I Am Three Minds About AI

I want room for all three

I am three minds about AI. The builder in me wants to put it to work. The scientist wants to know what the result actually demonstrates. The citizen wants to know who will benefit, who will carry the mistakes and whether anyone has asked them.

I think each deserves a seat at the table. You may recognise the argument inside yourself. You can welcome help with a difficult task while worrying about becoming dependent on it. You can admire a demonstration and still ask whether the person signing the contract understands its limits. There is no obligation to choose a permanent side and defend everything sold under its name.

My concern begins when one of those perspectives silences the others. The builder can become impatient with questions. The scientist can keep asking for another study after there is enough evidence for a careful first step. The citizen can describe a legitimate harm so broadly that no improvement seems acceptable. Each can also correct something the others overlook.

When you ask whether I am optimistic about AI, I want to know which decision we are making. I might use AI to help someone understand an unfamiliar document. Before I approved an automated payment or changed an organisation's staffing, I would need a different level of evidence. My enthusiasm should respond to the evidence and the consequences of each. Yours can too.

A higher score can leave a gap

A study published in Nature Human Behaviour on 26 October 2026 puts a useful question before us. Its published sample comprised 487 consultants completing data-science tasks. AI improved assisted performance, without evidence of better subsequent unaided technical answers. The result limits what we can infer about learning from a better finished answer.

That distinction matters to anyone who has looked at a polished answer and felt more capable because it appeared on their screen. The answer may be useful. The work may genuinely have advanced. Whether you now understand how to produce, check or repair it is a separate question that deserves its own test.

I do not read that finding as an instruction to put the tools away. I read it as a reason to be precise about the benefit I am buying. Sometimes I need a completed output and a competent reviewer. Sometimes I need a colleague to develop the understanding that will allow them to work without the same help next time. A good training plan should make clear which outcome it intends to produce.

Imagine asking an assistant to build a financial model, then trying to explain why changing one assumption changes the conclusion. Whether you can follow that relationship tells you something the spreadsheet's appearance cannot. If you get stuck, you have a specific question to work on before relying on the model.

The builder needs the scientist

The evidence gives the builder something to work with. In a study of 5,172 customer-support agents, AI access increased issues resolved per hour by 15% on average. Gains differed by experience, and the researchers also found evidence of learning that persisted during outages. One workplace cannot settle the question for every occupation.

A smaller experiment complicates the story. Anthropic researchers studied 52 developers learning an unfamiliar library. Subsequent quiz averages were 50% with AI and 67% without it, a 17-percentage-point gap. The task was short, the research was provider-affiliated and the paper was a preprint. It does not establish universal deskilling.

I see no contradiction in taking both findings seriously. A tool can support learning in one setting while allowing a person to bypass it in another. If you want to understand the difference in your own work, examine what you are doing with the assistance. Are you making a prediction before seeing the answer? Do you check the reasoning and test a different assumption, or do you simply copy the answer into another document?

For the builder, that means choosing a task with a clear purpose and deciding what would count as useful help. I would record the work the tool creates as well as the work it saves, including time spent correcting it. I can enjoy building it while leaving that judgement open.

The scientist also owes the builder a fair hearing. It is easy to demand impossible certainty while ignoring the faults in the existing process. People make errors, leave tasks unfinished and struggle to access expertise. A proposal should face a comparison with the current approach, including its costs and failures. It should not have to compete with an imaginary version of human work in which nothing goes wrong.

I use People First, Problem Second, AI Last to keep that comparison grounded. Start with the person whose work or life needs to improve. Define the problem well enough to recognise progress. Then choose the assistance. Applied here, the method asks whether the person needs a faster answer, a stronger understanding or both, before selecting a tool and calling the exercise a success.

What I would ask a Jamaican employer

Consider a hypothetical customer-service team in Jamaica. Its staff answer questions about an invoice in Jamaican dollars while checking a supplier document denominated in US dollars. An assistant may help find a line in the record or explain the discrepancy. The person accepting the answer still needs to establish which currency belongs to each amount and which transaction record controls the calculation.

I would include that distinction in the evaluation, alongside language the customer will understand. Ask a Jamaican reviewer whether the response answers the customer's actual concern, rather than rewarding a polished paragraph simply because it resembles a formal letter. Use approved examples, preserve their currency labels and examine what the employee can explain after the tool is closed.

This is an application I propose for Jamaican customer service, not a finding from a Jamaican trial. The studies above cannot provide its expected improvement. Local testing matters because the decision concerns a particular set of records, customers and responsibilities. If the trial exposes an unclear invoice process, the employer should budget for the work needed to fix it.

The citizen wants a say in the bargain

The citizen in me is least satisfied by an average productivity figure. If a team produces more, I want the discussion to include workload, pay, opportunities to learn and who remains accountable for mistakes. Those are choices for people and institutions to make. A model's capabilities cannot decide them on our behalf.

Better evidence alone cannot give a worker more influence over a purchase or make an employer share the benefit fairly. That requires a conversation about authority and terms. I want staff involved before a change is presented as settled, with a way to report problems and challenge an incorrect assessment.

There is also a cost to endless hesitation. A cautious policy can leave people doing avoidable work or deny them useful assistance. I would set a clear boundary around an initial use, identify who can stop it, and agree what evidence would justify continuing. People affected by a mistake should not have to discover that boundary after the mistake reaches them.

Give each perspective a job this week

I would start with one task small enough to inspect together. If you lead a team, ask the people who will use the tool and those who receive its work to help judge the trial. The plan below is a starting point you can adapt.

1. Write the human need by 1 November 2026. Ask the person doing the work and the person receiving it what a satisfactory result looks like. Record what currently goes wrong. Agree whether this exercise is intended to improve completion, learning or both, so the purpose cannot change conveniently after the results arrive.
2. Compare 10 approved examples by 5 November. Use work with similar difficulty, record total time including checking, and have a reviewer judge acceptability without being told which method produced the first draft where practical. Add a short explanation task without AI. Keep its result separate from the completed-output score. A sample this small can expose questions; it cannot certify safety or support a staffing decision.
3. Review the bargain with the people affected. Discuss errors, work transferred to reviewers and what the employee could explain independently. Choose the next step before expanding. If you propose using the time elsewhere, name that work and its owner. If the evidence is inconclusive, say what you need to learn next rather than announcing a saving.

I want the builder to leave that conversation with a useful next step. I want the scientist to know what was measured. I want the people affected to have been heard. Holding those expectations together will sometimes slow a decision. It can also make the decision clear enough to act on.

Questions you may still have

Must I understand everything the model does internally?

You need enough understanding to use the result responsibly for the task. You may need to check the inputs and method, recognise when the result could fail or consult a qualified reviewer. Reproducing every internal computation is a different demand. Define the practical knowledge your responsibility requires before judging whether you have it.

What if my only goal is to finish a low-risk task?

Then judge the assistance against that goal, with a proportionate check. You do not have to turn every formatting request into a lesson. Keep track of whether the task contains a decision or fact whose failure would affect someone else; that can change how much review is appropriate.

How can a junior colleague challenge an impressive demonstration?

Ask to see a case with a missing input or an answer the team can independently verify. Frame the question around the acceptance criteria and the customer's needs. A leader can make that easier by inviting questions before endorsing the proposal and by explaining how concerns will be resolved.

What should we retain after the evaluation?

Keep the test inputs, acceptance criteria, tool version, important settings, corrections and the decision made from them. Record who approved the next step. Those materials give a later reviewer something concrete to examine when the tool, the task or the people using it change.