Human Counterpoint in Practice

A city transit authority uses AI to redesign its bus network

The system analyzes millions of journeys, identifies the least-used routes, calculates the potential savings, and recommends eliminating three of them.

The analysis is excellent.

The routes really are expensive. The buses really are often half empty. Removing them really would make the network more efficient.

But one route connects a low-income neighbourhood to the nearest hospital. Another is used mainly by older people who no longer drive. The third is the only practical way for night-shift workers to get home.

The AI has answered the question it was given:

How do we make the bus network more efficient?

A Human Counterpoint might ask:

Efficient for whom?

A company uses AI to shortlist its future leaders

The system studies the characteristics of executives who have succeeded in the past, then identifies current employees who most closely resemble them.

The shortlist looks perfectly sensible.

The candidates have the right experience, performance history, and behavioural profile.

But who asks why every “high-potential” candidate looks remarkably like someone the company has already promoted?

The system is identifying patterns associated with past success.

A Human Counterpoint might ask whether the company is choosing the best leaders for the future—or simply reproducing the kind of leader it already knows how to recognize.

An AI tutor helps a student improve their test scores

The system notices every hesitation, offers a hint at exactly the right moment, adapts the difficulty level, and keeps the student moving forward.

The results improve.

The student completes more lessons and earns higher marks.

But who asks what happens when the help becomes so good that the student is rarely left alone with a difficult problem?

The system is optimizing learning performance.

A Human Counterpoint might ask whether the student is becoming more capable—or simply more dependent on never having to struggle.

A hospital uses AI to reduce appointment no-shows

The system identifies patients statistically most likely to miss appointments and recommends requiring deposits from them.

The numbers make sense.

But who asks whether the policy will disproportionately deter people with unstable incomes, unpredictable work schedules, or caregiving responsibilities?

The system is solving for attendance.

A Human Counterpoint might ask whether the hospital is unintentionally making access harder for the very patients most likely to need flexibility.

A company uses AI to redesign jobs for maximum productivity

The system analyzes workflows, removes duplicated effort, consolidates responsibilities, and recommends a leaner organizational structure.

The productivity gains are real.

But who asks what happens when every role is optimized so tightly that nobody has time to help a colleague, experiment with an idea, mentor a junior employee, or notice something outside their formal responsibilities?

The system measures output.

A Human Counterpoint might ask what kinds of human value disappear when there is no slack left in the organization.

A bank uses AI to decide which small businesses should receive loans

The system looks at cash flow, sector risk, repayment history, location, and thousands of other signals.

Its predictions are more accurate than those of many human loan officers.

But who asks whether the bank is gradually withdrawing capital from unconventional businesses precisely because they do not resemble businesses that succeeded in the past?

The model is reducing risk.

A Human Counterpoint might ask whether the institution is also reducing possibility.

A media company uses AI to personalize every reader’s news feed

Engagement rises. People receive more of the stories they are likely to read and fewer stories they routinely ignore.

The recommendation engine is working beautifully.

But who asks whether a news organization has responsibilities beyond maximizing relevance?

The system asks:

What will this person click?

A Human Counterpoint might ask:

What might this person need to encounter?

A retailer uses AI to make its customer service radically more efficient

The system resolves most complaints instantly, predicts what customers want, offers refunds automatically, and eliminates much of the need for human support.

Customer-service costs fall sharply.

But a small group of customers keeps returning with problems the system technically resolves but never really understands.

A parent trying to replace a gift before a child’s birthday. Someone whose order was meant for a funeral. A customer who doesn’t actually want compensation; they want somebody to acknowledge what happened.

The AI has optimized resolution.

A Human Counterpoint might notice that in some moments the real requirement isn’t resolution at all.