Post

Newsletter: I am sick and god damn tired of everybody pretending that generative AI is the next big thing. The media is complicit in accepting fantastical nonsense - both in the numbers put out by OpenAI and the silly jobs created by Anthropic - and it has to stop. www.wheresyoured.at/reality-check/

It's blatantly obvious that generative AI is a bubble. The returns and products just aren't there, and yet the media compliantly prints whatever OpenAI wants, even when it's insane things such as "they'll make $3bn from an agent product that barely exists in 2025." www.wheresyoured.at/reality-check/

The media needs to stop trusting every single thing OpenAI or other AI companies says. Blindly printing that OpenAI will make "$25bn from "new products" that don't exist in 2029 is propaganda, not analysis - and doing so only seeks to enrich the powerful. www.wheresyoured.at/reality-check/

Every story about OpenAI feels bereft of analysis or conscious thought. Their agents don't work, they will burn $6bn on inference alone in 2025, they have no path to profitability, yet most articles talk about them like a gifted child brimming with potential. www.wheresyoured.at/reality-check/

I'm tired of fielding the same half-assed critiques of my work, so I've condensed my answers into one helpful section - no, the cost of inference isn't going down, no, OpenAI isn't growth stage, no, it isn't like the early days of the internet. Come at me. www.wheresyoured.at/reality-check/

Generative AI is a remarkable con — a just-good-enough simulacrum of human expression to get it past the gatekeepers in finance and the media, knowing that neither will apply a second gear of critical thinking beyond "huh guess we're doing AI now." It's a disgrace. www.wheresyoured.at/reality-check/

The generative AI bubble - by which I mean the growth of ChatGPT - grew in part thanks to Google Search's failure to innovate. Most people I know who use ChatGPT do so for search, as Google has got so much worse at understanding what it is you're searching for. www.wheresyoured.at/reality-check/

I also briefly want to address anyone writing about AGI: stop. Stop doing it. Kevin Roose's "is AI conscious?" piece was a joke, a thinly-veiled ad for a new job at Anthropic, and every piece like this seeks to misinform the public and galvanize the powerful. www.wheresyoured.at/reality-check/

It is now actively illogical to act as if generative AI is a real movement. Analyst Josh Beck said last week that Amazon will only make *$5bn* from AI in 2025 - after spending $105bn in capex. And they're pulling back on data centers too. This isn't a revolution! www.wheresyoured.at/reality-check/

It's sad how little we invested on new architectures. Everyone put billions into training the same models to gain 0.1 % on MMLU. Sure, new architectures like Titans might power Gemini 2.5 which partly solved context degradation over 80k tokens. But the Agent dream isn't achievable, it's too late.

They won’t and they will.