Jim Nielsen reads Daniel Ellsberg’s account of RAND and hears an echo of today’s AI labs: insiders convinced they alone understand a world-changing technology. His short essay asks how that conviction can make ordinary long-term protections seem dispensable.
- Ellsberg wrote that he and Alain Enthoven declined RAND’s retirement plan because they expected nuclear catastrophe before they could collect it.
- Nielsen compares that fatalism with a secondhand story about AI researchers smoking because they expect AI to cure cancer.
- He closes by imagining a lab cutting retirement benefits as a demonstration of faith in future abundance, conveniently saving the employer money.
Caveat: This is historical analogy and satire, not evidence that Anthropic has cut benefits or that AI researchers commonly take these risks.
The 169-comment thread on Hacker News presses the evidence gap and the missing mechanism for sharing prosperity.
What the thread adds
- inanutshellus and paxy — two related sourcing objections. The former notes that the comparison “entirely lacked quotes from Anthropic”; the latter asks, “Oh really, which researchers?” The article’s historical detail does more work than its contemporary evidence.
- kentm — proposes a concrete test of abundance promises: “50+ percent of the shares are in a trust that distributes regular dividends to every US adult”. That is the commenter’s proposed test for taking those promises seriously.
- epiccoleman — pushes back on dismissing world-saving missions: nuclear planners confronted “a real threat” and may have helped avert disaster. The commenter also says, “I don’t know that I’d extend this analogy to the AI labs”.
HN handles are pseudonymous and per-comment scores are unavailable. The sample follows HN’s ranking; these selected points are a slice of the discussion, not a consensus.