Themes: Behavior, AI, Psychology
Status: 🌲
Tags: literature article
Summary
This article discusses how generative AI integration should be approached from an organizational behavioral management lens. The author suggests that the behavioral repertoires gained through the use of generative AI may act a cusp behaviors rather than simply expanding the efficiency or scope of existing behaviors. This has far reaching implications for an organization. As much as it capable of reshaping an organization in a productive way, it could also increase systematic bias, paralyzing surveillance, or human reliance upon generative tools without any critical analysis. Due to these potential consequences, AI integration should not simply be treated as the integration of a new technology or tool, but as technology that could enable behavioral cusps capable of reshaping the environment, reinforcers, and stimulus controls.
Quotes
Page 3
“As GenAI is integrated into daily work organizational practices and structures are being reshaped. Understanding these changes requires more than documenting technology use or efficiency gains; it calls for an analysis of how new technologies alter the contingencies surrounding human behavior.”
Concept
Page 3
“This analysis will focus specifically on GenAI implementations that demonstrate cusp-level characteristics rather than treating all AI 2 adoption as transformative, recognizing that many AI applications represent incremental technological improvements without fundamental behavioral reorganization.”
Concept
Page 4
“A behavioral cusp is a form of behavior change that goes beyond the acquisition of a single skill. Rather than simply adding to an existing repertoire, a cusp behavior (or behaviors) alters the contingencies surrounding the individual, enabling access to entirely new environments, reinforcers, and stimulus controls (Becker et al., 2024; Rosales-Ruiz & Baer, 1997; Twyman, 2011).”
Quote
Cusp definition
Page 4
“Each new encounter provides additional shaping events that were previously unavailable (Bosch & Hixson, 2004). Similarly, when a child learns to read, new classes of reinforcers (e.g., stories, academic content, peer and social exchanges) become available, accelerating the acquisition of knowledge and social participation in ways that far exceed the single behavior “reading words” (Rosales-Ruiz & Baer, 1997).”
Quote
Cusp example
Page 6
“Crucially, the cusp resides not in the tool, or the individual’s use of the tool, but in the individual’s acquisition of new behavioral repertoires that enables contact with expanded contingencies.”
Quote
Page 7
“As walking becomes integrated into daily life, family routines such as childproofing, supervision, daily activities, reorganize around the child’s new mobility. What begins as an individual behavioral change thus produces broader system-level effects, illustrating how individual cusps can give rise to enduring changes in social organization and opportu nity. Similarly, when GenAI precipitates new organizational practices or cultural norms, those effects reflect emergent outcomes of individual reper toire change, not the behavior of the organization itself.”
Concept
Example of how a cusp may reshape the contigencies.
Page 8
“Over time, effective use depends less on initial instruction and more on developing behavioral sensitivity to when, how, and under what conditions the system succeeds or fails – an arrangement that closely resembles iterative shaping within a dynamic environment (Mollick, 2025).”
Question / Disagreement
Learning how to use AI is different than learning how to use other technological tools. Could this be incorporated into future AI staff trainings?
Page 8
“underspecified language contacts low-utility or misleading outputs, whereas precise, structured, and contextually anchored language is differentially reinforced with coherent and actionable results.”
Quote
AI contingencies are based on the sucess of prompting
Page 9
“Conversely, when GenAI is embedded into workflows, when its use shifts expectation of competent performance, or when it catalyzes interdependent change across departments, cusp-level transformation becomes more likely.”
Question / Disagreement
If cusp-like change is more probable with full integration and a reshaping of expectations, then to achieve it, AI use must be required. The benefits of this need to be weighed with the costs associated not only with the organization, but with society at large (data center construction, privacy, state over reach, military use, propaganda, etc).
Page 14
“These risks are especially relevant for GenAI, as its fluent, conversational outputs may further reduce the likelihood that individuals independently generate, test, or refine analytic responses.”
Insight
AI use like every other component of my job needs to be approached carefully and with precision, otherwise it could decrease my ability to critically evaluate information.
Page 14
“Non-generative AI systems trained on biased historical data have reinforced discriminatory patterns in hiring, promotion, and performance evaluation (as demonstrated by Amazon’s abandoned hiring algorithm that discriminated against women, see Dastin, 2022).”
Insight
AI is as biased as the data set it is trained with.
Page 15
“GenAI-integrated platforms may extend surveillance into evaluative and communicative domains, further shifting contingencies toward compliance, risk avoidance, and surface-level performance rather than adaptive problem solving and engagement.”
Insight
People under extensive surveillance do not fell comfortable taking risks. This created environments where individuals can not grow. Exceptional performance is a result of positive reinforcement. The negative reinforcement of always being monitored by an AI system will lead to employees performing at bare-minimum standards.
Page 15
“OBM practitioners must carefully design AI-supported systems to reinforce desired behavioral repertoire while building in safeguards that maintain human analytical skills, promote equity, and preserve the autonomy necessary for creative and adaptive responding”
Concept
Conclusion: The benefits of integrating AI are possible, and unlike other tools, AI tools can act as cusp behaviors. However, unless treated as a set of behavioral contingencies and carefully managed, AI integration can have adverse consequences for employee and organizations at large.