superintelligence

Fully convinced that artificial superintelligence is coming, imminent, and world-changing — and structuring your beliefs, anxieties, or career accordingly. Superintelligence pilled people live in a constant low-grade awareness that the old rules may not apply for much longer. They read Bostrom, they track frontier model releases, and they have opinions about timelines. The term can be used earnestly within AI-obsessed communities or as gentle mockery of people who have made AGI their entire personality and can't get through dinner without bringing it up.

Artificial Superintelligence — AI smarter than the smartest humans across every domain. Distinct from AGI (general human-level). Doomers think we get one shot to align ASI; e/acc thinks ASI will solve all our problems.

Performance so far beyond the current state of the art that existing benchmarks can't meaningfully measure the gap — a result or capability that essentially breaks the evaluation framework rather than just topping it. Hyper sota describes the theoretical or emerging category of AI achievement that lives above SOTA the way SOTA lives above average: uncharted, benchmark-resistant, and only evaluable by capabilities that don't have tests yet. Often used speculatively about anticipated future models or results that are rumored but unverified.

Hyper agentic pushes autonomous self-direction beyond the normal register into something that starts to feel uncontrollable or unmanageable — a level of independent initiative so intense that it creates its own gravitational field. Where peak agentic is impressive, hyper agentic tips into slightly alarming: the person or system is taking on so much unprompted initiative that keeping up with them becomes the challenge. In AI contexts, hyper agentic describes systems with broad tool access and minimal constraints. In human slang, it's the person who solved the problem you hadn't mentioned yet.