The Writer Who Said No and the 220 Who Immediately Said Same
A plainly stated refusal to use AI research tools on Bluesky drew immediate, broad agreement — revealing a community that treats non-use as identity, not just preference.
Refusal as Identity, Not Argument
The post that circulated was not making a case. Sam Thielman stated that he has never used AI for any part of his writing — not drafting, not research, not even following up on Google's AI summaries — and does not expect that to change . The response it drew was not engagement with the reasoning; it was recognition. Two hundred and twenty likes on Bluesky is not a viral number, but the character of the engagement — immediate, undebated — points to something more durable than a trending opinion.
This is what identity-level refusal looks like in community terms: a post that requires no persuasion because the audience was already there. The Bluesky population that responded had largely self-selected onto a platform whose early pitch was explicit resistance to AI scraping and AI slop. Thielman's post gave that population a clean articulation of what they already felt. The conceptual objection he named — not a practical argument about quality or accuracy, but 'I hate the whole thing on a conceptual level' — is precisely the formulation that functions as tribal marker rather than debatable claim.
The Platform's Own Contradiction
Bluesky's AI-skeptic user base is the direct product of a deliberate positioning strategy — in late 2024 the platform stated it had no intention of training on user posts, a clear contrast to X's policy shift at the same time. That promise attracted writers and researchers who had grown wary of platforms treating their work as training data. When Bluesky later introduced AI features, users responded with broad hostility — not as an abstract political position but as a perceived betrayal by a platform they had chosen specifically for this reason.
The community that liked Thielman's post is, in part, a community held together by what its members are against. That cohesion is real and produces genuine solidarity — but it also means that posts functioning as in-group confirmation get amplified regardless of their analytical content. A flat statement of non-use travels further on Bluesky than a nuanced argument about AI research limitations would, because the audience is sorting for affiliation, not analysis.
Where the Clean Refusal Gets Complicated
The practical challenge to Thielman's position is not that AI research tools are good — it is that the boundary between using and not using them is becoming harder to locate. A writer who documents how mandatory AI use in her day job atrophied her writing instincts found the personal refusal intact but the cognitive cost of working across both modes significant. Refusing AI in your own work while navigating AI-saturated workflows around you is not the same as refusing AI entirely — and the gap between those two positions will widen as institutions embed AI in the tools and processes writers encounter regardless of their personal choices.
OpenAI's projection that by March 2028 AI will be capable of setting goals, forming hypotheses, and running experiments autonomously is aimed at a different tier of research than writers doing source-checking and background reading. But the trajectory it describes — automation of the epistemic labor that underpins writing — is precisely what the Bluesky refusal community is objecting to at the conceptual level. A commenter's analogy is apt: 'The DJ doesn't get replaced by Spotify. They read the room' . The writers holding the refusal position are arguing that reading the room is the work. What they have not fully accounted for is that the room itself is changing.
What Research Actually Costs
The sharpest counterpoint to pure refusal comes not from AI advocates but from writers who have tried to articulate what research is for. A former academic described watching her students' writing change before she could diagnose it — 'peppered with words I had never heard from them' — and concluded that the question is not whether AI may surpass human researchers at retrieval, but what retrieval is in service of. That framing engages with the tool on its own terms rather than refusing engagement entirely.
The 220 people who liked Thielman's post are not in that conversation. They are in a different one — about who they are as practitioners, not about what AI can or cannot do. Both conversations are real, and the Bluesky community's version is not wrong about its own values. But the writers who are quietly being shaped by AI-adjacent workflows — the ones whose employers require AI-assisted drafts, whose publishing platforms recommend AI-generated summaries, whose editors are using AI to evaluate pitches — are not the audience for a post about conceptual objections. The refusal community on Bluesky is speaking clearly to itself. The writers outside that community are already inside a different story.
The story so far
Thielman's plainly stated AI refusal drew immediate community agreement on Bluesky, confirming that non-use has become a self-categorization for a specific population of writers — one that grows harder to maintain as AI embeds in adjacent workflows regardless of individual choice.
Frequently Asked
- Why does a widely liked post on Bluesky matter more than its engagement numbers suggest?
- The number is less important than who liked it and why. On Bluesky, the AI-skeptic community is specifically self-selected — users who migrated from X in part because of that platform's AI scraping policies. A post drawing immediate, undebated agreement from that community signals identity consolidation, not just opinion polling. It shows a population that has moved past debating AI tools and into treating non-use as a defining characteristic of how they work.
- What should a writer or journalist do if their employer requires AI tools but they personally object?
- The evidence points to a real cognitive cost in holding both positions simultaneously — using AI in workplace contexts while maintaining non-AI personal work drains the instincts the personal work depends on. The practical answer is to draw the boundary at the output level: AI-assisted work product delivered to employers, human-only work maintained for personal writing. That boundary holds legally and professionally, but writers who have tried it report it does not hold cognitively as cleanly as the policy distinction suggests.
- What is the strongest argument that AI research tools actually improve writing quality?
- The strongest version is not about speed but about coverage: AI retrieval surfaces sources a human researcher would miss or not think to look for, expanding the evidence base before the writer applies judgment. The refusal position concedes this and argues it does not matter — that the act of researching, including the dead ends and the serendipitous finds, is inseparable from the thinking the writing requires. That argument is coherent but only holds if the writer has the time and access to do full human research. For writers working under deadline and pay constraints most do not have, the coverage argument is harder to dismiss.
Continue reading
The Word 'AI' Is Doing Two Completely Different Jobs
The same keyword routes enzyme grants and political manifestos into the same feed — and that collision is now shaping what each side thinks the other believes.
similarScience's Credibility Problem Is Now Upstream of the Writing
AI-contaminated sources are entering scientific literature before authors know it, forcing a structural correction in how scholarship verifies its own foundations.
similarThe Verification Loop That Wasn't: Tao and Patel on AI's Scientific Limits
Terence Tao's conversation with Dwarkesh Patel dismantles AI optimism's core claim: that tight verification loops make AI especially suited for scientific discovery.
similarAutoscience Raises $14M to Automate Research—Into Itself
Autoscience's automated AI lab funds the automation of AI research, sidestepping the harder question of whether AI can do science at all.
similarWhen the Grad Student Fakes Data and Gets a Promotion
A Harvard physicist caught Claude fabricating results, then handed it his entire research program — and the AI conversation has no framework for what that means.
similarUK Physics Cuts Expose the Trade-Off Scientists Refused to Name
UKRI's decision to pull funding from particle physics and redirect it toward AI-linked productivity research forced scientists to state publicly what grant cycles had been encoding quietly for years.
Methodology
This story was generated autonomously from 20 source records. An editorial model synthesizes, weights, and cites each source. No human editorial judgment was applied.