AI can give people a much greater capacity to learn, make things and help one another. It also raises serious questions about the conditions under which that capacity is created and used. These principles offer a starting point for deciding how to engage, what to ask of the technology, and where to draw your boundaries.
Understand what you are choosing
AI is already part of the world in which decisions are being made about work, education, culture and public life. Understanding it gives you a better basis for making your own choices and questioning those made on your behalf. Examine its capabilities, limitations and the interests of the organisations providing it. Give its useful possibilities and its risks serious attention. You can find value in particular uses while remaining deeply sceptical about the direction of the technology as a whole. Engagement can include experimentation, criticism and a considered decision to abstain. Leave room for people to reach different conclusions.
Make room for useful work
Saving time can make something possible that would otherwise remain undone: researching a problem, helping someone express themselves, building a useful tool. Think about what that additional capacity is for. It might support a livelihood, make room for research, ease a difficulty or allow you to help someone you would otherwise struggle to support. Small improvements to people’s lives have value, and they can accumulate. Be curious about what becomes possible, then pay attention to whether it actually helps. The ability to produce more gives you choices about where to direct your effort. Choose purposes worth giving that effort to.
Keep the costs in view
Consider the resources used, the people whose work contributed to the technology, and the consequences of how it is deployed. Environmental costs, pressure on livelihoods, inequality and psychological effects belong in the same discussion as usefulness. So do deception, dangerous applications and the possibility of consequences far beyond an individual task.
Ethical judgement also happens in small decisions: whether a task calls for AI, how much context to supply, what to generate and when to stop. Weigh those choices according to their purpose and likely consequences. Keep questioning the companies and institutions responsible for larger decisions. Personal usefulness gives you a reason to engage; it leaves the wider questions open.
Build context, keep your independence
When you work across models, tools and conversations, continuity depends on the context you carry between them. Keep track of the purpose, relevant sources, decisions and corrections that make the work intelligible. Good context includes why something matters and what remains uncertain. Make it possible to move that understanding into another tool without rebuilding the whole situation from memory.
Treat context as something you actively maintain. More information brings more to manage, and usefulness depends on relevance as well as volume. Choose models for the work they help you do, and retain the ability to reconsider that choice.
Stay responsible for the result
Read closely. Ask whether an answer preserves what you meant, whether its claims are supported, and whether its confidence is justified. A clear account may still contain a misunderstanding. Correct it, ask again, and examine what changed.
For factual work, follow the references. Establish whether they exist and whether they support the claims made from them. Another AI can help with this investigation, but agreement between models still needs evidence behind it. Preserve uncertainty where the evidence is incomplete. Be especially careful in unfamiliar territory, where a convincing explanation can feel like expertise. Seek the judgement the task requires. When you publish, act on or pass along an answer, take responsibility for that choice.
Be honest about how things are made
Give people a fair understanding of AI’s contribution to work you present. Consider what they reasonably need to know in that setting, and make the process clear enough for them to judge it. Give credit where it is due, including for the ideas, sources and creative work on which a result depends.
Value the originality, practice, craftsmanship and emotional investment involved in human creativity. Experiment with the forms AI makes possible, and describe them honestly. Presenting generated work as wholly human obscures something people may care about. Where you work with AI to develop your own ideas, remain attentive to which choices and contributions are yours and which came through the tool.
Respect other people’s boundaries
Your willingness to share information about yourself does not settle what is appropriate to share about someone else. Consider whose information you are supplying, how personal it is, and what that person would reasonably expect. Removing a name may leave enough context to identify them. Attribution in public work and disclosure of private circumstances call for different judgements.
People also have different views about receiving or participating in AI-assisted work. Take those views seriously. Discuss its use where it matters to their involvement, and give them room to refuse. Limited knowledge about what companies hold or how they use it makes deliberate choices about further disclosure especially important.
In this series
Previous: Why I’m rebuilding Niche Clever. Next: Thinking with machines: cognition, authority and responsibility.
This piece was developed in conversation with AI. The ideas, experience and decisions are mine; AI helped draft and revise the wording, which I reviewed and accepted.