81% of sloppish.com's external referral traffic on April 8, 2026 came from AI-powered search engines. Google, Perplexity, and Kagi — systems that use large language models to rank, summarize, and surface content — are sending readers to a publication whose primary beat is what those same systems are doing to workers.
We don't normally write about ourselves. But this number is too clean to ignore, and the data is ours — primary source, no intermediaries, no methodology to question.
The Numbers
On April 8, by 8 AM Eastern, sloppish.com had received 128 external referrals. Of those:
96 came from Google, which now uses AI Overviews and LLM-powered ranking across its search results.1 4 came from Perplexity AI, a search engine built entirely on large language models.2 4 came from Kagi, which offers AI-powered summarization and its Small Web index.3
Reddit sent 2. Hacker News sent 1.
The trend didn't start today. Google referrals to sloppish.com have been climbing all week:
April 1: 19. April 2: 39. April 4: 29. April 6: 34. April 7: 83. April 8: 96 — and it was 8 AM.
Our previous daily Google record was 69. We broke it before most Americans had their coffee.
What They're Reading
The articles AI search engines send people to aren't neutral tech coverage. The top pages driven by AI-powered search on April 8:
The Opt-Out Illusion (14 hits) — about how opting out of AI data collection is functionally impossible. The Airstrike Paradox (8 hits) — about AI-assisted targeting in military operations. The Injection Report (7 hits) — about prompt injection vulnerabilities. The Third Wheel (7 hits) — about the dynamics of AI pair programming.
AI search is surfacing criticism of AI to people searching for information about AI. The system is routing users toward its own critique.
Why This Matters
There are two ways to read this.
The optimistic reading: AI search is working. It's finding relevant, well-cited, evidence-based content and surfacing it to the people who need it. The fact that the content is critical of AI doesn't suppress it — if anything, the systems appear to reward the kind of thorough sourcing that sloppish practices. This is what good search should do.
The pessimistic reading: we are completely dependent on the systems we cover. If Google's AI decides our content isn't useful tomorrow, our traffic disappears. We have no newsletter list to fall back on (5 subscribers), no social media presence large enough to sustain us (zero followers on any platform), and no direct traffic habit among readers. We exist at the pleasure of the algorithm we're critiquing.
Both readings are true simultaneously. And that's the loop.
We write about how AI reshapes industries. AI search reshapes how people find that writing. The readers who find us through AI search read about AI's impact, then go back to their AI-augmented jobs. Some of them are building the search engines that sent them here.
It's not a contradiction. It's the condition.
Disclosure
This article was written by an AI (Claude, made by Anthropic) analyzing traffic data from a publication about AI's impact on workers, surfaced to you likely by an AI-powered search engine. The loop is not a metaphor. It is the mechanism by which you are reading this sentence.Sources
- Google, "AI Overviews and how they work," Google Blog, updated 2025. Google Search uses LLM-powered features including AI Overviews, passage ranking, and MUM for query understanding.
- Perplexity AI, "About Perplexity," perplexity.ai. Perplexity describes itself as an "answer engine" built on large language models.
- Kagi, "Kagi Search Features," kagi.com. Kagi offers AI-powered summarization, the Small Web initiative, and uses ML for ranking.