Amara's Law is an observation made by futurologist Roy Amara, stating that people tend to overestimate the short-term impact of new technologies while underestimating their long-term effects. AI’s near-term impact on leadership development is being overestimated because the hard parts of leadership development are not about information or insight. They are behavioural, contextual, relational, and political.
The current evidence base for AI-led leadership coaching and development is embryonic, uneven, and frequently confined to narrow outcomes such as structured goal attainment rather than durable behavioural change at work.
Why AI looks overrated in leadership development today
Leadership development has always struggled with a "last mile" problem: people learn the skills, but they don't always use them. Research shows that while training works, its success hinges on the company’s culture and design, not just the curriculum.
If the workplace environment doesn't support growth, AI won't be a magic fix. Simply generating more content or providing 24/7 coaching won't bridge the gap if the organizational context remains the same. Technology can’t automate a culture that isn't ready to change. Learning is easy; sustained behavior change is hard. Context determines ROI. Skills only "stick" where the culture allows. AI is a tool, not a cure. It can't overcome a broken learning environment
Kotliar’s typology of AI hype describes recurring claims about impact, speed, adoption, and technical attributes that precisely match the claims common in AI for L&D. Examples include “coaching for everyone, now” or “personalised leadership at scale” or “objective leadership measurement”. These create inflated management expectations and the inevitable tension when AI systems are expected to deliver more than their design and data allow.
Amara’s law: Overstating AI’s ability to develop leaders
Goal setting
In one prominent line of research, an AI chatbot coach designed for goal attainment appears capable of producing goal-attainment improvements comparable to human coaching under certain conditions. But goal attainment is a narrow slice of leadership development; it is closer to structured self-regulation than to complex interpersonal leadership, political judgement, or culture shaping. There seems to be an erroneous extrapolation from the results on goal-attainment to broader claims such as “AI can do executive coaching”. It is an example of sloppy thinking.
What about coaching?
Decades of research confirm that coaching works, but its success depends heavily on how it is designed and the environment in which it happens. When it comes to AI, the evidence remains a mixed bag. One study suggests that AI can match a human's ability to build a connection in a single, simulated session. However, real-world workplace experiments show the opposite: human coaches consistently outperform AI agents in building trust, sparking genuine insights, and helping employees actually reach their goals.
Ultimately, while AI can simulate a conversation, it struggles to replicate the "relational layer", the deep, trusting human connection that drives real change. Because humans still rate higher for confidentiality and goal attainment in practice, AI is currently better at delivering content than managing the complex emotional work of coaching.
Measuring impact
Leadership capability is notoriously hard to measure, and automated leadership assessment research notes that much existing workplace leadership measurement still relies on pre/post self-reports, which weakens causal inference about true skill acquisition. This is one reason “AI pilots” can look successful on engagement dashboards while failing to show behavioural transfer or organisational impact.
Transfer of learning
Research shows that training only leads to lasting change when the program design, the individual's traits, and the work environment, specifically support, reinforcement, and the opportunity to practice, are all aligned. This reality directly challenges the claim that AI coaches can "scale leadership development" on their own. No matter how sophisticated an AI’s feedback is, it cannot overcome a lack of manager support or poor psychological safety; if the surrounding organisational system doesn't incentivise growth or offer stretch opportunities, the technology will fail to translate learning into sustained behavior change.
Confidently wrong!
Current AI models pose significant risks as leadership development tools, starting with the danger of "confidently wrong" guidance. In high-stakes areas like conflict resolution or sensitive people decisions, an AI providing polished but incorrect advice can be more damaging than a simple factual error. Because these models often amplify harmful social biases and lean heavily toward Western cultural values, they may promote a narrow, skewed version of "good leadership." As I have shown in other articles, the concept of good leadership is a constantly evolving topic. In a global business context, this makes AI bias a critical governance and performance issue rather than just a theoretical ethical concern.
There’s a spy in the camp
Finally, there are serious concerns regarding ethics, privacy, and the risk of "development-as-surveillance." As leadership tools merge with people analytics and algorithmic management, the stakes for employees become incredibly high. Research warns that these systems can be life-changing when they get things wrong, and the potential for harm only grows as the technology becomes more powerful. When "coaching" starts to look like constant monitoring, the very data meant to help leaders grow can instead be used to control or unfairly penalise them.
Amara’s law: Underestimating AI’s ability to develop leaders
The long-term value of AI in leadership development isn't just about "smarter" bots; it’s about a fundamental shift in the economics of growth. By moving away from passive content consumption, AI can facilitate "deliberate practice" at scale. This means structuring repeated, low-cost scenarios with immediate feedback and spaced reinforcement—methods already proven to help skills stick. When paired with immersive simulations like VR and Mixed Reality, which show significant gains in behavioral and cognitive outcomes, AI creates a high-fidelity "flight simulator" for leaders to practice complex interpersonal skills before applying them in the real world.
Beyond practice, AI enables a shift from rare, episodic feedback to continuous assessment. Instead of relying on a single annual 360-degree review, organisations can use AI to analyse meeting interactions, written reflections, and project data to provide real-time "developmental telemetry." This allows for earlier interventions and a much clearer measurement of ROI. However, the goal is not to replace humans but to reconfigure the coaching ecosystem. As emerging standards from organisations like the ICF and NIST suggest, the future is a hybrid model where AI acts as infrastructure, supporting human coaches and ensuring safety and transparency rather than acting as a total substitute.
Finally, the rise of AI introduces a new core competency: human-AI teaming. Research warns that simply adding AI to a decision-making process can actually worsen outcomes if not managed correctly. Future leaders will need to be trained specifically in "trust calibration", knowing when to lean on algorithmic advice and when to override it. As international regulations like the EU AI Act make these systems more auditable and trustworthy, the path forward involves using tools like multi-agent simulations to create varied, low-cost training environments. This ensures leadership development is no longer a one-time event, but a continuous, data-driven, and socially responsible process.
Implications for leaders and L&D professionals
Based on the evidence, leaders should treat AI-driven development as a series of strategic interventions that must prove their value before scaling. Here are four key recommendations for a disciplined implementation:
Shift Metrics from Engagement to Transfer:
Move beyond measuring "user satisfaction" or "app usage." Instead, evaluate AI programs based on behavioral change and on-the-job outcomes. Use mixed-method assessments combining validated surveys with simulation-based data to track whether training actually translates into workplace performance.
- Design Hybrid Coaching Systems:
Rather than replacing humans, build an architecture that plays to the strengths of both. Use AI for high-frequency tasks like goal reinforcement, scenario practice, and session preparation. Reserve human coaches for high-stakes areas requiring deep trust, nuanced sensemaking, ethical judgment, and relational repair.
- Treat Governance as a Catalyst, Not a Constraint:
Recognise that robust oversight is a prerequisite for trust and scaling. L&D and HR should partner with legal and security teams to establish clear boundaries between "development" and "evaluation." Prioritise data minimisation, bias monitoring, and human escalation to prevent "helpful" analytics from becoming harmful surveillance.
- Develop Human-AI Collaboration as a Core Competency:
Stop treating AI literacy and leadership as separate tracks. Because human-AI teams can actually underperform without proper coordination, organisations must explicitly train leaders in trust calibration (knowing when to override an algorithm) and accountability in AI-mediated decisions.
References
Aboumoussa, L. (2024). Leadership development in the age of artificial intelligence (Working paper). Harvard Kennedy School, Mossavar-Rahmani Center for Business and Government
Baldwin, T. T., & Ford, J. K. (1988). Transfer of training: A review and directions for future research. Personnel Psychology, 41(1), 63–105.
Barger, A. S. (2025). Artificial intelligence vs. human coaches: Examining the development of working alliance in a single session. Frontiers in Psychology, 15, 1364054.
Bar-Gil, O., Ron, T., & Czerniak, O. (2024). AI for the people? Embedding AI ethics in HR and people analytics projects. Technology in Society, 77, 102558.
Barthakur, A., Kovanović, V., Joksimović, S., Zhang, Z., Richey, M., & Pardo, A. (2022). Measuring leadership development in workplace learning using automated assessments: Learning analytics and measurement theory approach. British Journal of Educational Technology, 53(6), 1842–1863.
Blume, B. D., Ford, J. K., Baldwin, T. T., & Huang, J. L. (2010). Transfer of training: A meta-analytic review. Journal of Management, 36(4), 1065–1105.
Bödding, R., Schriek, S. A., & Maier, G. W. (2025). A systematic review and meta-analysis of mixed reality in vocational education and training: Examining behavioral, cognitive, and affective training outcomes and possible moderators. Virtual Reality, 29, Article 44.
European Union. (2024). Regulation (EU) 2024/1689 (Artificial Intelligence Act). Official Journal of the European Union.
Gallegos, I. O., Rossi, R. A., Barrow, J., Tanjim, M. M., Kim, S., Dernoncourt, F., Yu, T., Zhang, R., & Ahmed, N. K. (2024). Bias and fairness in large language models: A survey. Computational Linguistics, 50(3), 1097–1179.
Giermindl, L. M., Strich, F., Christ, O., Leicht-Deobald, U., & Redzepi, A. (2022). The dark sides of people analytics: Reviewing the perils for organisations and employees. European Journal of Information Systems, 31(3), 410–435.
Göndöcs, D., & colleagues. (2025). Uncovering the dynamics of human–AI hybrid performance: Review of empirical studies (2018–2024). International Journal of Human–Computer Studies.
Harvard Business Impact. (2025). Fast, fluid, and future-focused: Building the collective intelligence of humans and machines (2025 Global Leadership Development Study research report).
Howard, M. C., Gutworth, M. B., & Jacobs, R. R. (2021). A meta-analysis of virtual reality training programs. Computers in Human Behavior, 121, 106808.
Kanbakan, A., Bozdereli Berikol, G., İlhan, B., Altıntaş, E., & Doğanay, F. (2026). Artificial intelligence based resuscitation simulation: A pilot study of a novel approach to team leadership training. BMC Medical Education, 26, 183.
Kotliar, D. M. (2025). Can’t stop the hype: Scrutinizing AI’s realities. Information, Communication & Society. https://doi.org/10.1080/1369118X.2025.2531165
Lacerenza, C. N., Reyes, D. L., Marlow, S. L., Joseph, D. L., & Salas, E. (2017). Leadership training design, delivery, and implementation: A meta-analysis. Journal of Applied Psychology, 102(12), 1686–1718.
Liang, W., and colleagues. (2022). Holistic evaluation of language models. arXiv preprint.
Logg, J. M., Minson, J. A., & Moore, D. A. (2019). Algorithm appreciation: People prefer algorithmic to human judgment. Organizational Behavior and Human Decision Processes, 151, 90–103. https://doi.org/10.1016/j.obhdp.2018.12.005
McKinsey & Company. (2025). The state of AI: How organizations are rewiring to capture value (McKinsey Global Survey report).
McKinsey Global Institute. (2024). A new future of work: The race to deploy AI and raise skills.
Milanez, A., Lemmens, A., & Ruggiu, C. (2025). Algorithmic management in the workplace: New evidence from an OECD employer survey (OECD Artificial Intelligence Papers No. 31). OECD Publishing. https://doi.org/10.1787/287c13c4-en
National Institute of Standards and Technology. (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0) (NIST AI 100-1).
National Institute of Standards and Technology. (2024). Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1).
OpenAI. (2025). Why language models hallucinate (technical report).
Passmore, J., & Daly, J. (2026). Are AI coaching agents learning friend or foe? A qualitative study of learning and development leaders’ perceptions of AI coaching agents using UTAUT2. International Journal of Training and Development.
Passmore, J., Tee, D., Palermo, G., & Rutschmann, R. (2025). Human coaches and AI coaching agents: An exploratory quasi-experimental study of workplace client attitudes. Journal of Work-Applied Management. https://doi.org/10.1108/JWAM-02-2025-0032
Passmore, J., & Woodward, W. (2023). Coaching education: Wake up to the new digital and AI coaching revolution! International Coaching Psychology Review, 18(1), 58–72.
Tao, Y., and colleagues. (2024). Cultural bias and cultural alignment of large language models. PNAS Nexus, 3(9), pgae346.
Terblanche, N. (2020). A design framework to create artificial intelligence coaches. International Journal of Evidence Based Coaching and Mentoring, 18(2), 152–165. https://doi.org/10.24384/b7gs-3h05
Terblanche, N., & Cilliers, D. (2020). Factors that influence users’ adoption of being coached by an artificial intelligence coach. Philosophy of Coaching: An International Journal, 5(1), 61–70. https://doi.org/10.22316/poc/05.1.06
Terblanche, N., Molyn, J., de Haan, E., & Nilsson, V. O. (2022). Comparing artificial intelligence and human coaching goal attainment efficacy. PLOS ONE, 17(6), e0270255. https://doi.org/10.1371/journal.pone.0270255
Vaccaro, M., and colleagues. (2024). When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour.
Vogel, B., Reichard, R. J., Batistič, S., & Černe, M. (2020). A bibliometric review of the leadership development field. The Leadership Quarterly.
World Economic Forum. (2025). The Future of Jobs Report 2025 (skills outlook).
Zhang, R., and colleagues. (2024). Systematic review and meta-analysis of AI-based conversational agents for promoting mental health and well-being. npj Digital Medicine, 6, 236.