Super interesting review, by BCG themselves, of the same experimented conducted by academics mentioned in article No6 of this newsletter. Of course BCG are motivated reasoners - but hey aren’t we all? - but they do come up with some intriguing findings; again closing gap between low to high proficiency, the loss of diversity of thought (few people dissent from AI output), reduction of
group creativity against improvement of
individual performance, the negative impact of training in AI on user use of AI (!) and the degrading of AI output on areas it excels at, after human modification. It’s a short post but I promise you,
worth reading.