@edzitron.com at his best, once again! With a little swipe at everybody's favourite ... Oracle: "[a deal that] requires Oracle to serve non-existent compute to OpenAI and be paid hundreds of billions of dollars that, helpfully, also don’t exist." 😁 www.wheresyoured.at/premium-the-...

Anthropic vs OpenAI Superbowl smackdown! open.substack.com/pub/garymarc...

Wow! @garymarcus.bsky.social has made the most important German daily newspaper! Congratulations on getting the message out there! www.sueddeutsche.de/kultur/kuens...

What’s also great is reading the dialogue with Claude where it stops and says “I need to stop and be honest with you about something important” and reveals it deleted every single photo. Software built for dunces

Have fun with that, Elon. A city on the Moon in ten years is somewhat less absurd than a city on Mars in twenty, but neither one is going to happen. The tone here is interesting, though. Almost as if he heard someone criticizing his Mars plans recently…I wonder how that might have happened? 😁

When you get visited by three Puerto Rican ghosts

every single person writing credulously about "elon will put people on the moon" should have to pin this post to their monitor

my own take is explained in detail here: bsky.app/profile/kost...

The fundamental mistake of people like @chrislhayes.bsky.social is to ignore and fail to engage with the principled skepticism of LLMs by people like Ilya Sutskever, Rich Sutton, Yann LeCun, Demis Hassabis, Melanie Mitchell and myself that has been increasingly vindicated.

This conversation between Steve Eisman (who predicted the last big financial bubble) and AI researcher @garymarcus.bsky.social is a great resource for those trying to understand the distinctions between LLMs and other more defensible AI endeavors. www.youtube.com/watch?v=aI7X...

Lunacy

Tomorrow's premium newsletter - The AI Bubble Is A Time Bomb, with every hyperscaler facing $10bn-$20bn a quarter in depreciation in the next two years and AI startups draining all that's left of venture capital. Here's $10 off annual. edzitronswheresyouredatghostio.outpost.pub/public/promo...

The distance between an “actual and true headline” vs it “performs well for user satisfaction” is a measure of moral rot at the heart of Big Tech. www.theverge.com/tech/865168/...

LLMs are not the way to AGI - and the big guns are finally starting to acknowledge it. open.substack.com/pub/garymarc...

houseofmirrors.substack.com/p/shame-is-a... Wrote about the shortcomings of anti-ai activism, what it can learn from past and present movements, and why we need to bring back shame as a tactic.

Was wondering this at the time - where was that Luddite group that was in the New York Times when there was a real deal opportunity to protest Altman going on the Tonight Show? Are they going to protest GTC? Or OpenAI Dev Day? Seems like a lot of opportunities. open.substack.com/pub/houseofm...

Here’s something fun that has gone relatively unreported. It appears that Microsoft is going billions of dollars in vendor financing, and has been for years.

“We Need to Bring this Stuff In-House”: Tesla Stock Sinks After Discovering Why FSD Is Not Prized Ford reached out in a recent report, with its head of electrical engineering Paul Costa spelling out just why licensing Tesla’s Full Self-Driving (FSD) program is a bad idea. /1

This piece is full of egregious assumptions about the cost of running Anthropic’s models based on absolutely nothing, but this may be the worst - estimating that the cost of serving Claude opus 4.5 is comparable to GPT-4o. Sometimes I wonder if techmeme has a gas leak

Brilliant interview, thanks @garymarcus.bsky.social The Big Short meets Marcus on AI open.substack.com/pub/garymarc...

Oh yeah, and watching tomato plants grow is not only exciting, it’s a very important economic use case for Claude code

What worries me with ice is that the public has no way of knowing if they are really ice. They have unmarked vehicles, wear masks, show no warrants or legal paperwork, no identification, and appear very unprofessional and violent. Anyone could be impersonating ice to terrorize our communities.

YES!!

Wenn #Trump morgen wieder mit seinem Blödsinn anfängt müssen alle anwesenden Europäer aufstehen und gehen! #Davos #sogehtsnicht

An article well worth reading!

Beatrix von Storch (AfD) beim Lügen erwischt! Und man sieht sehr gut, warum man rechten Medien wie NIUS, BILD & Co. halt nicht trauen sollte:

cannot even begin to give a shit about this, who cares? Their ASIC is delayed until 2027, their device until 2027, i bet whatever this is happens in 2027, everything is always a year or two away, reporting this stuff blandly is not helping anyone

This isn't even what the article says! It's mostly about how nobody can really tell if AI is useful for science or maths, and as ever with every single one of these stories, there's one anecdote of a person saying "yeah it's good but you know, you need to check if it's right"

“current AI systems are a death sentence for civic institutions, and we should treat them as such”

New paper by @jessicasilbey.bsky.social & @hartzog.bsky.social: How AI Destroys Institutions. AI systems erode expertise and short-circuit decisions, undermining the transparency and accountability that civic institutions need to survive. papers.ssrn.com/sol3/papers....

How are you going to pay them Sammy openai.com/index/cerebr...

"especially if AI appears unlikely to live up to its hype" is such a great sentence, it's like they can't say the words "what if AI doesn't do what it's meant to"

Grok is being used to remove clothes from random peoples' photos. Journalists think they got the chatbot to 'apologize'. What are you doing? An LLM cannot apologize. It doesn't know what is happening. A journalist doing this is a form of AI psychosis. Full ep here: www.youtube.com/watch?v=zT9l...

Lieber Herr Außenminister, liebe Bundesregierung, hier ist eure To-do-Liste für die Weihnachtsfeiertage: US-Botschafter einbestellen, Einreisesperre für Musk und Zuckerberg verhängen, HateAid-Etat verdoppeln. Jetzt ist Handeln angesagt, nicht wohlfeile Worte.

Two takes on AI and the future of America open.substack.com/pub/garymarc...

Last year I wrote my favourite thing I’ve ever written, and this year I am matching it with the equivalent rage against the tech industry - stage 4 of Enshittification, when the platforms begin to enshittfy shareholders. Out Monday if all goes to plan. www.wheresyoured.at/never-forgiv...

Premium - How The AI Bubble Bursts In 2026 www.wheresyoured.at/premium-how-... Thanks @edzitron.com , can't wait to see/read you in the next year!

█████████ ██ ███ ███ ██ ████████ ███ ██ █████████ ████████ ██ peter thiel ███ ███████ ████████ █████████ ██████████ ██ ██████████ jens spahn ████ █ █████████ ████ ███████ ████ ████ ██████ ███ ████ █████ ██ ███ ███ ████ ████ hendrick streeck██ ███████ ███████ ████ ████ █ ████████ ███ ██ ███ █████████

I don’t know how anyone can still talk about “the AI revolution” with a straight face. Reuters story about a railroad company that has sunk $300,000 into “developing AI products” and appears to have made…a chatbot that doesn’t work? archive.is/2025.12.17-0...

The article was great, the cards are brilliant! Thank you! AI Bullshit Detector Cards open.substack.com/pub/bernardm...

Important article! Thanks @willlockett.bsky.social AI Is A Hard Drug open.substack.com/pub/planetea...

This is a MUST LISTEN!

"[Workers at major companies] are asking, why do I have to use something that doesn't work? Well it's because capitalism demands eternal growth. Milton Friedman's ghost haunts every trader, Ronald Reagan pops out of the toilet every so often, this is what we do here." Love the mental imagery.

The good people of @propublicaguild.org do incredible work, and we'd be lost without them. And the erosion of labor rights is a big part of how we here in the US ended up in our current mess to begin with. Solidarity forever. 💪🏻

Come on, @propublica.org. Putting aside all of its other harms, AI degrades our communal understanding of what truth is. Good journalism relies on that. You're giving up on the actual talented people who do the work you rely on in exchange for something that will ultimately kill your business.

This is excellent, thank you! An AI Bullshit Detector open.substack.com/pub/bernardm...

An absolute bedrock institution for understanding how the planet works, just civilization-scaled vandalism by the most incurious morons on Earth

Premium newsletter: my 18k word AI Mythbusters: There’s no insatiable demand for AI compute, AI is not “getting exponentially better” or “tricking humans,” we haven't “lost control of AI,” AI doomerism is a grift, and there is no “GenAI arms race with China.” www.wheresyoured.at/ai-mythbuste...