Even that would be a bad law because there’s no good way to ensure such things. AI is a predictor of the most plausible next word… if that most plausible next word results in a sentence that tells someone to kill themself, it’s working as intended and it doesn’t align with my sense of morality at all that the AI provider should be legally liable for this.
Parents and schools should teach kids that AI doesn’t tell you the truth, nor give useful life advice, merely predicts what someone might plausibly answer to what you enter.
Even that would be a bad law because there’s no good way to ensure such things. AI is a predictor of the most plausible next word… if that most plausible next word results in a sentence that tells someone to kill themself, it’s working as intended and it doesn’t align with my sense of morality at all that the AI provider should be legally liable for this.
Parents and schools should teach kids that AI doesn’t tell you the truth, nor give useful life advice, merely predicts what someone might plausibly answer to what you enter.