This makes sense because they are seeing similar things and solving similar problems. When I told Will about this he said “it almost feels as if there is a lack imagination, right?” It does. It does. I’ve spent much time reporting over the last yearGoogle has taken direct aim at the intersection of AI and search by adding new AI features to Gemini’s search, as well as adding search features to Gemini. What struck me more in using both was that they are just about catching OpenAI’s ChatGPT. Google’s late appearance in 2025, March, is not a good sign.
Take AI Mode which was announced on March 5It’s cool. It works well. It’s a pretty close follow-up to what OpenAI already did. Don’t be misled by the name. Google already had AI Overviews, but AI Mode goes deeper. The company explained in a post that “This new Search Mode expands what AI Summary can do with more sophisticated reasoning, thinking, and multimodal abilities so you can get assistance with even your most difficult questions.”
Instead of a brief summary with links out, AI will dig in to offer more robust responses. You can also ask follow-up questions, which AI Overviews does not support. It feels like a natural evolution, so much so that I’m curious why it isn’t already widely available. For now, the feature is only available to those with paid accounts and even then, only through Search Labs’ experimental sandbox. Why wasn’t this available last summer, for example?
The second update is that it added search histories to its Gemini Chatbotand promises that even more personalization will be coming. Google says that “personalization allows Gemini, starting with Search to connect with your Google Apps and Services, will provide responses that are unique insightful and directly address you needs.” It’s been compared with Perplexity, a generative AI search engine startup.
Neither feature feels fresh or new. Neither feature feels innovative. ChatGPT has been using user histories to deliver results for years. Gemini can also remember certain things about you. However, I find it a bit surprising that Google took so long to integrate signals from other products. There are privacy concerns, but we’re dealing with an opt-in service.
Another thing is that ChatGPT, at least from what I’ve seen so far, is better at this. Here’s an example. I asked both: “What do You Know About Me?” ChatGPT responded with a really insightful and thoughtful profile based on our interactions. These aren’t only the things I explicitly told it to keep in mind about me. It’s largely based on the context of the various prompts that I’ve given it. It knows what kind of music I enjoy. It knows a few details about my film preferences. Some of it is oddly delightful. Google’s response was more suited to an advertiser than someone looking for tailored results. (“You enjoy comedy, music, podcasts, and are interested in both current and classic media”)
It’s clear that Google, and its parent company Alphabet,caninnovate in many areas. Take, for instance, this week’s announcement by Google DeepMind about Gemini Robotics. Or ride in a Waymo! Can it do this while still maintaining its core business and products? This is not the only problem that big legacy tech companies face. Microsoft’s AI strategies have been heavily reliant upon its partnership with OpenAI. John Gruber, a long-time Apple analyst, has a scathing critique of Apple.
Google is a company with billions of users, and a lot of money. It can leverage its current base in ways OpenAI, Anthropic (which Google owns a large chunk of) and Perplexity are not capable of. I’m also fairly convinced that it will have some painful days if it doesn’t become the market leader in this space, instead of a follower. Astra will be here soon. Let’s wait and see what happens.