Google’s careers site carries open roles tied to AI-generated content in Discover, Search Engine Journal’s Matt G. Southern reported on October 11. These are job ads, not a product announcement. Publishers who depend on Discover traffic should still read them as a signal of the direction the team wants to take the feed.
Damien Andell first spotted the listings and shared them on X. A Staff ML Engineer posting describes the Discover recommendation group as one that “leads by recommending AI-generated content (AIGC) to fulfill user interests.” Generated material is framed as the team’s core job, not a side experiment.
According to SEJ, that engineering role spans quality control, topic discovery, and content understanding. The hire would build agentic flows, meaning automated multi-step pipelines, that write content for “proactive” surfaces, with Discover and Notifications both named. The posting also covers review of accuracy, timeliness, coherence, and tone, along with feedback loops. A more senior ML engineering posting on content quality in the feed describes similar duties and says the team applies such content on several large surfaces.
The design listing sets the ambition. The Principal UX role sits at Google’s Director+ level, with offices in Mountain View and New York listed. SEJ reports a U.S. base salary between $275,000 and $382,000, with bonus and equity on top. The listing aims for a “proactive, hyper-personalized concierge experience,” and the designer would help users dig deeper and turn inspiration into action, coordinating with YouTube, Shopping, Chrome, and Android teams.
One detail matters for publishers. SEJ says the same posting wants the hire to weigh AI capabilities against a lasting place for people and their sources. It also lists a background in building “value exchanges” with creators and publishers as a plus. The listing does not define that exchange. Whether publishers become partners in the feed or raw material for it depends on that missing definition.
The roles build on earlier Discover moves, all as SEJ presents them. Google launched AI previews on trending topics in October 2025 for users in India, South Korea and the U.S., each linking out to web content. Google said in August that users would be able to steer their feed by typing out their preferences. Robby Stein, who leads product for Google Search as VP, posted on September 18 about a trial of “Dive deeper.” Tapping it brings up a brief overview of a topic plus related links, with video the first format. A data science listing, as SEJ describes it, puts daily active users of Discover above 500 million, which shows why changes to how the feed fills itself carry weight.
Measurement is thin. Search Console’s Discover generative AI report tallies how often your links appeared in those AI features, split by page, country, and date. Impressions are all the report’s help page covers. Search Labs experiments are excluded, and individual AI features are not separated. Clicks from those features do not appear there.
Control sits elsewhere. A site’s participation in Discover’s AI features is decided by the Search generative AI control, the same setting that covers AI Overviews and AI Mode. Google’s help page says an excluded site’s content stays out of those generative experiences, and everyone else’s remains eligible. That is an all-or-nothing choice, and the listings hint at nothing finer.
The listings leave major questions open. None gives a timeline, and none says what portion of the feed generated content would occupy. Google Careers does not show when the roles went up, and postings can be edited or withdrawn. Nothing in them says Google plans to replace publisher content; they describe generating and recommending alongside it.
Discover traffic is already hard to forecast, and a feed that mixes generated overviews with links adds a layer between a story and a click. Teams that rely on Discover should record current Discover performance and generative AI report impressions as a baseline, so any later shift in clicks can be traced to a specific feature instead of guessed at.
Search Engine Journal, Matt G. Southern, October 11, 2026, reporting on Google careers site listings.
