Chasing "perfect prompts" with AI resembles addictive behavior.

Intermittent reinforcement in AI can lead to compulsive dependence.

Recognize AI usage patterns to avoid compulsive dependency.

The promise of AI is everywhere, but that promise also portends peril. The problem with AI is that it doesn’t always give us what we want because of the way it’s applied. It’s a tool, not a panacea for everything, everywhere, all the time.

Most people use AI as an answer engine—asking questions, generating lists, filling the gaps in research, and the like. Used properly, generative AI prompts are a series of instructions that build upon one another to create an outcome. Therein lies the issue at hand.

Enter Intermittent Reinforcement

To review, intermittent reinforcement is when an action receives a reward, but that reward is random, not guaranteed. In the case of AI, we often fall into the trap of putting something out there—what, in AI-speak, is called a ‘prompt’—anticipating a concrete return (read: reward). The thing is, generative AI doesn’t work that way, because it relies on three things: the database(s) upon which it draws, the decision tree parsing the information in those databases, and, most importantly, context. Hence, intermittent reinforcement.

To put a finer—and Skinner-informed—point on it, we keep pressing the bar, waiting to get what we expect. The problem is that expectation is misplaced, because it’s not based in cause and effect, and can lead to a cycle of compulsive dependence. Simply put, we keep pressing the bar--or writing, re-writing, and re-writing the prompt—waiting for that reward.

Generative AI as a Category of Addiction

When we begin to think about AI as a prospective category of addiction, we need to consider addiction as compulsive dependence. Blending the notions of intermittent reinforcement with compulsive dependence, we find ourselves down an almost inextricable rabbit hole. Not getting what we want, we keep shaping the prompt and clicking the mouse until we get some approximation of our needs met, while likely not getting what we are genuinely seeking, but, nonetheless, accepting some approximation. Then we do it again…and again, and again, and again.

In some ways, addiction is simple. We engage in some behavior that ‘alters’ us—deflecting stress, or mood, or whatever else we find less than tolerable—and that has a physical impact, whether the immediacy of alcohol, or substances, or the less tangible, hormonally-informed responses associated with gambling, shopping, or sex.

In any event, we still land on intermittent reinforcement leading to compulsive dependence. Substances don’t last—so, the attendant ‘fix’ doesn’t last—and we keep on keeping on; returning to our panacea of choice in order to escape that which ails us, whether determinate or indeterminate.

If we consider the AI prompt as our drug of choice, it’s a fair gamble that the ‘perfect prompt’ we are pursuing pushes us toward some altered state through endorphins, oxytocin, or just straight-up adrenaline. We keep pressing the bar, seeking that perfect prompt to reach a perfect end.

© 2026, Michael J. Formica, All Rights Reserved

Ferster, C. B., & Skinner, B. F. (1957). Schedules of reinforcement. Journal of the Experimental Analysis of Behavior, 1(1), 1–50. https://doi.org/10.1901/jeab.1957.1-1

Hackenberg, T. D. (2018). Token reinforcement: Translational research and application. Journal of the Experimental Analysis of Behavior, 110(2), 257–274. https://doi.org/10.1002/jeab.461

Peele, S. (1985). The meaning of addiction: Compulsive experience and its interpretation. Lexington Books.

There was a problem adding your email address. Please try again.

By submitting your information you agree to the Psychology Today Terms & Conditions and Privacy Policy

© Psychology Today