- Today
- Holidays
- Birthdays
- Reminders
- Cities
- Atlanta
- Austin
- Baltimore
- Berwyn
- Beverly Hills
- Birmingham
- Boston
- Brooklyn
- Buffalo
- Charlotte
- Chicago
- Cincinnati
- Cleveland
- Columbus
- Dallas
- Denver
- Detroit
- Fort Worth
- Houston
- Indianapolis
- Knoxville
- Las Vegas
- Los Angeles
- Louisville
- Madison
- Memphis
- Miami
- Milwaukee
- Minneapolis
- Nashville
- New Orleans
- New York
- Omaha
- Orlando
- Philadelphia
- Phoenix
- Pittsburgh
- Portland
- Raleigh
- Richmond
- Rutherford
- Sacramento
- Salt Lake City
- San Antonio
- San Diego
- San Francisco
- San Jose
- Seattle
- Tampa
- Tucson
- Washington
MIT Study: AI Workers Barely Meet Minimum Standards, Not Replacing Humans
Research finds AI models only pass workplace tasks 65% of the time, struggling with complex work.
Apr. 3, 2026 at 7:55pm
Got story updates? Submit your updates here. ›
As AI models struggle to meet minimum workplace standards, companies must carefully integrate the technology to augment rather than replace human workers.NYC TodayA new MIT study tested 41 AI models on over 11,000 real workplace tasks and found that most AI outputs barely meet minimum standards, scoring well below 'superior' quality on complicated work. The research suggests AI is currently better suited for augmenting human workers on repetitive tasks rather than fully replacing them, as the technology still struggles with nuanced judgment and multi-step reasoning.
Why it matters
The findings counter the widespread narrative that AI is on the verge of automating knowledge work and displacing human employees. The data shows AI has significant limitations, with high error rates and an inability to match human-level performance on complex tasks. This has implications for how companies should approach AI integration in the workplace to avoid costly mistakes.
The details
MIT researchers evaluated 41 AI models from companies like OpenAI, Google, and Anthropic on over 11,000 text-based tasks drawn from official Labor Department job descriptions. Human experts then graded the AI outputs on a 1-to-9 scale, with a 7 defined as 'minimally sufficient' work. The AI models hit that 7 rating only about 65% of the time across all tasks. And the probability of an AI achieving a superior 9 rating never exceeded 50%, even with unlimited time. The models struggled most with tasks requiring multiple steps or nuanced judgment, common in skilled roles like legal services and IT. This tracks with high-profile cases of AI-generated content containing fabricated information or errors that have damaged companies' reputations.
- The MIT study was published on April 3, 2026.
The players
MIT
The Massachusetts Institute of Technology, a prestigious research university that conducted the study on the capabilities of AI models in the workplace.
OpenAI
An artificial intelligence research company that develops large language models, some of which were included in the MIT study.
The technology conglomerate, which also had AI models evaluated in the MIT study.
Anthropic
An AI research company that develops advanced language models, which were part of the MIT study.
What’s next
The researchers estimate that by 2029, AI could handle 80 to 95 percent of text-based tasks at a minimally sufficient level, suggesting the technology is on an upward trajectory but still has room for improvement.
The takeaway
The MIT study highlights the limitations of current AI models, which struggle with complex, multi-step tasks that require nuanced judgment. This suggests companies should focus on using AI to augment human workers on repetitive tasks rather than fully automating knowledge work, in order to avoid costly errors and reputational damage.
