Editorial: The ethics and behavioral economics of human-AI interactions
Bibliographic record
Abstract
Although some patterns already documented for interactions with previous generations of technologies are likely to extend to the current wave of AI, some of its features warrant specific examination. In particular, the ability of AI systems to continuously learn from new data and experiences means that they can evolve over time and even in real time, offering contextually relevant interactions and providing information that are tailored to the individual user's needs. On the one hand, this changes the performance expectations of the user, but on the other hand, it makes the outcomes less predictable, and the process more opaque, than in the interaction with older generations of automated agents. In essence, the special quality of AI lies in its mimicry of human learning processes and its adaptability to the user. This feature opens a space for strategic interactions on the both sides: Human users may adjust their behavior to generate desirable outcomes, for example, to affect individualized pricing; AI agents might adjust their behavior to increase engagement, for instance, by offering the information that the user is more likely to like, thus potentially fostering and amplifying biases, creating echo chambers, and spreading disinformation.These peculiarities raised questions and concerns not for a distant future; they are immediate and pressing as AI technologies become more capable and widespread. How, for example, is cooperation achieved when humans interact with "artificial agents"? What is different or similar as compared to human-human interactions? Do people display similar or different behavioral tendencies and biases (other regarding preferences, time preferences, risk attitudes, (over)confidence, etc.) when interacting with artificial agents as compared to humans? What are people's attitudes toward the use of intelligent machines for certain tasks or functions? What moral concerns does this raise? What are the reasons for any potential opposition to the reliance on AI-operated machines for certain tasks?Behavioral economics offers a lens to understand the nuanced ways in which interacting with AI affects human behavior. The papers in this special issue highlight the breadth of questions to be addressed: from the role of human personality traits for the hybrid interactions, to reliance on technology, intergroup dynamics and immoral behavior. The findings from these studies as well as from many ongoing research efforts remind us that this interaction is not a simple case of mechanical replacement but a fundamental transformation of the decision landscape. AI's influence on human behavior is intricate and often counterintuitive. The presence of AI alters the context in which decisions are made, the information that is available, and the strategies that are employed.Various foundational methods in behavioral economics, such as laboratory and field experiments, have been employed to provide causal evidence on the topic. These methods effectively abstract from and control for potential confounding factors that might be challenging or unfeasible to isolate using observational data. In addition, new tools -such as field-in-the-lab experiments with a learning factory -allows investigating real-world interactions in a controlled environment. Taking stock of existing evidence and theoretical contributions, moreover, conceptual analyses can offer unique insights from a number of the regularities documented in previous studies.The interaction with AI is dynamic and evolving due to the rapid pace of technological change. Although the exact sizes of the estimated effects might be context-specific and may change from one generation of a technology to another, we can and should study underlying behavioral regularities that are persistent and shape the general framework of the interaction with technology.The overarching narrative is clear: the rise of AI is not just a technological or economic phenomenon, but a behavioral one. The research presented here is united by a common goal: to navigate the ethical and economic implications of our deepening relationship with AI. The insights gleaned from these and many other studies to come can help pave the way for a future where AI and human behavior co-evolve in a manner that is beneficial and, above all, humancentric.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.042 |
| Meta-epidemiology (narrow) | 0.004 | 0.001 |
| Meta-epidemiology (broad) | 0.005 | 0.004 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.004 | 0.006 |
| Scholarly communication | 0.010 | 0.007 |
| Open science | 0.006 | 0.002 |
| Research integrity | 0.028 | 0.031 |
| Insufficient payload (model declined to judge) | 0.015 | 0.010 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".