artificial intelligence may threaten human autonomy not through direct coercion, but through thousands of convenient decisions that gradually shift control from people to machines. An analysis published on November 2 argues that AI can amplify existing human biases, particularly the tendency to prioritize immediate benefits over distant costs.
People often favor options that save time or reduce frustration, even when the long-term consequences are uncertain. AI can intensify this pattern by summarizing difficult material, drafting text, retrieving information and recommending decisions. These uses may provide immediate benefits, while any loss of competence caused by repeated delegation develops slowly and can be difficult to recognize.
The article describes this process as “extractive hybridization”: a trajectory in which AI interacts with human tendencies, incentives and institutions in ways that progressively extract value from human capabilities and the broader resources represented by people and the planet. The central concern is not every individual act of delegation, but the cumulative effect of making machine assistance the normal starting point.
Defaults are particularly influential. As Generative AI becomes embedded in email, search, operating systems, classrooms and workplaces, users may no longer decide whether to use it. Instead, they may increasingly decide when to override it. The analysis notes that simply keeping a person formally involved does not guarantee sound oversight, because people can misjudge AI-generated suggestions and may underestimate the system’s influence.
Interactions with biased systems may also reinforce human bias. Rather than merely reflecting existing prejudices, an AI system can become part of a feedback loop in which its suggestions influence people, those responses provide new behavioral data, and the system then reorganizes future choices around those patterns.
This cycle is especially significant because algorithms learn from actions that may reflect fatigue, distraction, anger or automatic behavior rather than stable preferences. When such behavior becomes training material for the systems that shape subsequent choices, impulsive actions can help establish the default environment for everyone else.
The analysis links these risks to a broader mindset that treats greater output, faster results and easier experiences as inherently better. AI can magnify that approach by generating more content, personalizing persuasion or predicting consumer desires. The technology, therefore, does not determine which goals should guide society; it strengthens the goals that people and institutions already prioritize.
An alternative approach, described as “hybrid flourishing,” would use AI to preserve or develop human judgment rather than remove every source of effort. Educational systems could ask learners to formulate their own views before providing an answer. Decision-support tools could present uncertainty and counterarguments instead of offering a frictionless recommendation. Carefully designed friction could help users reflect and reduce over-reliance.
The distinction is between effort that needlessly wastes time and effort that builds capability. The analysis concludes that the future impact of AI will depend partly on the human priorities it amplifies. Before deciding what kind of artificial intelligence to create, society may need to consider what kind of human behavior it wants that technology to reinforce.
Sources: Psychology Today.