5.2 Legal and Ethical Challenges of Artificial Intelligence

1. The bias's-and-the-discrimination's: the training-data's-legacy (the historical-bias's- reproduction (the hiring's-and-the-lending's-faces (the LAW5702-U1's-technical-discrimination (the "garbage-in"'s)))); the fairness's-definitions (the impossible-trio's-lore (the trade-offs'));2. The transparency's-and-the-explainability's: the black-box's (the deep-model's-opacity (the decision's-reasons (the credit-and-the-parole's (the contest's-right (the LAW5605-U7's- reasons-requirement's-AI-face))))); the XAI's-efforts'3. The accountability's-and-the-liability's: the who's-question (the developer-the-deployer- the-user (the chain's (the "many-hands"'s)); the product-vs-the-tort's (the defective-algorithm's (the LAW5505-U's-products-face)); the autonomy's-myth's (the "the-AI-did-it"'s-rejection (the human-in-the-loop's-requirement)); the vehicles's-and-the-agents's (the accidents'-cases (the comparative's-lore));4. The privacy's-and-the-data's: the training's-harvest (the consent's-absence (the scraping's (the copyright's-too (the LAW5502-U's-AI-face))); the inference's (the intimate-prediction's (the GDPR's-profiling-faces));5. The intellectual-property's: the authorship's (the AI-generated's (the human-author's- requirement (the comparative-rulings' (the names-only)))); the training's-fair-use's (the ongoing-litigation's (the exam's-frontier));6. The labour's-and-the-dignity's: the displacement's (the jobs' (the LAW5606-U's-faces (the automation's-debate))); the deepfake's-and-the-likeness's (the identity's (the election's (the LAW5610-U8.5's-face)));7. The ethics'-frameworks' the principles'-menu (the beneficence (the non-maleficence (the autonomy (the justice (the explainability)))); the UNESCO's-2021 (the recommendation's (the ROAM's (the human-rights-oriented (the inclusive (the trustworthy)))));