"As organizations race to adopt AI - especially generative and agentic systems - the risks are becoming more complex, more unpredictable, and more distributed across everyday workflows. The challenge isn't just understanding what AI can do, but what it can break, distort, or unintentionally amplify. And according to Reid Blackman, founder of Virtue, author of Ethical Machines, and one of today's most practical voices in AI ethics, traditional approaches to "responsible AI" are no longer enough."
"Blackman argues that companies must shift from debating abstract principles to confronting something far more concrete: ethical nightmares. Instead of asking teams to memorize values like fairness and transparency, he suggests something radically simpler and far more actionable - define the nightmare scenarios, build resources to avoid them, and train people to recognize when those risks are emerging. This framing cuts through organizational hesitation, intellectual clutter, and compliance theater."
"Before founding Virtue, Reid Blackman spent a decade as a professor of ethics. What drew him to AI wasn't just its novelty - it was its intellectual difficulty. Traditional corporate ethics problems, he notes, are rarely complex in principle. But AI introduces unknown unknowns, where even well-intentioned teams can unintentionally create discriminatory or privacy-violating systems. Discrimination in an automated hiring model, for example, can emerge even when engineers actively try to prevent it."
AI adoption creates complex, unpredictable risks that can break, distort, or unintentionally amplify everyday workflows. Companies should shift from debating abstract principles to defining concrete 'ethical nightmares' and building practical resources to avoid them. Training staff to recognize emerging risks enables earlier intervention and reduces compliance theater, intellectual clutter, and organizational hesitation. As AI moves toward agentic, multi-tool systems, outcome-oriented risk management will determine which organizations innovate safely versus those that encounter avoidable incidents. Practical, scenario-driven policies provide clearer operational guidance than memorizing abstract values like fairness or transparency.
Read at Medium
Unable to calculate read time
Collection
[
|
...
]