Armando Da Silva Vieira, Lecturer of Applied AI In May 2026, researchers from Oxford, Google DeepMind, OpenAI, Anthropic, and Stanford published a paper (https://arxiv.org/html/2605.10310v1) with an unusually candid premise. AI alignment, they argued, has been dominated almost entirely by preventing harm — safeguards, controllability, compliance. Necessary, but incomplete. Like a psychology that only studies illness […]

Artikkel avaldati algselt Novaatoris. Ehkki tehisaru teeb üha rohkem inimeste elu mõjutavaid otsuseid, ei suuda see neid alati inimesele arusaadavalt põhjendada. Tartu Ülikoolis kaitstud doktoritöö osutab, et inimesed ootavad masinalt realistlikke ja usaldusväärselt mõjuvaid seletusi. Tehisaru teeb järjest rohkem otsuseid, mis mõjutavad inimeste elu. Seda kasutatakse haiglates diagnooside toetamisel, pankades laenutaotluste hindamisel ning hariduses õppijate […]

Mohamed Abdelrahman shares his journey as a PhD student in the Data Systems Group at the University of Tartu, where he focuses on Automated Machine Learning (AutoML) and aims to make AI more trustworthy and accessible. His story was introduced by Professor Jaak Vilo, Head of the Institute of Computer Science. Describe your thoughts when […]

Lecturer of Applied AI Armando Vieira shares his thoughts on Medium, an online publishing platform. The dominant discourse on AI alignment rests on a quiet but catastrophic assumption: that morality can be imposed on systems that do not experience anything. The debate unfolds as if values were transferable objects — encoded, optimized, constrained — rather than lived […]