Humans or AI? Study shows which tasks benefit most from using artificial intelligence

As artificial intelligence (AI) advances, there was a world push for people to study to work alongside it to maintain tempo with its speedy adoption in each private {and professional} settings.

Nonetheless, a current examine from MIT’s Heart for Collective Intelligence (CCI) discovered that whereas human-AI collaboration will be helpful, sure tasks might yield higher outcomes when dealt with independently.

“There’s a prevailing assumption that integrating AI right into a course of will at all times assist efficiency – however we present that that isn’t true,” Michelle Vaccaro, lead creator of the examine and a doctoral scholar at MIT, mentioned in an announcement.

“In some instances, it’s useful to depart some tasks solely for people, and a few tasks solely for AI,” she added.

To achieve a clearer understanding of when people and AI work most successfully collectively, the researchers analysed 370 outcomes drawn from greater than 100 experimental research involving human-AI mixtures throughout varied tasks.

The research coated three approaches to dealing with a process, which included people alone, AI programs alone, and human-AI groups.

The researchers discovered that whereas human-AI groups tended to outperform people working independently, they didn’t exceed the outcomes achieved by AI programs working alone.

Human-AI collaboration not at all times greatest

Printed within the journal Nature Human Behaviour, the examine additionally discovered no proof of “human-AI synergy,” which means that for sure tasks, relying solely on both people or AI might produce higher outcomes than collaboration.

However the meta-analysis discovered some particular instances the place human-AI collaboration might be simpler than people or AI alone.

“Although our principal consequence means that – on common – combining people and AI results in efficiency losses, we don’t suppose which means combining people and AI is a foul thought,” the authors wrote within the paper.

“Quite the opposite, we expect it simply signifies that future work must focus extra particularly on discovering efficient processes that combine people and AI. Our different outcomes counsel promising methods to proceed”.

The researchers recognized decision-making tasks, similar to detecting deepfakes and diagnosing medical instances, the place human-AI groups underperformed in comparison with AI working independently.

In distinction, in addition they discovered that people and AI labored higher collectively on artistic tasks, like producing new visible or written content material.

In these instances, the human-AI groups even surpassed the efficiency of the best-performing particular person people or AI programs.

The researchers theorised that this might be because of the nature of artistic work itself.

Artistic tasks, they defined, require a mixture of creativity, information, and perception, which are qualities which are inherently human and tough for AI to completely replicate. But these tasks additionally require repetitive, routine processes the place AI is extremely efficient.

Consequently, a mix of human and AI strengths might have allowed for higher outcomes in artistic work.

“Let AI deal with the background analysis, sample recognition, predictions, and knowledge evaluation, whereas harnessing human abilities to identify nuances and apply contextual understanding,” Thomas Malone, co-author of the examine and director of the MIT CCI, mentioned in an announcement.

He concluded: “As we proceed to discover the potential of those collaborations, it is clear that the longer term lies not simply in changing people with AI, but additionally to find revolutionary methods for them to work collectively successfully”.