AI Agents Make Humans More Productive, MIT Researchers Say

AI agents can make the workplace more productive when fine-tuned for different personality types, but human co-workers pay a price in lost socialization, ongoing research by MIT information technology specialists has found.

MIT researchers led by Sinan Aral, a professor at the MIT Sloan School of Management and Harang Ju, a Sloan postdoc associate, found that humans using AI raised their productivity by 60%—partly because those workers sent 23% fewer social messages.

AI agents are complex artificial intelligence models that can work autonomously to complete tasks assigned to them while using reasoning and memory. They are increasingly being deployed in the workplace: In a recent PWC survey, 79% of senior executives said AI agents were being used at their companies; 66% said they were increasing productivity. They are being deployed as executive assistants, for customer support, and for analyzing complex tax forms among other uses.

Aral and Ju created a platform called Pairit, previously known as MindMeld, to better understand how the rapid adoption of AI agents is likely to change the workplace. In one experiment, more than 2,000 participants were assigned randomly to teams with other humans while another roup was paired with AI agents. Thee groups were then tasked with creating advertisements for a think-tank.

“Humans on Human-AI teams sent 23% fewer social messages, creating 60% greater productivity per worker and higher-quality ad copy,” the researchers found. The human teams working with other humans produced higher quality images. The personality of the humans also mattered, with humans in the “conscientious” group producing enhanced images while working with open AI agents, while those in another group—branded “extroverted”—produced lower quality image and texts in their work while working with “conscientious” AI agents.

Humans working with other humans sent more messages that were social and emotional including ones to build rapport and showed concern, the experiment found.

The researchers concluded that “AI traits can complement human personalities to enhance collaboration.”

“Together, our results highlight the transformative potential of AI agents in collaborative workflows while underscoring the importance of aligning their design with human traits and task requirements,” Aral and Ju said in their paper.

Spinning Off

Devanshu Mehrotra, an analyst with Gartner, said the ability to “hyper-customize” AI agents for individual workers would be revolutionary.

“This is a preliminary entry into this type of research. This type of methodology does open up the doors for a lot of further discussions of this kind, because I genuinely think that we have only started to scratch the surface,” he said.

The researchers are now focused on using their platform to analyze negotiations between humans and how AI can augment them. They are also experimenting with AI agents for consumer support, coding, data analysis and creative writing.

Ju said their study has proved that the same kinds of AI agents might not work for every person.

“Personalized AI, we believe that that’s going to be a huge thing, he said. “So not everyone’s going to get the same AI, like you don’t work with every one of your coworkers the same. Everyone’s different, and so AI should be different too.”

Getting the right fit between AI agents and human workers “will be just as important as it is between humans, he added.

Aral and Ju have started a company called Pairium AI to build personalized agentic AI. The company promotes itself as “personalizing the Agentic Age with AI technology that adapts to your unique work style.”

“We naturally learn to adjust our styles according to our teammates,” Ju said. “That’s going to have to happen a lot with AI.”

That, he said, will lead to improved productivity, teamwork and performance.