AI has both well-intentioned and fundamentally problematic applications.
How is AI applied?
- Policing: i.e. increased breadth of surveillance. This has some issues such as worse facial recognition performance on minorities due to lack of data which leads to wrongful arrests / stop-and-search. There is a lack of accountability on AI systems.
- Military: searching for specific (human) targets Propagates bias into war: may result in more deaths. Automated defence dehumanises war, leads to loss of moral engagement, simplification of complex moral choices, easier escalation, and arms races.
- Political Control: state surveillance of opposition, ethnic & religious minorities; strengthening authoritarian states; human rights losses in privacy, freedom of speech, religious freedom. Also leads to increased censorship and disinformation.
- Revival of phrenology: disproved pseudoscience that assumes predictions of gender, intelligence, criminality and internal states can be made from facial analysis. Allows oppression of already marginalised communities. Reinforces harmful stereotypes, mis-identification, micro-aggressive ad-targeting.
Who has access to AI?
AI is prohibitively expensive, requiring lots of computation power, expensive hardware, money to train and house. There is a Big Tech monopoly over this industry. Universities can’t compete, resources used for Big Tech’s priority problems.