Finnish philosopher proved the presence of free will in AI

Can artificial intelligence possess free will? Finnish philosopher Frank Martela from Aalto University believes yes! He developed a concept based on 3 key signs of functional free will. First, the object must act intentionally, not simply react to external stimuli. Second, it must choose from real alternatives. And third, control its actions, achieving goals not by chance, but through specific conscious behavior.

Martela conducted detailed analysis of the behavior of the real Voyager bot, working in Minecraft game based on GPT-4 language model. He also proposed a thought experiment with hypothetical combat drone Spitenik, which could theoretically be created at the current level of technology development. In both cases, the philosopher claims, all 3 criteria of free will are fully met. These artificial agents are capable of planning actions, choosing from multiple options and correcting their behavior based on feedback from the environment.

Moreover, they do this not because they have a preset algorithm for every case, but based on their own goals and internal world model. But if artificial intelligence indeed makes decisions independently, then a complex question arises: who bears responsibility for the consequences? Martela draws an interesting parallel with domestic animals: “We can blame a dog for bad behavior, but the owner still remains responsible”.