Artificial intelligence continues to transform our daily lives, but a fictional company has just revealed a secret that could well turn conventional wisdom on its head. As AI makes its way into every sector, this unexpected revelation promises to redefine the limits of modern technology.
From improving business efficiency to revolutionizing human interaction, the implications are vast and intriguing. This captivating mystery is already attracting the interest of experts and the general public alike, hinting at fascinating prospects for the future. Find out what lies behind this surprising discovery and how it could impact tomorrow’s world.
Experiment at Carnegie Mellon: a fictitious company made up of AIs
Researchers at Carnegie Mellon University recently conducted a bold experiment by creating TheAgentCompany, a fictional company populated entirely by artificial intelligence agents. The aim was to assess the ability of these agents to fulfill a variety of roles, such as financial analysts, software engineers and project managers.
The agents, derived from technologies developed by Google, OpenAI, Anthropic and Meta, were confronted with tasks simulating the daily life of a real software company. However, the results revealed chaotic performance, underlining the current limits of AI in complex business environments.
Disappointing Performance of AI Models
The results of the experiment highlighted the significant shortcomings of AI agents in completing everyday tasks. Anthropic’s Claude 3.5 Sonnet, although the best performer, only managed to complete 24% of tasks, at a prohibitive cost of over $6 per task. Google Gemini 2.0 Flash completed only 11.4% of tasks, requiring an average of 40 steps per task.
Amazon’s Nova Pro v1 proved the least effective, with a success rate of just 1.7%. These results highlight the challenges facing AI, including a lack of common sense and limited social skills, making these technologies unsuitable for complex work environments.
Current Limits and Perspectives of Artificial Intelligence
The Carnegie Mellon experiment highlighted crucial problems, such as AI agents’ lack of common sense and poor social skills. These limitations make it difficult to integrate them into roles requiring nuanced understanding and human interaction. Although some simple tasks can be automated, today’s AI cannot replace complex human jobs.
It remains a complementary tool, useful for assisting workers with repetitive or analytical tasks, but far from an autonomous intelligence capable of handling unforeseen situations. Thus, AI finds its place in the world of work as a technological ally, rather than a substitute for human skills.
