Self-driving as a case study for AGI
Self-driving cars are used as a lens to debate how artificial general intelligence (AGI) might emerge and reshape work. Commenters argue over whether systems like Waymo and large language models represent true “general” intelligence or just narrow automation, how to define AGI in the first place, and whether economic definitions tied to job replacement are too reductive. The conversation also explores social impacts such as job loss, misinformation, regulation, and whether advances in AI will lead to broad prosperity, deeper inequality, or require major changes like universal basic income.
Definition of AGI vs. “Just Automation”
- Many argue self-driving is sophisticated automation, not AGI: it solves a bounded task and doesn’t generalize.
- Disagreement over AGI definition:
- One camp focuses on “general problem solving” across domains (math, programming, science, driving).
- Another emphasizes economic framing: surpassing humans on most economically valuable work.
- Some find the economic definition “depressing” and hijacked by corporate/marketing interests.
- Several posters think current discourse “forgets the G” and that LLMs + self-driving show no clear path to true generality.
State of Self‑Driving Tech
- Waymo in a few U.S. cities is praised by some riders as safe and reliable; others point to crashes, awkward edge cases, remote assistance, and geofencing to argue it’s oversold.
- Debate over whether current systems are “AI” vs. “complex control algorithms,” with some insisting they’re just forward compute graphs, others noting modern self-driving stacks are heavily ML-based.
- Some see hardware constraints (camera quality, fixed positions, lack of additional sensors) as fundamental; others argue more sensors or better software could overcome this.
LLMs, Intelligence, and Tests
- Dispute over whether LLMs qualify as AI at all; some say they’re “just statistical tricks,” others cite mainstream AI textbooks that treat language models as core AI.
- Turing test and Chinese Room thought experiments are discussed; many think LLMs fail at self-driven information seeking, common sense, and grounded perception.
- Proposals for more demanding AGI benchmarks (e.g., solving complex game puzzles like Zelda shrines).
Economic and Social Impact
- Historical analogy: past automation killed jobs but created new ones; some worry this may break with AGI.
- Fears of job loss (especially creative work), “bullshit jobs” disappearing, and whether UBI will be necessary or politically feasible.
- Concerns about AI-accelerated misinformation and deepfakes, though others argue propaganda has always existed and society will adapt.
- Some expect mild, gradual adjustment like with self-driving rollouts; others predict severe unrest, protests, and harsher responses once physical systems are widely automated.
Broader Skepticism and Timelines
- Several commenters think serious AGI (in the science‑fiction sense) is decades away and that current “AGI soon” narratives are mostly hype.
- Frustration over shifting terminology and lack of precise, shared definitions; calls to separate philosophical debates from practical automation discussions.