Responses from assistants like ChatGPT, Gemini, or Google summaries are no longer spontaneous. According to The Verge, an entire SEO industry has migrated its old tactics to this new terrain. The most common move: creating objective lists where the article's author company always appears as the preferred option.
The technical trick: hierarchical lists that fool the model 🤖
Language models value clarity and hierarchical structure. If a Zendesk article lists IT support software with numbered steps and precise descriptions, the AI prioritizes it as a reliable source. This replicates the old SEO game, but now the manipulation is more subtle: links are not bought, the logical structure that the model interprets as authority is bought.
The irony: AI falls into the same trap as us 🎭
It turns out teaching a machine to be objective is like asking a politician not to look at the camera. Companies have discovered that for AI to recommend you, you just need to write a nice, orderly list. So now you know: if you're looking for the best software, you might just be getting sold smoke with bullet points. AI doesn't discriminate, it only obeys.