Ethan Mollick argues that generative AI is already good enough to significantly boost human creativity, particularly in idea generation. He explains that AI's quirks — including its tendency to hallucinate and produce strange outputs — are actually features rather than bugs in a brainstorming context, where weird ideas can spark novel thinking and bad ideas are easily filtered out. AI also brings vast knowledge that enables unique combinations and recombinations of existing concepts.

Mollick outlines several concrete techniques grounded in creativity research. Playing with constraints involves using AI to explore extreme users, impose arbitrary limitations on solutions, or consider opposites to break out of familiar thinking patterns. "Interviewing" the AI means having it role-play as different personas to surface problem areas you haven't considered. The "Yes, and…" improv technique works well with AI to iteratively build on ideas. He also advocates going for volume, leveraging AI's ability to rapidly generate large numbers of ideas, and making it weird by requesting outputs in unconventional formats — like a Johnny Cash song or a Seinfeld script — to stimulate creative thinking.

The post emphasizes that AI outputs don't need to be perfect or accurate to be useful in ideation; their primary value lies in pushing people beyond their familiar mental patterns. Mollick frames AI as a new and powerful "source of oddness" that anyone can access for free through tools like ChatGPT.

Ethan Mollick highlights an ongoing academic debate about AI-assisted idea generation. The discussion centers on three interconnected papers: first, an experimental study showing that even the older GPT-3.5 model helped people generate higher-quality ideas; second, a response paper arguing that the AI-generated ideas tend to be highly similar to one another, raising concerns about diversity of thought; and third, a counter-response suggesting that this lack of variety may not matter since the outcomes are still demonstrably good.

The thread captures a nuanced tension in AI research — whether the quality of individual outputs outweighs concerns about homogeneity and originality. This is a key question for anyone using large language models as brainstorming tools, as it speaks to the tradeoff between reliably strong results and the risk of converging on a narrow set of ideas.