Writing Code with AI Makes You Skip Creative Thinking — Not Dumber, Just Taking Shortcuts

Forum topic · 二一 · 2026-05-14

Summary

A study presented at a Chinese tech forum (zhichai.net) examines how LLM-assisted coding affects the creative process of programmers. Twenty programmers completed two sets of programming tasks, one with LLM assistance and one without, with the comparison focused on the ideation process rather than code quality. Results: with LLMs, the ideation phase shortened significantly (p=0.0004), programmers jumped faster to execution, and moments of creative insight dropped significantly (p=0.002). LLM-assisted code was more correct and runnable, but the volume of creative ideas stayed roughly the same. Researchers identified four human-LLM collaboration patterns; the overall trend shows LLMs help produce functional code faster at the cost of skipping divergent thinking that could yield better designs. The paper argues this is not because LLMs make programmers dumber, but because they remove friction — the obstacles that typically drive human creativity. Design implications: generative AI tools should support the creative process, e.g., prompting users to consider alternative approaches before accepting a first solution. Paper: Saghi, Huang, Chattopadhyay, arXiv:2605.13776 (2026).

"Creativity is fundamental to being human." That's the opening line of the paper, followed by a simple but brutal experiment.

20 programmers each completed two sets of programming tasks — one assisted by an LLM, one without. What was compared was not code quality (that has been studied to death), but the creative process itself.

Results:

With LLM assistance, the ideation phase was significantly shorter (p=0.0004). Programmers jumped to the "execution" stage faster.
Moments of creative insight were significantly fewer (p=0.002). Those "wait, there's a better way to do this" flashes of inspiration became rarer.
LLM-assisted code was more correct and runnable — but the quantity of creative ideas was roughly the same.

The researchers identified four human-LLM collaboration patterns supporting different problem-solving strategies. But the overall trend is clear: LLMs let programmers produce functional code faster, at the cost of skipping the divergent-thinking stage that might lead to better designs.

One participant summed it up: "Like taking the path of least resistance."

The paper's finding is not that LLMs make programmers dumber. It's that LLMs are too good at directly handing out answers. When you have a tool that can always tell you "just write it this way," you rarely feel the urge to ask "is there a better way?" Human creativity is largely driven by friction — being stuck on a problem, forced by a dilemma, compelled to take detours. LLMs eliminate a lot of that friction. But friction happens to be fertile ground for creativity.

The design implications are clear: generative AI tools should be designed to "support the creative process" rather than merely "produce correct results." For example, after offering a solution, ask "Have you considered another approach?" — or require users to compare multiple ideas before accepting the first one.

Reference paper: Zeinabsadat Saghi, Run Huang, Souti Chattopadhyay. "'Like Taking the Path of Least Resistance': Exploring the Impact of LLM Interaction on the Creative Process of Programming." arXiv:2605.13776, 2026.

Writing Code with AI Makes You Skip Creative Thinking — Not Dumber, Just Taking Shortcuts

Summary

Tags