Odgovorno AI znači razvijanje i korištenje AI-ja etično i sigurno — obraćajući pažnju na pravednost, transparentnost, privatnost, odgovornost i sigurnost. Kako AI postaje moćan i sveprisutan, graditi ga odgovorno je važna obaveza s pravim konsekkvencama.
Ključni principi odgovornog AI
✓ FAIRNESS → avoid bias/discrimination; treat groups equitably (see AI bias)
✓ TRANSPARENCY/explainability → understand and explain AI decisions (esp. important ones);
avoid opaque 'black box' for high-stakes uses
✓ PRIVACY → protect personal data; responsible data use; consent
✓ ACCOUNTABILITY → who's responsible for AI decisions/harms? humans accountable, not 'the AI'
✓ SAFETY/reliability → AI works as intended, doesn't cause harm; guardrails; testing
✓ HUMAN OVERSIGHT → human control over important decisions (don't over-automate critical choices)
