Artificial intelligence: do you know the brain fry, or when AI overwhelms the human brain?

The consultants at the Boston Consulting Group have given a name to the phenomenon, the “AI brain fry,” the brain cooked by AI, or “the mental fatigue linked to the excessive use or supervision of artificial intelligence tools, beyond our cognitive abilities.”

After the ChatGPT revolution in 2022, this year saw the arrival of AI agents, which can perform a multitude of tasks at the user’s request, suddenly turning the user into both a conductor and a producer. “People who burn out don’t just use AI. They create 100 agents that they need to manage continuously,” recently wrote X Tim Norton from the AI integration consulting firm nouvreLabs.

A new mental burden

Many, including the Boston Consulting Group, however, refuse to use the term burnout, which often corresponds to a state of distress associated with a loss of motivation. Their study of 1,488 professionals in the United States showed a decrease in the burnout rate when AI takes over repetitive tasks.

“It’s a new type of mental burden,” suggests Ben Wigler, co-founder of the startup LoveMind AI, which works on human-AI interactions. “With these models, you have to babysit.”

For now, “brain fry” mainly manifests among computer developers, with programming being the most obvious application for AI and its agents. “The cruel irony is that the code generated by AI requires more careful scrutiny than that written by humans,” described computer engineer Siddhant Khare on his blog.

According to the Boston Consulting Group study, employees suffering from “brain fry” make 39% more major errors.

“It’s really scary”

“Relying on hundreds of lines of code written by AI is really scary because there is a risk of security flaws or simply not understanding the whole program,” adds Adam Mackintosh, also a programmer at a Canadian company.

Moreover, as Ben Wigler reminds us, activating agents is very costly in terms of rented computing power from remote IT providers. “If a model misunderstands something and embarks on a mission, it’s money down the drain,” he explains.

Adam Mackintosh also admits sometimes succumbing to the temptation to go further, losing track of time. “Many entrepreneurs work late into the night,” and “when you have productivity on this scale, it encourages you to stay even later,” says the entrepreneur, who mentions finishing days at 2 or 3 in the morning. Adam Mackintosh recalls spending 15 hours straight working on an application, with 25,000 lines of code at the end. “I couldn’t code anymore,” he says. “I was irritable and didn’t want to answer questions about my day.”

“Defining limits”

A musician and teacher, who prefers to remain anonymous, confesses to having a hard time disconnecting. “Instead of putting my brain on pause and watching a series on TV, I will spend my evening trying different things with AI.” It is worth noting that all the people interviewed by AFP see a significantly positive contribution from AI, despite the drawbacks.