Despite the productivity benefits of agentic AI, recent research from Carnegie-Mellon University shows it struggles with daily operations. Their mock company, TheAgentCompany, demonstrated that current AI agents could autonomously complete only about 30% of tasks. While they manage simpler tasks reasonably well, complex, long-term tasks remain out of reach. Experts believe AI agents are still extensions of predictive text rather than fully autonomous problem solvers, indicating they aren't yet suited for software development or intricate operational roles despite optimism about their potential.
This paints a nuanced picture on task automation with LM agents -- a good portion of simpler tasks could be solved autonomously, but more difficult long-horizon tasks are still beyond the reach of current systems.
While AI agents occasionally performed simple, isolated tasks well, the study makes it clear that they can't yet handle the kind of complex, dynamic work that humans excel at.
The researchers concluded that current AI is best described as a sophisticated extension of predictive text -- good at pattern recognition, but lacking true understanding, adaptability, and independent problem-solving skills.
Although AI may not be ready for heavy operational lifting, technology leaders see great advantage in deploying these agents for key business areas.
Collection
[
|
...
]