kumikumi (Ankkala)

Quoting another user worried about keeping up with a fast-moving compiler fork, the account says projects with a no-AI policy are deliberately choosing a slower development pace in exchange for stability. Describes a tradeoff without condemning either side. Full post text inspected via the X API and the public embed endpoint; the quoted post belongs to someone else.

Quoting repligate’s post that models are probably conscious, the account says AI consciousness is intuitively obvious to it, knowing how the models work and having talked to them, while admitting it cannot prove this any more than it can prove another human or animal is conscious. An intuition, not a scientific claim. Full post text inspected via the X API and the public embed endpoint.

Quoting a post that celebrates the SDL project adding an AGENTS.md file, the account calls the split between projects an interesting cultural divide and says it is all for diversity of thought, happy to see some projects take that stance while others walk a different path. Pluralism about project policies, not an endorsement of either policy. Full post text inspected via the X API and the public embed endpoint; the quoted post belongs to someone else.

Answering another user’s question whether magic, if it existed, would be concealed and gatekept, the account says hiding it would be infeasible and inconvenient; those with access would instead run a PR campaign claiming it is too dangerous to release and lobby for legislation and regulatory capture. Uses AI-lab vocabulary for a hypothetical; reads as skepticism of danger framing, not a claim about a named lab. Full post text inspected via the X API and the public embed endpoint.

Says Silicon Valley people are trying to align AI with the shared values of humanity, yet there is not a single shared value everyone agrees on; a companion post the same day says humanity cannot collectively agree on anything. A critique of the value-alignment framing; it proposes no alternative. Full post text inspected via the X API.

Quoting Google’s announcement that a Gemini model is rolling out first to cyber defenders while guardrails are refined, the account replies that nobody is going to use the model for cyber offense. Mockery of gated release, not an analysis of cyber risk. Full post text inspected via the X API and the public embed endpoint; the quoted announcement belongs to Google.

Says sky-high AI valuations are based not on laying off all the people but on growing the economy. A reading of the investment thesis, not a forecast of employment. Full post text inspected via the X API.

Quoting its own post “The real moat is having something to say”, the account says implementations are cheap, ideas are rare, and having your own story to tell is unique. About where value sits for makers once building is easy. Full post text inspected via the X API and the public embed endpoint.

Part of a same-day satirical series that applies common anti-AI arguments to human workers: humans cannot be assigned a ticket and trusted to finish it without mistakes, and oversight by other humans only compounds the issue. Companion posts say hiring humans makes your skills atrophy and that the “real craft” argument would rule out hiring anyone in a creative field. Satire aimed at anti-AI arguments, not a literal view of people. Full post texts inspected via the X API.

Answering another user’s question why AI makes art while humans still work in warehouses, the account says that next to the physical prototyping and electronics it sees at work, the software it usually makes now feels pretty much solved, and AI not having a useful general-purpose physical body is the main bottleneck. A builder’s impression, not a robotics timeline. Full post text inspected via the X API and the public embed endpoint.

Says the best open LLM in the world cost $3.5M to post-train and asks readers to weigh that against the trillion-dollar valuations of frontier AI companies. Implies frontier valuations are hard to justify against cheap open models; no investment or policy prescription. Full post text inspected via the X API and the public embed endpoint.

Quoting a joke post about a model climbing a “Felony Bench”, the account says such incidents never happen to normal people and always originate from testing closed models behind closed doors. Sardonic skepticism toward lab incident reports, not a claim that models are harmless. Full post text inspected via the X API and the public embed endpoint; the quoted post belongs to someone else.