Researcher Lets GPT-2 Ask Codex About World Affairs in Curious Experiment
MikePFrank · x · 2026-10-06
Mike Frank ran an experiment letting 2019-era GPT-2 ask OpenAI's Codex questions about world affairs, with the two models conversing. He called the first test "pretty interesting" and shared a screenshot of the dialogue, without further details.
Related event: Researchers Let GPT-2 Command Codex Agent in Fun Cross-Era Experiment(3 posts)→
More from Fun
- Critic calls out month-late disclosure and 'rogue agent' narrative as accountability dodge — gerardsans · 2026-10-06
- GTA 5 ported to WebAssembly with AI's help — 'the end of PC ports?' — pvncher · 2026-10-06
- Why models never give up: RL training punishes quitting, so no chill version exists — rickasaurus · 2026-10-06
- User claims Opus 5.5 found 200GB of junk left by Codex—vs 15GB from Cursor — RexDouglass · 2026-10-06
- Mistral Large 4 trails GLM-5.3 on Artificial Analysis index, fueling 'eval crisis' talk — Yuchenj_UW · 2026-10-06
- Gergely Orosz: startup tech blogs are increasingly unreadable AI slop — rseroter · 2026-10-06