First off, a quick apology — I’ve been under the weather. Most of last week was spent offline, recovering from what was likely a mild case of covid. I tested negative twice, but with the virus constantly changing, who knows what that really means.
Anyway, normal service should now be resumed… fingers crossed anyway.
1. No, Google hasn’t proved that AI is fine for the environment now
This week Google released a study which “proved” that the water and power usage of Gemini is teeny weeny and definitely nothing to worry about. Only one problem: the author of at least one of the studies which the company cited said its report “spread the wrong message to the world.”
Shaolei Ren, an associate professor of electrical and computer engineering at the University of California, Riverside, accused Google of “just hiding the critical information”. In other words – surprise! – the company has been highly selective about the data it’s used in order to create a positive view of its AI.
Greenwashing is nothing new, but I can’t remember a recent example where a business released what purported to be independent data about its AI systems. Expect much, much more of this in the future.
2. And speaking of companies that lie their way to success
Remember when the good folks at Meta got caught inflating video engagement numbers? Remember how it took publishers down a rabbit hole of “video first” content which cost them a lot of money, and many people their jobs? Of course you do! And you wouldn’t think Meta would do the same thing again, would you?
Well of course it’s going to do the same thing again.
Where last time it tried to fool publishers, this time it’s retailers that have been the target. A whistleblower has claimed the company artificially boosted the numbers for performance of its Shops ads. It by included shipping fees and taxes in sales figures, subsidised ad auctions, and applied undisclosed discounts, making the ads appear more effective than they were. It also tried to get around Apple’s App Tracking Transparency (ATT) rules. ATT, introduced in 2021, has cost Meta billions of dollars a year in revenue – money which the company was getting from abusively monitoring the behaviour of people.
The whistleblower complaint, brought by former product manager Samujjal Purkayastha, claims these practices inflated return on ad spend (ROAS) by up to 19% and misled advertisers, especially after Apple’s 2021 privacy changes reduced Meta’s access to user data.
Oh and don’t forget that recently a California jury ruled that Meta violated the state’s wiretap law by collecting sensitive health data from the Flo period-tracking app without user consent, for targeted advertising.
This is a company that just can’t help itself: it wants to avoid being bound by any rules which stop it from violating user privacy, even if it means calling in The Big Orange One to battle governments on its behalf. Crooks appealing to crooks. Whatever next?
3. Skibidi dildos
The recent spate of sex toy-throwing incidents at WNBA games are acts that aren’t so much protest as pure provocation, designed to go viral rather than make a point – the point being that women are sex objects. But it’s also connected to the rise of meme culture, fuelled by platforms like TikTok and shaped by the logic of viral web series like Skibidi Toilet, creating a climate where disruption is its own reward.
Underlying this spectacle is what the author calls the death of shame, a cultural shift where the humiliation of acting stupidly no longer sticks to the perpetrator, but to the target. The article draws a direct line from the rise of shameless public figures to a generation raised on irony, unable or unwilling to distinguish between trolling and activism.
The result is a society addicted to performance, where attention is the only real currency and even acts of humiliation become viral content. As the author puts it, “Donald Trump’s most enduring legacy isn’t a policy but a persona as the shameless troll who made humiliation a political strategy.” In this world, everything is bait, and the line between rebellion and ridicule no longer exists.
4. The AI takeover of universities is nearly complete
The first generation of students who have never worked without AI are just about to enter their final year. Ian Bogost takes a long look at them, and how AI poses problems and questions for higher education more broadly.
After reading this I started wondering whether the issue with AI isn’t really about what it’s good for – it’s what are humans good for? I would love to live in a world where all of the drudgery in life is taken care of by machines, where people work for pleasure rather than for money. Fully automated luxury (gay) space communism, and all that. Where, to tackle the worries of those students, you study not because it’s the gateway to a job, but because you want to do it.
Capitalism, of course, can’t ever get to the point of a world of leisure. The point of capitalism – which is often lost on people – is the division of the world into the people who own capital, and the people who have to work for a living. One vision of a post-capitalist world, that held by the tech bros, does away with the people who work, replacing them by machines, machines which the tech bros own. In this view, the only people who get that fully automated world of luxury are them.
5. The problem with machines is they are predictable
You may have noticed that agentic browsers – ones which can take actions on your behalf on websites – are the hip thing that every company seems to be making. I’ve been playing around with a couple (Perplexity’s Comet and The Browser Company’s Dia) and they’re interesting. I can see the point – who really wants to spend an hour researching hotels for a business trip? – but they’re about as reliable as most first generation AI products.
So it’s really not a surprise they are susceptible to scams, to a degree that’s likely higher than most human beings. All code has flaws, and vulnerabilities, and unlike those of humans they are 100% replicable. Use a flaw in an agentic browser and it will work on every instance of that code. Find a method which works on one human being, and you can never guarantee that it will work on every other.
How serious is this? Well… at the moment, not very. Agentic browsers are in their very, very early days, and none of them are all that good or that widespread. If you’re using one, you’re likely to be experimenting rather than using it on your day to day work.
6. What did Jesus do the moneylenders, again?
One of the things which has changed in my lifetime is the generalised worship of people who shuffle money around as “successes”. It kicked off in the 80s, of course, when Reagan and Thatcher turned the world into a paradise for obscure neoliberal economists.
Anil Dash notes something that should be obvious: someone who moves money around is not the same as someone who actually makes stuff. Sure, they can enable creative work: but they don’t actually do it themselves.
And yet the people who seem to be running the world right now do exactly that: shuffle bits. We laud the idea of “passive income” (which mostly means making money by having money) without actually considering the morality involved. It’s weird, and I don’t like it.
7. What happened when Ford’s CEO drove a Chinese EV?
He liked it. A lot. And he realised that Chinese EVs pose an existential threat to Western car makers.
Chinese companies like Xiaomi have a lot of advantages over US car makers. First, they’re market entrants, which means they don’t have the dead weight of their existing models or the profit margins they make. Second, of course, production costs are lower. And third, they have a government which is supporting them, rather than one which is wedded to the primacy of the gas guzzler, and even attempting to stop states enacting their own emissions mandates.
8. Vote Labour, get, erm, free AI
Of course, it makes sense that the Minister for Science, Innovation and Technology would be a bit enthusiastic about tech. It comes with the territory. But you would hope that there would also be a degree of healthy scepticism about big tech companies, at least from a Labour minister.
Enter Peter Kyle, a man who claimed he would “advocate” for Amazon when the online retailer was having issues with the government’s own competition regulator. He’s notably enthusiastic about AI, wanting closer partnerships with the USto “boost” AI investment. No doubt, some kind of lobbying job is his once his political career is over.
But it seems that even he wasn’t swayed by the idea that the government should hand over a few billion pounds to give everyone in the country access to ChatGPT Plus. Well at least he didn’t accidentally buy a bank.
9. Some sad news
Margaret Boden has died, at the age of 88. Maggie was one of the founders of the Centre for Cognitive Science at the University of Sussex, which was a big influence on AI as a discipline. I met Maggie a couple of times at conferences back when I was a student, and she was a incredibly smart person. Her book The Creative Mind was a big influence on me, and I would recommend anything and everything she wrote.
10. Your next book just landed
R. F. Kuang has a new book coming out, and this profile in The New Yorker is a great read. If you haven’t read Babel you definitely should.