AI as conspiracy theorist

My conversation with davinci GPT-3 text generator

I’ve been talking to an AI about conspiracy theories. If you would like to interact with this AI, you can sign up for the beta here (tip: say you’re an AI researcher so it lets you join). The model I’m interacting with is called davinci and it’s one model of GPT-3, which Wikipedia describes as is “an autoregressive language model that uses deep learning to produce human-like text.” I’ll post my entire conversation, but without much prompting it gave me this result (questions are mine, answers are AI generated):

Interesting that this AI is saying things that got Alex Jones kicked off the internet and sued. Below is the entire conversation leading up to the question about Sandy Hook. The AI does give results sometimes based on things you said previously in the conversation, so you could argue I was leading it down the road to conspiracy theories. But still.

Also notice that it seems to know who I am but makes up facts about me that aren’t true, but are plausible. It makes up episodes of the show I’ve never actually done but on topics I have covered.

As you can see, eventually I broke it. It couldn’t handle the level of truth I was asking it to reveal. Still, it telling me about crisis actors was something I didn’t expect. Try it out for yourself if you want to go down this AI rabbit hole.