Mat Honan argues in MIT Technology Review that Google is catching up on AI search

Mat Honan argues in MIT Technology Review that Google is catching up on AI search

Mat Honan’s item in The Download was published March 17, 2025. He argues that Google’s newly announced additions of Gemini AI features to search, and search features to Gemini, look more like catching up with OpenAI’s ChatGPT than setting a new direction. He connects this to a broader pattern: the major AI players seem to be converging on agents, deep research and lightweight models. Honan says what strikes him is not how well Google’s features work, but their resemblance to ChatGPT’s offerings and their belated appearance. He takes that timing as a poor sign for Google. The newsletter does not assess how well the features work or explain what competitive consequences may follow. It points readers to a full story but does not include that story’s evidence or further argument. Whether the newly announced features will change Google’s position in AI search remains unknown from this reading.

Publication: ee6715e9-f6b6-d80b-0239-e0744fba0116

attribution — pass: The issue is one newsletter piece with its own commentary, not a bare listing. The Google search item is correctly credited to Mat Honan at MIT Technology Review, and its arguments stay attributed to him. The convergence observation originated with Will Heaven, as Honan mulls it, but Honan adopts and states it himself, so crediting him with describing it is acceptable.

I’ve been mulling over something that Will Heaven, our senior editor for AI, pointed out not too long ago

summary — uncertain: The summary faithfully conveys Honan's catch-up thesis, his focus on timing over quality, and the convergence pattern. However, central terms are left unexplained, and the definitions list is empty. These include agents, deep research, lightweight versions of models, and Gemini, which is never identified as Google's AI model and chatbot. A general reader may not grasp what Google added or what the converging features are. Smallest repair: add term-library tooltips for Gemini, agents and deep research. Optionally note that Honan builds on Will Heaven's observation.

Agents. Deep research. Lightweight versions of models.

date — pass: The March 17, 2025 date matches originalPublished at day precision. The source's own reference to March 2025 is consistent with that date. The reading presents no hindsight and makes no use of later sidebar items, such as the Deep Dive teasers about later events. The capture date is disclosed, and the evidence is a publisher page rather than an archive copy.

their belated appearance in March of the year 2025 doesn’t seem like a great sign for Google

support — pass: All three claims match their verbatim excerpts. The summary's caveats are accurate: the newsletter does not assess how well the features work and only links to the full story. The closing question follows from Honan's argument without asserting an outcome. No unsupported benefits or costs are implied.

What strikes me more than how well they work is that they are really just about catching up with OpenAI’s ChatGPT.

media — pass: No media was selected, and no verified candidates were supplied, so nothing could be misidentified.