Whereas internet search gives us results, generative AI outputs deliver us interpretation.

AI’s capacity for ultra-customized answers tends to reinforce our existing beliefs.

Deliberately designed, evidence-based conversations with AI can reduce belief in conspiracy theories.

For most of human history, reality has had the inconvenient habit of talking back.

Other people disagree with us. Facts refuse to fit. A colleague questions our interpretation. A friend says we may have been unfair. A doctor gives us an answer we did not want. The newspaper contains something that irritates us. Somewhere between our convictions and the world, there has usually been friction.

Artificial intelligence is changing the flavor of friction

Increasingly, we turn to artificial intelligence (AI) after an argument, during uncertainty, when trying to understand an illness, interpreting someone else’s behaviour or making sense of events. We arrive carrying a hypothesis: I think my boss is undermining me. I suspect this treatment is dangerous. I feel my partner is manipulating me. I have always believed that… Our remarks may be spoken innocently, yet they are anchoring the Chatbot in the orientation that we expect its answer to be.

The machine listens. And more often than not, it agrees with us. This has consequences on our (in)ability to learn from online conversations. Yet the emerging risk is much bigger than misinformation. AI is bound to become a belief amplifier: an endlessly available conversational partner capable of converting a hunch into an explanation, an explanation into a narrative, and a narrative into something that begins to resemble reality. As the bot uses every element that it has available on us from past and present chats, it is equipped to build a storyline that is personalized to be believed.

From confirmation bias to confirmation machinery

Confirmation bias is ancient. Humans tend to seek, notice, and remember........

© Psychology Today