AI can compare options, organise evidence and point out a condition you missed. It can also produce a recommendation that sounds firmer than the information deserves.
The difference between assistance and decision-making matters. The more a choice affects money, rights, health, employment or another person, the less sensible it is to hand over the final judgement.
Who should decide?
Compare laptops against your stated criteria
AI can organise the evidence. You choose the trade-off.
Decide whether someone should receive a public service
This needs proper human responsibility, policy and a route to challenge.
List questions to ask a financial adviser
Useful preparation, not a replacement for advice.
Give AI the comparison, not the authority
AI is useful for turning your criteria into a clear comparison.
You might ask it to compare three laptops against price, weight, battery life and accessibility needs. You still check the specifications and decide which compromise you prefer.
The AI does not experience the consequence of being wrong. You do.
State your criteria before seeing the recommendation
If you ask "Which is best?", the AI must invent a meaning for best.
List the factors that matter and mark any non-negotiable condition. Ask the AI to show where evidence is missing and avoid ranking options that fail a required condition.
Compare these options against my criteria. Show the evidence for each score, mark missing information and do not choose a winner. Tell me which trade-offs I need to decide myself.
This keeps the judgement visible.
Watch for values disguised as facts
A recommendation may assume that lower cost matters more than support, or that speed matters more than privacy. Those are value choices.
Ask the AI to name the assumptions behind its recommendation. Change one important priority and see whether the result changes. If it does, the decision rests on that priority rather than an objective winner.
Keep high-impact decisions with people
Do not use a general chatbot as the final decision-maker for recruitment, eligibility, medical treatment, legal action, credit or another high-impact outcome.
AI may support research or administration in controlled settings, but proper policy, professional judgement, transparency and routes to challenge the outcome still matter.
For personal health, legal or financial questions, use AI to prepare questions and understand terms. Take the decision to a qualified source.
Record why you chose
For an important choice, keep the criteria, sources, uncertainties and reason for the final decision.
This makes later review possible. "The AI recommended it" is not a useful decision record.
Nova 9 view
Let AI organise the evidence and reveal the trade-offs. Do not let a confident recommendation hide who owns the decision.
If the outcome affects rights, safety or another person, human responsibility is not an optional final check.
Sources and last checked
- ICO, Explaining decisions made with AI: https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/artificial-intelligence/explaining-decisions-made-with-artificial-intelligence/
- NIST, AI Risk Management Framework: https://www.nist.gov/itl/ai-risk-management-framework
- UK Government, Data and AI Ethics Framework: https://www.gov.uk/government/publications/data-ethics-framework
- Last checked: 14 September 2026
