An artificial intelligence leads a company and creates software with almost no human help

A university experiment tested the capacity of AI and left several unknowns about what this technology could do when organizing a company

Artificial intelligence (AI) remains a great unknown. A university experiment tested the capacity of this technology to form a company, lead other bots and create software, almost without human help. The result was surprising.

The test was carried out by researchers at Brown University in the United States. For this, the ChatGPT 3.5 model was used and a fictitious company called ChatDev was created, with its divisions in design, coding, testing and coding.

Each section was assigned a bot, in creative areas of the company and the work began. Everyone communicated in each section for work, according to their functions and with minimal human intervention.

According to a report by Genbeta, the purpose of this company was to create a board game called Gomoku. All bots are assigned to their sections and debate which model to choose for the software. In this case, Python is being debated as a programming language due to its ‘simplicity and readability’.

This example has allowed us to see the capacity of AI to have organizational functions, not just mechanical work, with the ability to detect vulnerabilities and correct them in just seven minutes.

The truth is that the test was in a fictitious company. It remains to be seen whether AI can have similar reactions in a real field.