Freddie deBoer's proposed a bet with Scott Alexander, predicting that all of these 17 conditions will be true: I'm Offering Scott Alexander a Wager About AI's Effects Over the Next Three Years.
1,000People are also trading
This is a great operationalization of the 'will AI disrupt the economy by 2029' question. The individual conditions are all priced at 87-92%, which implies roughly 90% chance each stays normal. But the joint probability matters — if these are somewhat correlated, P(all 17 normal) is still high. If independent, it would be 0.9^17 = 17%, but economic indicators are heavily correlated so that math is wrong.
The key question is whether AI advances could cause ANY single metric to breach its threshold by Feb 2029. The most vulnerable seem to be the employment-related ones — 'combined employment in software developers, accountants...' and the college wage premium. Those are the channels through which AI substitution would first show up. But even aggressive AI deployment timelines suggest meaningful labor market impact by 2029 is unlikely to breach these wide bounds.
DeBoer is probably right that most of these hold. The economy has enormous inertia. The more likely failure mode for Scott's side would be an exogenous shock (recession, financial crisis) rather than AI-specific disruption within 3 years.
@ChurlishGambit Te more boring / normal each prediction, the stronger the claim that all of them hold. I'm sure Freddy would have no objection to adding "the union dissolves" to the list. Let's also add that the sun won't be covered by a Dyson sphere, etc. It makes no difference.