GenAI At Work

Does GenAI Really Boost Productivity?

Is GenAI actually useful in the workplace? Researchers at the Harvard Business School just published a study that seems to answer the question with a solid YES.

The study, conducted between May and July last year, examined the use of GenAI at Proctor & Gamble. It analyzed the work output of over 775 randomly selected participants who used generative AI on their teams.

The survey has quite a bit of detail, but here are some of the key findings.
– AI significantly boosts performance – individuals using AI matched the quality of teams working without AI, effectively replicating key benefits of human collaboration.
– AI breaks down expertise silos – without AI, R&D professionals created more technical solutions while Commercial professionals developed more commercial-oriented proposals. With AI, both groups produced balanced solutions regardless of background.
– AI improves emotional experiences – professionals using AI reported higher levels of positive emotions (excitement, enthusiasm) and fewer negative emotions (anxiety, frustration) compared to those working alone.
– AI + teamwork creates exceptional results – teams using AI were three times more likely to produce top-tier solutions than the control group.
– AI saves time while improving output – individuals with AI spent 16.4% less time on tasks while producing solutions of higher quality and substantially greater length than non-AI users.

While the study shows very positive results from using AI, it is important to note that the rollout of AI was done systematically with training and clear guidelines. There is no mistaking the positive impact the GenAI had on team performance. Still, it does not match the level hyped for the predicted productivity revolution with mass job eliminations. The current level of capability makes it an effective business tool with positive ROI, but we are still a long way from “the robots taking over”. Do you think differently?

You can find the study at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5188231