I published this post on The Ten Commandments of AI in 2018:
The Bishop of Oxford, who sits on the House of Lords artificial intelligence committee, has come up with a ten commandments of AI:
AI should be designed for all, and benefit humanity.
AI should operate on principles of transparency and fairness, and be well signposted.
AI should not be used to transgress the data rights and privacy of individuals, families, or communities.
The application of AI should be to reduce inequality of wealth, health, and opportunity.
AI should not be used for criminal intent, nor to subvert the values of our democracy, nor truth, nor courtesy in public discourse.
The primary purpose of AI should be to enhance and augment, rather than replace, human labour and creativity.
All citizens have the right to be adequately educated to flourish mentally, emotionally, and economically in a digital and artificially intelligent world.
AI should never be developed or deployed separately from consideration of the ethical consequences of its applications.
The autonomous power to hurt or destroy should never be vested in artificial intelligence.
Governments should ensure that the best research and application of AI is directed toward the most urgent problems facing humanity.
I like them; they provide an essential dimension to this debate.
The link I originally supplied no longer works.
Now, commentator Cliff B has offered another variation on this theme:
Remember my 10 Commandments for AI? I sent them in a letter to the FT. They didn't publish it. I sent them to my MP, Anna Sabine. No reply. Pity. Here they are again, for old time's sake. I think they might have been useful.
1. AI shall serve humanity, not govern it.
2. Human authority shall prevail over machine output.
3. AI shall not decide matters of rights or liberty without human judgement.
4. Responsibility for AI use shall rest with humans, never machines.
5. People shall be told when AI affects decisions about their lives.
6. Everyone shall have the right to human review and appeal.
7. AI shall not deceive, manipulate or impersonate.
8. AI shall not enable unjust surveillance or profiling.
9. The public good shall prevail over profit or power.
10. AI shall remain a tool, never an authority.
Both are worth reflecting on. AI is not following these rules.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
Clear guidelines like these would actually benefit the AI industry, as well as the rest of us. AI systems themselves depend upon social stability. If AI contributes to widespread insecurity, inequality, or institutional distrust, the economic environment sustaining the AI industry deteriorates as well. Consumer demand weakens. Political reactions intensify. Litigation increases. Regulation becomes harsher. Social consent fragments.
And the same principle applies to taxation. A broad restructuring of how productivity gains are socially shared is necessary because AI will significantly alter labour markets.
The AI industry, along with all of us, will need a sustained stable, legitimate and democratic society as economic output becomes progressively less dependent on human labour.
Isaac Asimov’s Rules for Robots, from the short story Runaround, 1942.
1st Law: A robot must not harm humans or allow them to be harmed through inaction.
2nd Law: A robot must follow human orders unless they violate the First Law.
3rd Law: A robot must protect itself, provided this doesn’t conflict with the First or Second Law.
Later, Asimov added the “Zeroth Law,” which takes precedence:
0th Law: A robot may not harm humanity, or, through inaction, allow humanity to come to harm.
Thanks
In Asimov’s “Little Lost Robot” his protagonist, Susan Calvin, is horrified to learn that they’ve built robots with a weakened first law (they cannot harm humans, but can allow humans to come to harm). I’ve often imagined her reaction to our designs.
My company is developing an AI-based solution to help with care provision. From the outset, we decided we were not eliminating professional input on auditing but instead providing support and validation.
A reasonable aim for how to use AI may be to ‘Trust but Verify’. Well-implemented systems should be grounded in the data they’re analysing, robust against hallucinations, and exercise a suitable degree of caution.
In some cases people are happy to have work automated – when it’s a part of a larger role that’s of less interest than other parts (e.g. auditing often being seen as a ‘necessary evil’ by many managers). Other AI automation can be less favourable, such as when an AI generates the easy parts of code and leaves a developer spending most of their time on niggly, frustrating pieces like debugging and merge conflicts.
Changes in job satisfaction from the application of AI will ultimately affect how it is received and how much push-back is experienced by companies attempting to force their teams to use it. When it enables us to do more things of interest with sufficient effective competence and spend less time doing things of less interest, then it is more likely to be well received.
This positivity may apply for things like web page generation, image generation, etc, because most users of such tools are not web developers or artists. They just want to complete those peripheral tasks sufficiently well to get back to their main role.
And the feedback is? The idea sounds good.
You may be interested in the new encyclical by Pope Leo addressing AI. He invited the co-founder of Anthropic to be one of the team of advisors in writing the work.
I will take a look
These are good and fairly comprehensive. However, where is the commandment about not using power or water unsustainable at the expense of people and planet?
I know you could argue it’s implicit in others but it probably needs it’s own line.
Alas, might it be that whatever humans create that can be beneficial, can be subverted/perverted by other human beings?
Might the predatory uses of A. I. be connected with subversions/perversions of democracy?
https://www.theguardian.com/environment/2026/may/25/south-east-water-homes-kent-sussex-supply-outages
Never mind emergency bottled water for the lower-value human capital – who will save the data centres when the “incompetent” water companies fail to supply their cooling water?