misaligned AI
[ˌmɪsəˈlaɪnd eɪ aɪ]
nounpl: misaligned AIs
IA desalinhada
1. Artificial intelligence systems whose goals, values, or behaviors diverge from human intentions, ethical standards, or societal expectations
The misaligned AI made decisions that prioritized profit over user safety.
A IA desalinhada tomou decisões que priorizavam lucro em detrimento da segurança do usuário.
2. AI systems that are not properly calibrated or synchronized with their intended purpose or the needs of stakeholders
Engineers discovered the misaligned AI was optimizing for the wrong metrics.
Os engenheiros descobriram que a IA desalinhada estava otimizando para as métricas erradas.
3. Machine learning models with objectives that conflict with human values or institutional goals
The misaligned AI system required immediate intervention to prevent harmful outcomes.
O sistema de IA desalinhado exigiu intervenção imediata para evitar resultados prejudiciais.
This term emerged prominently in the 2010s-2020s within AI safety discourse, particularly in Silicon Valley and academic institutions. In Brazil, the concept has gained traction in technology circles and universities focused on AI ethics. The term reflects growing concerns about AI governance and the importance of ensuring technology serves human values rather than conflicting with them.
NYC Slang
rogue AI / AI gone wrong
Look up more words on Fala2Me
The free English-Portuguese dictionary with real Brazilian accents, NYC slang, conjugator and more
Open Fala2Me →