Generative AI is dangerous — but not for the reasons you might think

ChatGPT, Open AI’s new chatbot, is, quite frankly, amazing. Generative AI — machine learning systems that respond to a user’s prompts by generating text or images — looks poised to revolutionise the media, education, and communication more generally. Lack of responsibility, not reliability There is, however, a deeper reason why the outputs of generative AI are bullshit, which also applies to generative AI that outputs images, like Dall-E — even though we wouldn’t normally criticise images for not being “true”. This, rather than any lack of reliability, is why the outputs of generative AI are fundamentally bullshit. How bullshit begets bullshit Finally, the outputs of generative AI are dangerous bullshit because bullshit spreads.