AI Research
Apr 3, 2026
MIT study finds AI workers often perform at minimal standards in various tasks
Apr 3, 2026
AI Summary
Research from MIT indicates that while AI technology is improving, it often meets only basic performance standards in workplace tasks. The study analyzed 41 AI models across over 11,000 tasks, revealing that AI struggles with complex jobs and typically requires human oversight for quality output.
- A recent MIT study evaluated the performance of 41 AI models on over 11,000 text-based tasks relevant to various job roles.
- The study found that AI models achieved a 'minimally sufficient' score in about 65% of tasks, indicating they can produce acceptable work but often lack quality.
- AI's success rate for more complex tasks, requiring creativity or precision, did not exceed 50%.
- The research highlights that AI is more effective in routine tasks, while skilled roles in legal and IT fields showed lower success rates.
- Companies using AI have faced challenges, including instances of inaccuracies in automated outputs.
- MIT researchers noted that AI's performance is improving, with an estimated increase of up to 11 percentage points in success rates annually.
- By 2029, it is projected that AI could achieve 80% to 95% success in text-based tasks at the minimally sufficient level, but superior performance remains uncertain.
mitai workerstask performancehuman comparisonai improvement