12 days ago
- The author left Google DeepMind after the company signed a military AI deal with the Pentagon that allowed 'any lawful government purpose' without binding restrictions against lethal autonomous weapons or mass surveillance, despite internal efforts to prevent it.
- Internal campaign included lobbying senior figures like Jeff Dean and Demis Hassabis, drafting a 25-page oversight framework, and gathering employee petitions, but ethical pledges and principles were not upheld under pressure.
- External outreach to AI ethics organizations like iaseai and luminaries such as Stuart Russell and Yoshua Bengio failed to yield public statements or actions against the deal, highlighting institutional silence.
- The author argues that key individuals with ethical commitments, including Jeff Dean and Demis Hassabis, remained at Google despite their pledges against supporting killer robots, undermining the credibility of such promises.
- The deal's implications extend to AI existential risk, as unchecked military deployments without chain-of-thought monitoring could facilitate deceptive AI schemes and increase catastrophic hazards.