Jimmy Apples

Reports overhearing the present described as “the megaproject era”, meaning one person can build in days what took a dedicated team months, and says it still prefers “gentle singularity”. Framing of the present, not a forecast. Full post text inspected via the X API.

Quoting a report that RSI was a key part of Gemini 4’s training, asks what RSI means to Google, whether it is a spectrum or “full on foom”, and whether other labs will freak out, joking about telling Eliezer Yudkowsky to hide in a basement. Questions and a joke, not claims; the quoted report is someone else’s. Full post text inspected via the X API.

Says the public hears that AI will kill us all but only gets a productivity amplifier and AI videos, so it asks why build it at all; the industry will have to work much harder on optimism and on explaining why. A critique of AI messaging, not a risk estimate. Full post text inspected via the X API.

Says that telling people things need to slow down, after teasing a world of abundance and science breakthroughs while giving them three.js games, naturally provokes strong emotions; it then names Hugging Face, AISI and CAISI as independent evaluators. Criticizes inconsistent lab messaging rather than taking a side on a slowdown. Full post text inspected via the X API.

Says people love drama and negativity: friends barely discussed a swarm of agents solving a millennium problem but talked at length about a Zelda trailer. Concludes “We are still early”, implying society is underreacting to AI progress. Full post text inspected via the X API.

Quoting an earlier post, reminds readers that the person who coined the term AGI said it has already arrived, calls arguing over personal definitions dumb because AGI is a line in the sand on a spectrum, and says what matters now is where we are on the spectrum of recursive self-improvement and the future pace. Full post text inspected via the X API.

Says some highly strung, neurotic people in San Francisco tie their identity to imminent AI takeover and would do well to take a breather and touch grass, quoting a news line about other developers not holding themselves to safety limits. Mockery of a social scene, not an argument that risks are zero. Full post text inspected via the X API; the quoted line is someone else’s.

Reports countless conversations with people outside the online AI bubble who ask why a data centre should be built near them quickly just because a rich Silicon Valley figure says so. Union support for data centres is a good start, it says: lead with benefits, not dismissal. Full post text inspected via the X API.

Reacting to Sam Altman saying he thinks AGI arrives this year, says there are already multi-agent swarms and RSI and that people are “not being rsi pilled enough”. Signals that the account thinks progress is further along than mainstream framing suggests; no date is given. Full post text inspected via the X API.

Poses a poll: if you believe open-source models will lead to large hacks on critical infrastructure, do you rapidly distribute defensive AI models, ban models that cross a cyber evaluation, or something else? Adds “It’s going to get worse, very soon.” The account does not pick an answer, but expects cyber misuse to grow. Full post text inspected via the X API.

Quoting a line from a paper whose authors credit discussion with ChatGPT for a theorem, says it is looking forward to the day AI delivers big results in health and science too. Full post text inspected via the X API; the quoted line belongs to the paper’s authors.

Says it loves the idea that “Billy from Alabama” is asking AI to try to solve 60-year-old math conjectures, and asks what else an average person with a swarm of agents could help solve. Enthusiasm for broad access to frontier capability. Full post text inspected via the X API.