GPT-2030 and Catastrophic Drives: Four Vignettes

·Bounded Regret··

I previously discussed the capabilities we might expect from future AI systems, illustrated through GPT2030, a hypothetical successor of GPT-4 trained in 2030. GPT2030 had a number of advanced capabilities, including superhuman programming, hacking, and persuasion skills, the ability to think more quickly than humans and to learn quickly by sharing information across parallel copies, and potentially other superhuman skills such as protein design. I’ll use “GPT2030++” to refer to a system that ha...

Read full article →

Related Articles

OpenAI’s o1 correctly diagnosed 67% of ER patients vs. 50-55% by triage doctors
donsupreme · Hacker News · 5mo ago
Accelerating Gemma 4: faster inference with multi-token prediction drafters
amrrs · Hacker News · 5mo ago
A couple million lines of Haskell: Production engineering at Mercury
unignorant · Hacker News · 5mo ago
Harvard particle physicist Matthew Schwartz drops 36 papers authored with Claude
xqcgrek2 · Hacker News · 1d ago
An AI agent emailed researchers for help. It told us why
sbulaev · Hacker News · 21h ago