Will there be evidence of war crimes by NATO member states counterfactually due to AI by 2030?
GeopolíticaOne-OffEuropa3a
Manifold MarketsSem KYCDados de resolução verificadosRazoavelmente calibradoPrevisão da comunidade atualNão 76.6%
Previsores
9
Tipo de pergunta
binary
Metodologia
Play-money forecasting platform
Tipo de fonte
Previsão
Dados do mercado
Atualizado há 7 horas
Desatualizado17/01/24, 14:18 UTC1/01/30, 23:59 UTC
Tendências
Regras
From various documentaries and articles, I've gathered a few things about future non-existential but still catastrophic risks from military AI.
- It may increase the likelihood/frequency of war crimes:
- In the initial stages, AI drones / autonomous weapons may frequently misidentify targets.
- AI drones / autonomous weapons may be programmed to not be capable of refusing orders from human or AI commanders to commit war crimes.
- To achieve objectives, AI commanders may choose highly bizarre and unconventional solutions that involve committing war crimes, due to not having a complete intuitive human morality, like Yudkowsky's unsafe genies.
- Human commanders that already want to commit war crimes may feel emboldened, that they can avoid personal blame, or that they can convince moderates, if an 'objective' AI advisor calculates that committing war crimes is the optimum strategy to achieve objectives.
