Accept 'bad things' in return for benefits of AI, says Sam Altman
Claims by OpenAI CEO Sam Altman that society should accept “bad things” in exchange for the benefits of AI provoke intense skepticism, with many questioning who decides these trade-offs and who bears the risks versus the rewards. Commenters compare AI’s externalities to those of chemicals, surveillance tech, and social media, arguing that regulation, corporate accountability, and social safety nets are necessary rather than blanket acceptance of harm. Underlying the debate are concerns about concentration of power, erosion of justice and trust, and whether AI’s touted benefits meaningfully outweigh its social and environmental costs.
Tradeoffs and “accepting bad things”
- Many object to the framing that society must “accept bad things” for AI benefits, arguing benefits never excuse harm or antisocial behavior.
- Others say tradeoffs are unavoidable for any technology, but debate what “accept” means: tolerating that harms occur vs socially condoning specific abuses.
- Some see Altman’s argument as “ends justify the means,” with costs borne by the public and gains privatized.
Power, inequality, and who decides
- Strong concern that AI’s benefits and harms are unevenly distributed: upside for the wealthy and tech elite, downside for workers and marginalized groups.
- Repeated theme: the rich and powerful, not “society,” effectively decide which tradeoffs are made.
- Extreme wealth is described as driving a desire for power, not just money.
Regulation, guardrails, and accountability
- Several argue regulation should constrain “bad things,” just as with chemicals, finance, and other high‑risk industries.
- Criticism that AI companies want strict rules when it slows competitors but oppose third‑party guardrails or liability when their models cause harm (e.g., assisting crime).
- Suggestions include corporate “death penalties,” banning negligent executives from leadership, and legal responsibility for harmful model outputs.
Violence, justice, and social stability
- Thread debates political violence: some condemn any targeting of individuals, warning it leads toward failed states.
- Others argue systemic injustice and nonresponsive institutions naturally push the oppressed toward violence, noting that state and political violence already exist.
- Disagreement on whether justice systems are merely slow or fundamentally “clogged” and illegitimate.
Local, environmental, and surveillance impacts
- Concerns about data centers (noise, siting, power, water use) but some see this as secondary to broader risks: labor disruption, political power concentration, and bio/terror misuse.
- Worries that AI-enabled surveillance drastically lowers the barrier to intrusive monitoring, even if surveillance tech predates modern LLMs.
Benefits and capabilities of AI
- Skeptics question whether AI’s real benefits justify social and cultural costs, dismissing many outputs as low‑quality “slop.”
- Others cite substantial gains, including claims that frontier models are transformative for areas like mathematical research.
Alternatives and social safety nets
- If large-scale disruption is “inevitable,” some argue the obvious response is building robust social safety nets; the perceived lack of support is taken as evidence of misaligned motives.