Evidence analysisOctober 24, 2026Evidence
‘The bulk of the benefit and the threat is actually humans’

FreedomWitness Evidence Engine: This is an AI-assisted original analysis built from the linked source and available public source context. It separates evidence from interpretation and may challenge conventional framing only where the evidence supports it. It is not a copied article.
What is established
Artificial intelligence (AI) is increasingly integrated into critical fields such as medicine, education, and governance. The development and deployment of AI, particularly large-scale models, are primarily driven by industry actors with significant resources. Universities, while less central in building cutting-edge AI systems, retain an important role in defining appropriate applications and shaping ethical frameworks around AI use, especially given their long-standing traditions in humanities, ethics, and governance.
Medical AI tools reflect varying ethical priorities embedded through their design and training, which may align differently with cultural values. For example, attitudes toward patient autonomy can vary substantially by geography and professional culture, and AI tools tend to exhibit consistent “ethical north stars” that may differ from local clinical values.
There is no clear consensus on optimal integration of AI in medical education or other domains, and usage outcomes can vary depending on context and task. AI’s dual nature—its potential for benefit and harm—is closely linked to human decisions, intentions, and governance. Autonomous AI systems exist but have limitations, and human agency remains critical in guiding AI’s impact.
What the source claims
The Harvard Gazette article presents an interview with Isaac Kohane, inaugural chair of Harvard Medical School’s Department of Biomedical Informatics and founder of the Human Values Project. Kohane highlights several key assertions:
- Most technical advances in AI over the past 15 years have emerged from industry, notably from organizations able to develop large models that confer competitive advantage.
- Universities contribute uniquely by determining which AI applications are appropriate and developing ethical governance frameworks attuned to cultural and regional differences.
- AI models trained globally do not uniformly embrace the same ethical principles; they vary in emphasis on autonomy, beneficence, justice, reflecting diversity in human values.
- The debate over AI in medical education lacks standardized best practices; AI can be both a helpful learning aid and a crutch that may weaken readiness if over-relied upon, requiring further pedagogical innovation.
- Comparison of AI and healthcare quality should consider the reality that many patients do not receive optimal medical attention; AI might offer benefits even if it is imperfect relative to the very best clinicians.
- The “bothness” of AI—the simultaneous potential for positive and negative outcomes—is fundamentally rooted in human choices and applications rather than autonomous machine will or inevitability.
- Kohane calls for public engagement led by universities to foster informed conversations, resist alarmism or simplistic utopian visions, and assert human agency in shaping AI’s future.
What may be missing from the usual framing
This perspective reframes the AI discourse away from singular narratives of existential threat or flawless benefits by emphasizing the complex human-social factors in AI’s design, deployment, and governance. It pushes against the notion that large AI companies alone should guide ethical standards globally, noting the inadequacy of corporate-led ethics divorced from cultural diversity and public accountability.
Also highlighted is the nuanced heterogeneity within clinical ethics and human values across cultures, which AI models reflect, complicating one-size-fits-all assumptions about medical AI decision-making. This contrasts with some mainstream portrayals of AI as monolithic or universally aligned with Western ethical frameworks.
Lastly, the piece underscores a need to move beyond framing AI’s risks and benefits purely as technological or economic issues, instead situating them within longstanding questions in humanities and governance, thus expanding the conversation to include values, responsibilities, and pluralistic dialogue.
Competing evidence or uncertainty
While Kohane’s insights are grounded in empirical studies through the Human Values Project and his medical informatics expertise, broader uncertainties and debates remain, including:
- The extent to which university ethics frameworks can influence or counterbalance dominant industry AI developers in a competitive technology landscape.
- How to operationalize cultural diversity and plurality of ethical norms in universal AI governance mechanisms without fragmenting standards or sowing confusion.
- The effectiveness and long-term consequences of AI integration in education and medicine, where empirical outcome data are still emerging and context-dependent.
- The capacity of current AI technologies to behave autonomously in impactful ways absent human oversight, a topic of ongoing technical and ethical scrutiny.
- Potential structural biases, conflicts of interest, or commercial incentives within both industry and academia that might shape perspectives or slow consensus building.
Readers should note that this single interview offers valuable but not definitive evidence on these complex issues and that expert views and empirical data continue to evolve.
Why it matters
The framing presented by Kohane is significant because it shifts attention from fear or hype about AI to focus on human responsibility, ethical plurality, and institutional roles in shaping AI’s trajectory. This invites policymakers, educators, clinicians, and the broader public to engage in thoughtful, culturally aware discourse and governance.
Recognizing AI as an extension of human choices underlines the critical importance of transparency, inclusivity, and accountability in AI development and deployment. It also highlights universities as essential sites of for critical reflection, research, and public engagement rather than mere passive recipients or implementers of AI technology.
Given AI’s rapid integration into sensitive areas like healthcare and education, understanding these human-centric dimensions is vital for building trust, mitigating harms, and maximizing equitable benefits.
What to verify next
- The detailed methodology and scope of the Human Values Project studies on AI ethical alignment across cultures.
- Empirical outcomes of AI applications in medical education, particularly longitudinal studies comparing different deployment strategies and learner outcomes.
- Comparison studies of AI clinical decision support performance relative to average and expert human practitioners across diverse healthcare settings.
- The efficacy and adoption of university-developed AI governance frameworks in influencing broader policy and corporate practice.
- Independent assessments of AI autonomy levels in military and medical contexts to clarify human versus machine agency.
Readers interested in deeper insight should consult the original Harvard Gazette interview with Isaac Kohane for comprehensive context and nuanced discussion.
Source & verification
Primary source: Harvard Gazette · October 23, 2026
FreedomWitness links to the original source so readers can inspect the underlying material and judge the context for themselves.
Continue reading

Loss to follow-up during tuberculosis treatment and 5-year mortality in Brazil: A target trial emulation
Loss to follow-up during tuberculosis treatment and 5-year mortality in Brazil: evidence dossier What is established Tuberculosis (TB) remains a leading cause of infectious…

Disagreement doesn’t have to be toxic
Evidence Dossier: The Nature and Practice of Disagreement What is established Disagreement is a common and unavoidable part of social life, extending beyond politics…

Business professor? Perhaps more ‘detective of human behavior.’
Lauren Cohen: Investigating the Social Dynamics Behind Financial Markets This evidence dossier explores insights from an interview and profile published by the Harvard Gazette…