Nov 3, 2024 ยท 2024 #39. Read the transcript grouped by speaker, inspect word-level timecodes, and optionally turn subtitles on for direct video playback
Edit labels for this show, save them in this browser, or download a JSON override for the production folder.
Transcript Playback
Disrupt Edition
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Speaker 2
Hello everybody, it is Sunday, November the 3rd, 2024. First Sunday in November, two days off what many believe will be the most disruptive American election in history. And when it comes to technology, for our That Was The Week show, we're focusing also on disruption. Keith Teer, my... a friend who is our eyes and ears on Silicon Valley, was at TechCrunch's Disrupt event this week in San Francisco at the Moscone Center. And the focus of his editorial, and indeed the That Was The Week newsletter, is indeed disruption. So, Keith... I know you and I, in the old days, we used to go to Disrupt when it was a rather small event. These days, it's enormous. I mean, I saw that they're getting massive stars, huge amounts of people. Was it exciting or was it just another blah, blah media event?
You know, interesting question was exciting. I don't think you'd call it exciting because it's very much taking the pulse of what's going on in the industry and They do have a startup competition that runs through it with a final and a winner. So that's kind of exciting. But the content itself is very much taking the pulse. So there's a lot of AI stuff, a lot of health related stuff. Big surprise. A lot of venture stuff. It's held in the Moscone Center in San Francisco, which has three huge floors that they take over in Moscone West, it's called. And they have about seven stages, all of which are pretty full with people watching. I think that draws around 10,000 attendees.
Are those 10,000 people, are they all kind of wannabe people who want to get into the tech business? Or is it 10,000 leading influencers, 10,000 of the most disruptive people in the world, at least in tech?
He wasn't. I was there as the founding shareholder.
Words and timings
Hewasn't.Iwasthereasthefoundingshareholder.
Speaker 2
You were there on behalf of Mike Arrington, although I'm sure... Not really. Mike would never give me that honor. Yeah, I mean, it... I remember in the old days, didn't he start Disrupt with Jason Calacanis? And then, as always with Mike Arrington, they had a huge row and a big split. So much has changed. What year was Disrupt founded?
Was it 2007 or 8? The history is interesting. It started in 2005 in Mike's backyard in Atherton. Menno Park, which was an acre ranch. So we used to have these bonfire parties with burgers.
It started in Mike's garden with a kind of a wiki where you signed up. And by the time the third one of those happened, he had a huge tent and almost 1,000 people showed up in the garden. And then it moved to an event which was called TechCrunch 40 because there were 40 startups that were profiled. And then it became TechCrunch 50. And then it became TechCrunch Disrupt. So I'm going to guess you're looking at maybe four, three, four years in. So that would be 2008.
No, Calacanis was hired as a, as a kind of a, and paid as a, as a kind of a stage compare along with Mike. And what happened is when, when they, you know, 2A type personalities, it didn't really work very well. Calacanis.
Yeah. I mean, the history of Calacanis and, Mike Arrington is also an interesting history. But you work off this theme in your editorial. You believe that we are indeed living in an age, maybe disrupt, TechCrunch disrupt is no longer particularly disruptive, but you believe we're living in an age of profound... You don't write about the election. Maybe we'll talk about that later in the show. Why did going to disrupt this year, Keith, convince you that we live in such a profoundly disruptive age, even if the event itself isn't particularly disruptive?
I don't think going to disrupt persuaded me. It's this week's news that persuaded me. If you look at this week's news, and by the way, hi to David Birch, who just texted us to say he had dinner with an old friend of mine last night. But the news this week is super disruptive. Everyone is making moves, everyone. So OpenAI launched... a new Mac client along with its iOS client that can also do web search.
But why is it different this week from any other week? No, it isn't. It's like Chinese water torture. It's drip, drip, drip, drip, drip. And eventually the drips become heavy. And that's kind of what's happened this week.
Oh, this is supposed to be a family show, Keith. We don't want to talk about Chinese water torture. You talk about in your editorial five points about why we live in such a disruptive age. Most of it deals with AI.
AI development, you argue, is going to continue to accelerate, making traditional software development increasingly obsolete. And companies are also going to deeply integrate AI into their core operations like Meta. Meta features in one of the pieces of the week in Stratechery, which you always like, about Meta's AI abundance. Why does Meta have more AI abundance than anyone else?
His point is that Facebook now has 3.2 billion people that engage with it monthly. And that, therefore, Meta's AI, which is... not quite as good as OpenAI is, but it's good, is going to reach more people than anyone else is, and therefore have a greater impact than anyone else. And I think he then draws the conclusion that Meta is, you know, the most important AI company. I disagree with him on that.
I think OpenAI is... Well, also, they have more of their own content. I mean, they've got the the training wheels or whatever term people use. I mean, if they've got whatever it is, 3.5 billion people on the Facebook platforms, then those people are producing content which can be put into Meta's AI. Is that fair?
It is, but it doesn't seem to make much difference. I mean, one of the things I did this week is downloaded a tool called LM Studio, which lets you run the meta AI models locally on your Mac. And you get your own, if you will, you get your own kind of open AI interface in your house. And it It just isn't as good, to be honest, as Claude or OpenAI. I use it for many engineering tasks, and it was really bad. I tried to use it, and it just wasn't good enough.
No, I do agree with that. But I think he's missing the bigger picture, which is, You know, Apple putting OpenAI into iOS and Google putting Gemini into Android means most of those three and a half billion people are going to have direct access to AI without going on Facebook.
It's going to be interesting to see how it plays out. You've always been very, very bullish on OpenAI. This was an interesting week where OpenAI, and this is according to TechCrunch, launched its Google Challenger chat GPT search. You've always believed that OpenAI was kind of an alternative to search anyway. So why is chat GPT search any different?
It basically marries a trained AI to current information available via a web search. So I've got it open here. And if I do a web search, you know, who is Andrew Keen and what is Keen on?
It, it, it, you know, Andrew Keen is a British American entrepreneur, author, commentator, known for his critical perspectives on the internet, digital culture, born around 1916, Hampstead, London, studied at University of London and political science at Berkeley.
Yeah, you learn something every day, Keith. But in all seriousness, do you think that this move by OpenAI to launch formally a Google Challenger, is this for real or is it just another way that OpenAI is trying to tweak Google's nose?
I think it's for real, Andrew. What it's going to do, it's going to take what currently is a cloud-delivered service, Google Web Search, and push it closer and closer to us individuals on our devices. The fact that I can now run a huge trained model on my Mac here at home, I could also run models on my iPhone, means that we're living at a time when distributed intelligence renders the cloud irrelevant. But why would you want to render the cloud irrelevant? Well, because it's better. It's better. You know, 3 billion smartphones computationally is very powerful. And I think that it's not fully being understood.
No, the models are trained. And then, you know, just to give you a sense of it, the Lama model that Meta built is a download of about five gigabytes. So it can run on an iPhone. But it was trained in the cloud. But that training is private. It's kind of behind the scenes, what Facebook does privately. Once it's trained the model, the actual model shrinks down to a consumable size.
Is this, though, a sort of cannibalization of these companies? Why would Facebook or OpenAI or Google, why would they want People to work essentially offline, away from...
They don't, but they have no choice because open AI has no legacy in the cloud to preserve. So it can push this intelligence out to us without any negative consequences for itself. So Google and Microsoft and the others are forced to follow suit. And that's the innovator's dilemma. Part of this, they're going to have to give up some of their previous gains because we users aren't going to use them.
It's interesting. It's certainly the innovator's dilemma. I'm not sure if you had in your news of the week, the Google numbers, which were very good. And the The best part of Google's numbers for this quarter were their cloud numbers. So Google Search and YouTube, actually, the numbers weren't particularly impressive. So what does that mean for Google, that those cloud numbers you're suggesting in the long run don't mean anything because we're moving away from the cloud, a post-cloud digital economy?
The old page rank. Because once you have AIs that consume and learn and know, the more reliable they get, the less a dated index is going to make any sense. And so by definition, you know, you're moving to a more real-time, interactive kind of knowledge transfer as opposed to type a keyword,
get some results. Excuse the, maybe the vulgarity of this question. When Google announced that their numbers this week, and they were very, very strong in cloud services. And at the same time, the search numbers were less impressive.
Well, the cloud is a word that disguises a lot of different business models, actually. But the developer cloud where Google makes tools available for developers, including trained models, that is not going away. Things like Copilot and GitHub that Microsoft owns isn't going away. But what is going away is the architecture where services are built in the cloud and made available through a web interface. That is probably going away. Not quickly. I mean, it's a bit like radio and video. It's not going to absolutely go away. But the preponderance of use cases that can use AI is going to grow. And as that happens, less and less of these cases will survive in their current form in the cloud.
So Google's, so do we have a word for this post-cloud economy? How would you describe it? Is there a single word or it hasn't emerged yet, so to speak, out of the cloud?
One of the things I go away is the value of, computational power. I mean, the arms race in big tech now is in terms of tooling up for computational power. That's why to launch an AI platform requires billions of dollars.
The M4 chip, the M4 Max, can run any AI that we currently have. So the idea that you need cloud computational power for running AI is not true. So, OK, so we're going to jump.
We usually do post of the week at the end. But let's go to that because your post of the week is from JF Martin. Why the M4 Mac Mini will be my best Mac ever. And it sounds like the classic Apple news story is that the latest Mac, desktop computer or phone or watch or laptop is the best ever. Why should this make any difference to ordinary people? How are people going to use this new M4 Mac Mini in any different way from they use their current computer?
Well, it's going to be open to lots of use cases that the old ones can't. For example... Just to use a little example, its ability to run models, it is fast enough that you can have real-time conversations, voice conversations with an AI that sits on this Mac mini.
So you can't do it on a traditional Apple MacBook Pro?
Words and timings
Soyoucan'tdoitonatraditionalAppleMacBookPro?
Speaker 1
It would be too slow. You can do some things on the MacBook Pro. I've got a Mac Pro, which is very overpowered, but it's running on an M2 chip, and it can't do some things that that M4 chip can do, which is much cheaper. The other thing, Andrew, is just to answer your previous question, one more thing. The input-output It has Thunderbolt 5, and Thunderbolt 5 can operate at 120 gigabits per second. Now, an entire AI model is only five gigabits.
Now, I take your point, Keith, but you spent this week at TechCrunch Disrupt. Maybe you're going a bit geeky on me. I mean, you're a hardcore geek, but most of us aren't. I mean, is there really a market now for this type of use for ordinary people of the M4 Mac? I mean, maybe it's a big deal in the long run, but for 2024 or 2025, people just are going to use their computers as they've always done.
Well, I'm not really focused on the end user yet because I think the end user, you're right, will be slow to change behavior. I'm more focused on what developers will do with it that will then give to end users very simple tools. We talked a few weeks ago, for example, about how the classroom in schools might change because of AI. This is now $600, this device. You put one of those in every classroom and developers deliver to it the curriculum appropriate to that classroom with an agentic teacher capable of teaching the subject. And it has huge transformational ability at very low cost. Why can't that be done now on the computers? Because a MacBook Pro costs $4,000, not $500.
You also talk about healthcare and other traditional industries will see massive disruption. That ties in with Disrupt, the startup of the week, was Salva Health that also won Disrupt 2024. Is Salva Health an example of how healthcare is going to be dramatically changed by AI?
Yeah, in lots of different ways. What they've produced is... a device that can do a mammary scan and detect breast cancer at very, very high quality and very low cost linked to results that come on an iPhone.
So I have to go to a hospital to get an MRI for this now.
Words and timings
SoIhavetogotoahospitaltogetanMRIforthisnow.
Speaker 1
No, this is going to be cheap enough and easy enough that any doctor surgery can have one. And so that's why they won. And the AI is both in the diagnosis and the equipment.
I don't know. The answer to that question depends on software developers. Basically, the hardware and the models are now in exist. The models are free. and the hardware is cheap. So now software developers have to build actual use cases to leverage that. And Salva Health is one of those. It's a use case built on top of largely free and inexpensive hardware, free models and inexpensive hardware. So that's gonna be the trend. By the way, even the training is getting cheaper and cheaper. You're getting better models with more parameters that can be trained at lower cost. And the semiconductor industry is also stepping up. You're getting competitors to Nvidia now that have better performance at lower price points. So we're in a whole wave of reinventing the meaning of tech.
Let's end with the looming election. As I said at the beginning, some people see it as the most disruptive election in history. One of the pieces you feature in this week's newsletter, Jeff Bezos' note that Americans don't trust the news media. Meanwhile, For all these advances in AI, no one has any idea who's going to win the election. The polls are so close. Some people believe that the pollsters have basically chickened out by coming out with these polls that suggest that the election is essentially tied. What do you make of... I don't want to make this into a political podcast, but what do you make of the fact that on the one hand, Technology is clearly about to fundamentally disrupt the world again or continue this radical disruption. And on the other hand, it doesn't seem to be helping us when it comes to politics. America is as divided as ever, and we have no idea who's going to win the election. You talk about Salva Health figuring out whether or not women have breast cancer, but none of these posters have any idea of how America is going to vote.
Yeah, politics is the last refuge of the analog world in a way. And the way these polls are conducted is somebody calls you on your phone and talks to you.
which kind of tells you that you can't really trust these polls because they're biased towards whoever answers the phone, which is unlikely to be a representative sample of normal people. So I don't trust any of these polls. I also think the media, Jeff Bezos' point, if you go back to the Vietnam War when the media held government accountable or even Watergate where they did it again, You know, you eventually get to the Iraq War where the media doesn't really question the weapons of mass destruction story and becomes more sycophantic. And then it becomes divided into two camps. But that's been going on for ages. Well, so you get to a point where the media is not trusted and politics is...
really a measure of our trust of institutions, which is, I think, at an all-time low. And so tech could transform that. I mean, how easy would it be to just vote on your phone and see the results immediately? It would be super easy as long as an identity is a solved problem now, so you could stop people voting twice.
Trump people might not trust it on our phone because they'll argue that Apple or Google or OpenAI are a pro Democrat. Meanwhile, there was an interest. You didn't have it in this week's newsletter. There was a BBC headline. from today, and I'm quoting, Gabriella is an undecided voter. Here's the very different content TikTok and X showed her. How do you think this election is going to be remembered? We always remember elections as the blogging election or the internet election or the social media election. How is technology shaping the November 2024 election?
The most obvious way is the use of podcasts to replace mainstream media. Joe Rogan getting Trump on, and Kamala Harris has done a whole bunch of them. And, you know, the rise of the bulwark and lots of other YouTube channels that have...
We always hear that, but the big news today is that Kamala Harris was on... SNL Live, which is anything but a podcast. It's a traditional television show.
Yeah, but I don't think today is trend. That's not the trend, is it? I mean, if you're looking at the trend of the last few weeks, it's her absence from...
The election reported... the truth or the reality in two profoundly different ways. Is this the first election where the media is essentially fragmented in terms of just reporting on what actually happened?
I think that was already true for the last two elections as well. I think that's been a trend. I think I think there's two questions here. What do I think? And then what will historians conclude? I think historians will conclude that America is the most divided it's ever been. And depending on the result, pollsters will be either ridiculed or reinforced. I suspect they might be ridiculed. What I think is this election is the third election... that is taking place in the context of america's self-awareness of its own challenges you could call it america's decline like trump does but i think america isn't really in any absolute decline but it is in a relative you know self-awareness crisis well it's in a transformational moment it's in a transformational moment and in this transformational moment Some people hate the change. I think both sides actually hate the fact that that's true. Trump leverages it, stirring up anti-immigrant sentiment. Kamala Harris hates it for different reasons, but deals with it better by, at least she has, you know, clearer gender on women's rights for example she is fairly pro-immigrant so she isn't going for what trump is saying she's easily the more human of the two candidates trump represents the anger of america uh kamala harris represents probably the i don't know lack of clarity about how serious america's near future
is but i wonder Keith, whether the reason why there's so much personal insult and antipathy is actually that they're not that divided on policy. They pretty much agree on most things. I don't think that's... I mean, with China, they're both sympathetic to a stronger state. The old sort of neoliberal progressive division between the two parties has gone away.
I think I see the differences more than the commonness, but I do agree that there's a large crossover in the middle of maybe 80% where there's not much disagreement. But I think when it comes to, is America's future a policeman of the world, leader of NATO, with Russia and China as the enemy's future, which is the Biden-Harris kind of view? Or is America's future less global, which is the Trump point of view? Deep within that are probably the debates of the next 10 years.
And finally, Keith, we will know who wins the election this time next week when we'll meet again. Should the tech community really care? You were at TechCrunch Disrupt, 10,000 people, everyone excited about their new businesses, huge amount of energy, investment, money, excitement about AI in particular. That's not going to change next week. Whoever wins, should the tech community care or are these parallel realities, the hysteria over politics, the dynamic nature of our tech economy?
I think they shouldn't care. I mean, I was looking this morning at what's likely to happen with the House and the Senate. It looks as if the Republicans are going to be two seats to the good in the Senate. And it looks as if the Democrat and Republicans are going to be pretty much tied in the House. It could go either way. So, you know, if Trump wins the election, that's kind of the most dangerous because they could win both houses as well. If Kamala wins, she definitely is not going to win both houses, according to the polls. And that probably means not much will happen because there'll be a standoff. And I think tech probably works best when it isn't interfered with. We've talked a lot about Lina Khan.
Your friend, who we haven't talked about. We'll have to do a show on Lina Khan. So in other words, everything's changing dramatically in tech, but it's likely the election isn't going to change anything. You still have stalemate, balance of power. This time next week, key things will still be the same.
Well, we will have a clue next week. We will know, or we might know. God knows. whether there'll be violence on the streets or disputes. But we will see you next week. Have a safe week, Keith. And we will talk again next week. That was the week for technology, a disruptive week. It's going to be a very disruptive week this week, but a political one. We will see its outcome this time next week. So keep well, Keith, and we'll talk again in a week. Will do. Thanks, Andrew.