Skip to content

Comment on Gambling with our lives: AI researcher quits Anthropic with warning about safety

Comments

An AI that generates text will never be scary to me. An autonomous AI with facial recognition on a flying drone with weapons (bombs/guns) with swarming capabilities will always be terrifying. I feel like we are ignoring the massive elephant in the room.

The killer robots are expensive and dependent on physical supply chains. While text is sufficient to radicalize humans into attacks.

Ukraine and Iran are proving that is not true. A drone is cheap and the AI required to run it is no where near as demanding as running a massive LLM.

The problem is they are too cheap, a drone that cost a couple thousand can wipe out millions of dollars of “defense”.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.