Real-time driving gameplay isn't new: it was doable in 2016, dev says
gandamu_ml · x · 2026-09-18
Developer gandamuml pushes back on hype around a model playing a driving game in real time: this was roughly achievable back in 2016 and easy to train — essentially a continuous classifier he built in 2 days as a first project. He notes outputting all action values in a single pass is a known, more efficient approach.
Related event: Real-Time Driving Game Play Is No Breakthrough, Researcher Says(2 posts)→
More from Models
- Router flaw spotted: it picks models but never grades answers — grade-then-escalate fix proposed — airesearch12 · 2026-09-19
- Every frontier lab is having an urgent 'Jev' meeting today, observer says — altryne · 2026-09-19
- Jev classifies the entire Bluesky firehose for animal posts in near real time — danshipper · 2026-09-19
- Critics say OpenAI's eval environments aren't airgapped and barely monitored — MoonL88537 · 2026-09-19
- Codex Reset Coupon Changes Usage Reset Cycle, User Flags Billing Quirk — algo_diver · 2026-09-19
- 28 hashes logged: crowd audits Gensyn's open-1b, first verifiable training run — benfielding · 2026-09-19