AI in the room: individual spark, collective blandness?

I am using AI a lot in the training I deliver now - in fact there's not a single course or seminar I give without at least some element of AI.  I'm getting clearer now on the fantastic opportunities - but also some of the risks - that come with using it in our day jobs.

This week I have been helping people use Copilot for critical thinking.  Specifically: developing logical structures for problem solving; brainstorming solutions to questions; and scenario planning across multiple possible future outcomes.

It is truly amazing how well, how quickly, and how creatively it can help in these tasks.  In a matter of seconds, it will come up with novel, relevant, and well-structured ideas.  In a matter of minutes, training participants have been able to refine and improve the outputs to the stage where they are genuinely useful.  They have been blown away!

But - as always with AI - there are some risks.  Of course, we all know there's the temptation to take AI outputs as "gospel" and not spend the time checking they are as relevant as they can be (or even accurate!)

But there's also a risk that we lose the important diversity that comes from different human perspectives, too.  I came across a very interesting article, "Generative AI enhances individual creativity but reduces the collective diversity of novel content", that powerfully argues that while individual AI-supported outputs are generally more creative, AI-enabled content is more similar to other AI-enabled content than work produced by humans alone.  In other words, AI can sharpen our creativity as individuals, but dull our novelty as a team.

How might we address this risk?  One way might be to treat AI more like an individual team member, rather than an enabler for everyone's thinking.  For example, before meetings, workshops or problem solving sessions, ask one team member to work with AI to develop content, but ask the others to come up with ideas on their own (no AI allowed!)  Or conduct a (human) brainstorming session, and use the outputs as part of an AI prompt live in the meeting.  These approaches would seem to offer the best of both worlds: the efficiency, com

prehensiveness and clarity of AI, but with some genuine unfiltered human creativity in the mix as well.

It's certainly something I'm going to try - but what do you think?