Shawn Smucker’s “Please Use AI” landed because it pokes at a deeply awkward truth: a lot of people in tech don’t actually disagree about AI’s usefulness—they disagree about what it says about us if we use it.
<> The article appears to have struck a nerve./>
That much is obvious from the response. The discussion around the post reached 727 points and 379 comments on Hacker News, which is the kind of engagement reserved for topics that feel bigger than a single essay. In other words, this wasn’t just content. It was a trigger.
My read: the appeal of a post like this is not that it reveals some hidden technical insight. It’s that it gives permission to stop pretending AI is a niche curiosity. The examples surfaced in the article summary—like using AI for a meal plan instead of calling a friend who cooks—suggest a deliberately cheeky argument: stop overthinking the tool and start using it in ordinary life.
That framing is clever because it sidesteps the usual sterile debate about model quality and jumps straight into behavior. Should we use AI for writing, planning, summarizing, brainstorming, and other low-stakes cognitive labor? The ecosystem around AI tools already answers that question with a loud yes. Summarization and writing products are aggressively marketed as productivity layers for everyday work, which shows how mainstream this category has become.
But there’s a catch: mainstream does not mean trustworthy.
The strongest criticism of AI-assisted workflows is still the same one, and it hasn’t gone away just because the tools got better: they can flatten nuance, omit context, and produce confident nonsense. Even the surrounding product ecosystem leans on words like summary, key points, and efficiency, which are useful promises but not guarantees of accuracy.
That’s why posts like Smucker’s become polarizing. On one side are people who see AI as a pragmatic layer on top of knowledge work. On the other are people who hear a demand for adoption and translate it into a demand for lower standards. Both reactions are understandable, and both are partly wrong.
The productive position is less ideological: use AI where it saves time, but do not outsource judgment.
- Use it to compress noise, not to replace original reading.
- Use it to draft, not to decide.
- Use it to accelerate thinking, not to simulate having thought.
That distinction matters for developers especially. AI is already useful for summarizing docs, generating first drafts, and roughing out research, but the moment you confuse output with understanding, you’ve created a very expensive illusion.
So yes, use AI. Just don’t confuse enthusiasm with discernment. The strongest version of the argument is not that AI should replace human effort; it’s that refusing to use it at all is starting to look less principled and more performative.
