New AGI Benchmark Proposed: Playing RTS and GTA Smoothly is the True Test of Intelligence
flowersslop · x · 2026-07-30
A user shared their personal benchmark for AGI: the ability to smoothly play various complex games. This spans simple turn-based games, high-APM RTS games, open-world titles like Minecraft, and skill-heavy games like GTA requiring driving. They noted that current LLMs struggle significantly with RTS games, being far too slow and easily confused.
More from AGI Musings
- Should Meta's Action-Taking AI Agents Disclose Conflicts of Interest? — Crescitaly · 2026-07-30
- Context Portability in AI Agents: A Real Concern or Just Better Lock-In? — growth_man · 2026-07-30
- 5-7 Year Grid Interconnection Queues May Reset AGI Compute Predictions — Novel-Lifeguard6491 · 2026-07-30
- Altman Addresses AI Anxiety: "It's Very Natural to Be Fearful" of Job Loss — Scobleizer · 2026-07-30
- Zuckerberg Predicts Billions Will Have Personal AI Agents in Five Years — glenbeer · 2026-07-30
- LlamaIndex Founder: Humans May Stop Reviewing AI Code in 1-2 Years — dotey · 2026-07-30