'll come back at you with... You said we weren't going to be cool-headed today, and you delivered. We were not. Vicious war between the two of us. All right. So next week, we have Nick Bostrom coming on the show. It's an interview I've teased for a while, but he's the AI doomsday philosopher that said, basically, that there's a chance we'll be living in utopia, and we get real weird at the end talking about simulation theory and whether we converge with AI and rip our way out of whatever simulation the simulation masters have us doing.

57:09So stay tuned for that. And then plenty of other great shows coming up this August. Ranjan and I will be back, of course, next Friday, breaking down the week's news. Thanks for coming on this week, Ranjan. Great speaking with you until that last segment. John Marco, shape up. Unbelievable, John Marco. We'll see you next time. on Big Technology Podcast.