- Boston Consulting Group gave AI tools to hundreds of its individual contributors in a study.
- In some cases, the consultants with AI were much more productive. In others, they made errors.
When I think of white collar workers, I think first of consultants.
Armed with MBAs and case studies, these consultants parachute into the biggest companies in the world to find efficiencies, sharpen strategy, and justify tough decisions.
There's been a lot of talk recently around how AI might impact these kinds of so-called knowledge workers. That's why I was fascinated to see the results of a study that gave hundreds of consultants access to AI.
The experiment involved 758 individual contributors at Boston Consulting Group. One group had no access to AI, and two groups were given access to GPT-4 with different approaches. The takeaways:
- AI can be incredibly powerful. There were 18 consultant tasks where AI proved capable. Across these tasks, those using AI quite simply crushed it. They completed more tasks, completed their tasks more quickly, and the work was of a higher quality.
- AI can't do everything. For tasks deemed to be "outside the frontier" of what AI is capable of, those using it were significantly more likely to make mistakes than those not using AI.
- AI benefits the lower performers most. Everyone using AI tools in the right places benefited, but it was the lower performers who did so most, allowing them to close the gap to their higher-performing peers.
The study supports earlier research that's found AI increases productivity overall while elevating the less experienced and lower performers most.
The findings also hint at how AI might impact knowledge workers everywhere. After all, the management consultant advising your CEO might just have been a part of this study.