I thought it valuable for any of us involved in AI implementation as it contains some key questions to ask:

• What is the actual business problem you’re trying to fix?: Usually, the question executives ask is “How can we use AI?” Instead, consider, “Where are we slow, wasteful, error-prone or stuck?” AI amplifies what you have and must be congruent with it. He notes that good starting points are boring ones such as where do decisions bottleneck; where do people spend time on low-value work; and where do customers experience friction?

• What work do we want humans to stop doing?: Most AI strategies talk about adding more while very few specify subtraction. “If AI is supposed to help, what exactly should it replace? Drafting? Summarizing? Triage? Forecasting? Quality checks? If the answer is ‘everything, eventually,’ you’re not doing strategy – you’re doing science fiction,” he says.

• Do we trust the data this will be trained on?: He notes that AI doesn’t hallucinate out of nowhere. It reflects the quality of what you feed it. So ask uncomfortable questions about what your data really represents – how it was created, filtered and rewarded.

• Who carries responsibility when AI goes wrong?: If, for example, an AI system rejects a job candidate or flags a customer as risky, who owns the outcome? If no human is accountable, he says you’re not innovating but are “responsibility laundering.”

ENJOY!
AI Implementation G&M May 4 2026.docx