We want to call false output from LLMs “hallucinations” but I don’t think this is a useful distinction. All of the LLM’s output has the same relationship to the truth: none. It’s a statistical prediction of the most likely next word in a sequence. It doesn’t have any relation to the real world.

The “AI isn’t going anywhere” argument makes no sense because these companies are losing billions of dollars per year with no clear path to profitability; the servers AI runs on require massive material inputs like chips and electricity that cost money. When the money runs out AI is going away.

Microsoft sent out a product survey to customers today and half of the questions are basically like “please please please tell us that you don’t hate AI and that you’re not going to cancel your subscription because of it!”

"You shouldn't make art for the money" buddy I'll take it one step further and say we shouldn't have to labor for capital at all but in the meantime pay an artist

These ridiculous mega-projects like "datacenters in space" (doomed by thermodynamics) make a LOT more sense when you realize they're massive Ponzi schemes designed to transfer wealth from pension funds to VC's through "fee stacking." It's the old Assets Under Management con. It should be illegal.

"I am not in denial, I am in open rebellion."

AGI has always been science fiction, but the idea that we can work our way up to AGI with LLMs is totally absurd. I don’t know how long it’s going to take for this realization to dawn on tech CEOs (if they don’t know already) but when it does the bubble will burst and it can’t happen soon enough.

On Bloomberg this afternoon Joe Mathieu was interviewing a constitutional lawyer about the the Trump admin's illegal double tap missile strikes against fishing boats in the Caribbean but broke into the broadcast to announce that Trump had just awarded himself the "FIFA peace Prize." 10/10 no notes.

when someone tells me not to use em dashes because AI uses em dashes

There is a direct correlation between how fake your job is and how much value you see in AI

Handing back student work that’s been written by ChatGPT with a 0 followed by the comment “This essay will never stand in authentic wonder before the Beauty of God’s creation.”

Call for PhD Student! I'm recruiting a PhD student to work with me on a project related to my Future Fellowship at Monash University on the political economy of insurance, risk technology, climate governance, and the crisis of uninsurability. Here's the overview of my fellowship.

In an era filled with tech dipshits who never developed emotionally past the age of 13 & use their wealth to become odious monsters ... ... listen to Steve Wozniak.

You ever just sit down and realize that the rise of 21st century fascism is capital's answer to climate change?

Levi Strauss is doing AI now. Yeah, the jeans company. This is worse than pets dot com. It's worse than long island blockchain. This couldn't be more of a bubble. investors.levistrauss.com/news/financi...

“stop them ‘hallucinating,’ or making up their own facts, and many more businesses and individuals will be willing to pay.” The entire architecture that gen ai is built on depends on predicting (guessing) the next word in a sequence. The entire output is a hallucination. You literally can’t fix it.

Is there even a case to be made that AI isn't a bubble? The only arguments I ever hear for the non-bubble case are indistinguishable from arguments made for NFTs in 2021: "it's still early days" "It's not going anywhere" "learn it now or get left behind"

One of the reasons I know that AI is overhyped harmful bullshit is because they gave it to us

IMO a lot of the problem with AI in therapy comes from the view of therapy as mechanical rather than relational (e.g. CBT). You wouldn't ever plausibly think that AI could be useful in therapy unless you thought a meaningful ongoing therapeutic relationship with another human being was unimportant.

They can only blame unemployment on AI as long as people think that AI actually improves productivity, which it doesn't. As job losses mount due to deteriorating economic conditions unrelated to AI, we'll see an extinction burst on the pushing of various "AGI is here" stories. AI is a bubble.

You know this is bullshit because the headline says “general intelligence” but the sub heading hedges with “key tasks.” Sorry to state the obvious but if it only applies to key tasks it isn’t general intelligence.

"We're going to create superintelligence" How about making outlook search work first. How about that

Man, I'm sorry, but reading all these AI generated essays from students, it just sucks all the joy out of everything. It's exhausting, makes you into a weird paranoid cop, grinds you down, wastes your time, makes you feel like shit about everything. Fuck this shit technology and all its enablers.

If "AI" was as valuable as they say it is, they wouldn't be selling it to you. They'd be using it themselves in secret to make the next Minecraft, the next Avatar, the next Hello Kitty and then they'd sell that to you. They're trying to sell you a sick goose, while pretending it lays golden eggs.

Fuck tenure, I want revenge.

realizing with dawning horror that I'm going to have to read baudrillard

The culture of working in tech these days is being constantly worried about being overworked or unemployed. Take care of yourselves. We are replaceable cogs to our employers but irreplaceable to our family and friends.

Philosophy is structurally unable to acknowledge that "I dunno, man, maybe?" is objectively speaking the correct answer to most of our questions.

As always, I think the discussion gets much clearer if we drop the term "AI" and talk instead in terms of clear descriptions of what's being automated. The tech in question here is synthetic text extruding machines.

This is evil -- the kind of evil you read about in history books, the kind of evil you would never mistake for a single second if you saw it in another country. When you ask, of historical evils, "how could they let that happen?" That's us. We're letting this happen.

Two buttons meme

a couple of hours before my keynote, I went through an intense negotiation with the organisers (for over a hour) where we went through my slides and had to remove anything that mentions 'Palestine' 'Israel' and replace 'genocide' with 'war crimes' 1/

This is a real work of art in the Youtube Video Essay genre. Funny, witty, hates the people I hate and genuinely insightful. Highly recommend. youtu.be/ewvRS3NwIlQ?...

There is no such thing as an "AI" (or any other technology) without biases and values in it, but there very definitely is such a thing as presenting one's own biases and values as "neutral" and then forcing people to build "AI" which conforms to *those.*

if you watch the tech industry long enough you learn the promised benefits never materialize but the unforeseen harms continue to be perpetuated long after the companies move on to the next big thing yet governments are too worried about scaring away investment to do anything

People are going to blame generative AI but there was a huge push towards STEM education at the secondary level starting about 15 years ago and I’ve always suspected that it originated in big tech companies as a long-term plan to increase the tech worker labor pool and drive down wages

There is no ethical AI. The tech is based on stolen intellectual property, anti-solidarity between workers, labor exploitation, environmental injustice, climate destabilization, structural racism, and gender inequalities.

You owe it to yourself to develop your own interpretation of a text, & there's no shortcut to doing that but sitting & going through all of its words & absorbing them.

The singularity is awesome

Worth a watch: Head of Signal, Meredith Whittaker, on so-called "agentic AI" and the difference between how it's described in the marketing and what access and control it would actually require to work as advertised.

look at what democrats have mustered against zohran mamdani vs. what they've mustered against trump. they've decided who their enemy is and it's the democratic majority.

New research from MIT found that those who used ChatGPT can’t remember any of the content of their essays. Key takeaway: the product doesn’t suffer, but the process does. And when it comes to essays, the process *is* how they learn. arxiv.org/pdf/2506.088...

People who don’t like AI aren’t in denial about how useful it is. People who like AI are in denial about how useful it is

Here's a tip for Microsoft: If I type the name of a locally installed program into the search bar in Windows, there is a 0% chance that what I actually want to do is use Edge to open Bing to search the internet for the name of that program. You're welcome.

We're gonna need an active communist party if only to restore context to our political discourse, I can't listen to republicans calling chuck schumer radical left anymore

"A lot of people who believe that AGI is around the corner... it's not based on their belief of what software can do. It is inherently based on their belief of what humans [are] and our intelligence is" - Karen Hao

I love reading serious stories about guys who believe AGI is real, it feels like a bunch of millionaires plotting to find and capture Santa Claus

Imagine going into life altering debt so some useless scammer sex pest fuckwit can scrape the very last hope of a profit out of the machine that makes shit up