Exploring the Risks of General-Purpose AI: The Role of the Brain’s Reward Mechanism and Nearsighted Goals in Processes of Decision-Makings

The introduction of general purposes AI (such as, but not only, LLMs- large language models AI) in our daily lives draws everyone’s attention due to its almost unlimited possibilities. Concerning the risks AI brings to us humans, collectively and individually, we argue that especial attention should be given to the vulnerability of our processes of decision-makings. These processes depend on many things, and we focused on two. First how AI can take advantage of things like our brain’s rewards mechanism. Second, we address the question of how the type of goal, if nearsighted or farsighted, is relevant for our rewards mechanism, due to its fast production of dopamine, acting as a weak spot for an easier manipulation of our decision by AI. We conclude with three suggestions for reducing our exposition to AI’s manipulation.