Ethical and Philosophical Questions Surrounding AI

Philosophy might not produce definitive answers. It can, however, ask difficult and profound questions: questions that can re-frame issues, expose assumptions, and challenge us to think deeply and critically.

Some of the following questions touch upon a number of concepts that have a long tradition in philosophy, for example, justice, autonomy, moral obligation, and rights. However, some others touch on relatively new concepts, such as privacy, personal data, robot rights, and non-human decision-making.

Decision-making

• Should an AI system make decisions that affect human lives?

• Are there any sorts of decisions an AI should not be allowed to make?

• Should there always be a human in the loop?

• What is valuable about a human in the loop?

• Is a human error more acceptable than system errors?

• Should we hold AI systems to the same standards as human decision-makers, or hold them to a higher standard?

• Is a black-box process acceptable, i.e. where there is a lack of transparency and explanation?

• Should we prefer better performance or explicability, if there is a trade-off between them?

• What does it mean to explain a decision?

• What makes a decision fair?

• Who is ultimately held responsible for the decisions made by an AI?

• Does accountability require transparency?

Personal Data

• What kind of rights, if any, do we have over our personal data?

  • What right do we have over personal data that involves other people, e.g. transaction history?

• How does personal data relate to personal identity?

  • Is personal data more akin to property or an extension of the self?

  • Can I own someone else’s personal data?

• Is data a public good?

• Can we meaningfully consent to giving away our personal data, if doing so is effectively compulsory for civic life?

• Why is privacy valuable?

• Should AI aim to preserve privacy?

Justice and Ethical AI

• What is data justice?

• Must AI systems counterbalance historical injustice or bias?

  • Should AI reflect or improve society?

• How can AI undermine human flourishing?

  • Should it be allowed to?​

• Is diversity important in AI development?

• Why should we demand that AI be ethical?

  • What about AI means that it has this high standard compared to other technologies?

• Why ought AI be beneficial to humanity?

Trust and Safety

• What is trust?

• What makes something trustworthy?

• What is the point of making an AI system trustworthy?

• Should we treat trust as a means to an end, or an end in itself?

• What makes a system or product safe?

• How safe should an AI system be?

• What would be the sufficient precautions, evidence, or tests that would ascertain its safety?

To AI or not to AI

• When ought we use AI systems?

• At what point does it become unethical to not use AI?

• Who has an obligation to make use of the best available resources, including technology?

• Should certain AI systems never exist?

• Are some uses of facial recognition always morally unacceptable?

• Do we risk a new era of physiognomy?

• Are some uses of affect recognition always morally unacceptable?

• Should some things always remain private and personal?​

• Should I be cognitively enhanced by AI?​

• Would this be fair?​

Availability

• Should AI tools and systems become widely available?

• Should there be compulsory training or education on AI?

• Are there risks to democratising AI?

• Are there risks to monopolising AI?

Warfare

• Should we ban the development and use of autonomous weapons?

• If there is a human in the loop, is it more acceptable to use such weapons?

• If the weapons are only semi-autonomous, is this more acceptable?

• If civilian casualties are reduced as a result of using autonomous weapons, does this make it morally acceptable?

Workers rights

• What moral rights do developers have over the subsequent use of the AI systems they helped create?

  • Given the often piecemeal nature of AI development, is this right diluted?

Robot Rights

• What is autonomy?​

• How is a human autonomous?

• Can an AI be autonomous?

  • Can an AI be autonomous in the same way as a human?

• If an agent is autonomous, do we owe that agent moral obligations or rights?

• Should we ever give an AI rights?