I know a lot of people who love technology, which has left them vulnerable to being seduced by the exaggerated claims about AI. The level of hyperbole has been immense. 

Just two days ago, I had a calendar reminder I set from something I heard 1.5 years ago on the Jim Rutt podcast. According to Wikipedia, Jim Rutt was "the CEO of an American technology and Internet company."

On the podcast in question, Jim interviewed Adam Levine, on the topic of AI-powered programming for developers.  A large part of their conversation can be summarised by this comment from Jim:

As I say, mama, don’t let your babies grow up to be coders because they’ll be fucking obsolete in about eighteen months - Jim Rutt, March 6, 2025

So it is now eighteen months since Jim said this. According to what I can find out, there are more jobs now for coders than there were when Jim made this pronouncement.  

I could go into all the reasons I felt this was wrong at the time, but that isn't what is interesting. The main point was that we seemed to be in a moment of collective delusion.  This wasn't the delusion of the public or the uneducated.  This was the delusion of many of the people we would expect to know the most about the topic. The podcast in question was from the ex CEO of a technology company, interviewing someone who made his living partly by talking to the world about where AI was headed. 

In this podcast above this article with Cory Doctorow, the clearly out-of-his-depth interviewer quotes Geoffrey Hinton as an authority, saying he is the godfather of AI. Let us check a similar quote Hinton made about advances in AI:

people should stop training radiologists now...it is just completely obvious that within 5 years deep learning will do better than radiologists - Geoffrey Hinton, 2016

There are now far more radiologists than there were in 2021, when his prediction was meant to come true, and it is an incredibly highly paid and important profession. Imagine if people had listened to what this godfather of AI said was "just completely obvious".

Again, however, the point isn't to talk about Jevons ' paradox, or the long history of poor claims about AI advances, or the long history of poor claims about other revolutionary technologies.  It is the groupthink that made people unable to apply basic Bayesian reasoning to their speculation about what would emerge from the latest AI advances. The nature of technological advances seems to leave people free to reason in a fact-free universe. They don't need to be constrained by the normal rules of logic, because they are just assuming certain things will happen, which themselves are based on a string of tenuous claims.  

Doctorow, in this interview, has to go through the chore that anyone who has some doubt about current claims of AI must, and state that they are a materialist who thinks that theoretically sentient AI is possible. This caveat, of course, has almost nothing to do with the current state of AI or where it will be in the foreseeable future.  I have had to do this myself in conversation with an AI security researcher.  I felt very much as I do when I am talking to someone who believes in conspiracy theories about extraterrestrials, when I have to caveat that it is likely that in the seemingly infinite expanses of space there must be intelligent life somewhere. This has virtualy no bearing on whether someone got probed by aliens in Kansas last Tuesday.

The cultural moment with AI is part of a larger phase that I call Information Pornography. This is the tendency of the intelligentsia to enjoy talking about things that are exciting and new. Most importantly, these topics of conversation and thought should not have a clear solution or end point, so they are useful for going on and on about.  The whole debate about trans rights, which potentially helped Donald Trump win a second term as US President, was very much like this. One side shouted that trans women are women; the other that they are men. Neither is completely right, which is important because there is no clear end point to the conversation, which ended up provoking a lot of clicks, engagement, stand-up routines, and votes. Great information pornography, especially because it didn't really affect people's lives very much, so it was kind of fun to talk about, in the same way a rollercoaster is fun...speaking of things that haven't changed our lives very much, back to AI.

Many of the things we need to do to improve the world are old and boring.  In comes something like AI, which is very exciting to talk about, and allows almost limitless possibilities for intelligent people to have clever imaginings about. In a world with so many problems - poverty, environmental, human and animal rights violations people are instead talking about "fully automated luxury communism".  80,000 Hours, previously one of my favourite podcasts, has become almost singularly obsessed with AI.  Eleven of the top twenty of their most recommended careers for helping the world are focused on AI.  What isn't in there?  Anything directly related to alleviating poverty, curing cancer, removing animals from factory farms, peace and war (other than nuclear war), or anything related to the biodiversity or climate crises.  I posit that these things aren't getting as much attention at the moment, because they aren't as enjoyable for intelligent people to prognosticate about. They are also a little depressing, because they are consequential problems that require a lot from us. Better to talk about AI, or maybe trans women in sport.

There are many ways in which higher intelligence can mislead us. A simple one is that it can be used to create status, in much the same way as the bully uses violence to assert their own status.  Another, more relevant here, is that it can make us think we are the ones on the cutting edge who see the future clearly, and the masses of sheeple just aren't aware of it. I deliberately used the language of conspiracy theory there, because I think the pornography of information has multiple strains of greater and lesser rationality. The AI discussion I am talking about is definitely on the greater rationalist side, but it leaks into both sides.

OK, and a final caveat: of course, AI is important to talk about, but it's just another incredibly powerful technology. What we have to worry about is not AI loosing its chains and enslaving us all, but the rooms of hackers and criminals in parts of the world who now have a better tool to attack us with.  Of course, we also have a better tool to defend ourselves with. It may take some time to come to a new equilibrium, but as someone who has been on the Internet for decades, this evolution of attack and defence has been going on for a long time.

So perhaps that is enough of my quickly typed-out ramble that no one will read. The main point is that new technologies are highly provisional.  It seems almost every day I get an email from a reasonably reputable science source talking about the latest way we are going to solve the plastic pollution crisis, from microbes that eat plastic, to new sorts of plastics made from other things, to things to collect plastic in the ocean and so on. I have been seeing these articles for decades, and the pile of plastic pollution gets ever larger.  I understand why I see them; I want to see action on the environment, I like scientific sources, and I'm interested in technological advances. As a liberal, the algorithms even preference articles that show a thirteen-year-old non-Caucasian female has discovered the new thing that is going to fix the plastic problem.  Nothing ever comes from these things, except perhaps encouraging pacifity in people who think technology will save us.  What we need to do is stop using plastic, but this prompts negative feelings and powerlessness.  Much easier to get a quick endorphin hit from thinking that someone is simply going to fix the problem. This is science as a distraction rather than as a guiding light for the way forwards.

We need more of the work that 80,000 and effective altruism was previously more focused on: smart thinking about practical things we are going to do to fix the world. We also need to create a cultural landscape, including democratic consent, for these things to happen. Technology will help us fix the world, and it will also cause problems, which we will figure out how to deal with. Techno-utopian dreams and techno-sci-fi nightmares are not helpful at the moment, though they sure keep people talking and get people clicking...