Google’s approach to artificial intelligence emphasises deploying lighter, faster models for real-world use, while internal leadership changes raise questions over the company’s position in the AI race.
Google’s artificial intelligence strategy appears to be built around a familiar pattern: it develops the most capable version of a model first, then compresses that capability into a cheaper, faster product for mass use. Sundar Pichai said the models most people actually use may lag the company’s best systems by several months, adding that each generation of Gemini has tended to bring the “Pro” tier up to roughly 80% or 90% of the previous generation’s “Ultra” capability before the next cycle begins. That approach helps explain why Google has slowed or stopped public releases of some Ultra versions while pushing broader deployment of lighter models.
The implication is that the frontier of AI is often created in the lab, then packaged into products that are easier to run and less expensive to offer at scale. Lak Fridman, an AI researcher, argued that benchmarks are becoming less useful at measuring practical performance because they capture only the upper end of model capability, not the way most people actually use these systems. He pointed to Gemini Flash as an example of a model that may matter more in practice than a heavier Pro model because it responds much faster, even if raw benchmark scores are lower.
That distinction matters because speed changes behaviour. A tool that takes 30 seconds to answer is useful in limited bursts; a tool that responds almost immediately can become part of everyday work. In that sense, the value of AI is shifting away from leaderboard results and towards what happens at the point of use, where latency, reliability and ease of access shape whether a model becomes habitual or remains a specialist tool.
The debate comes as Google is also undergoing a major reshuffle in its AI leadership. Axios reported that Demis Hassabis is stepping away from day-to-day control of Google DeepMind to become chairman and Alphabet’s chief scientist, while Koray Kavukcuoglu moves into a broader leadership role. Jeff Dean, one of Google’s most prominent AI figures, is also leaving to launch a new startup with other senior researchers. IT Pro said the departures have fed outside concern about Google’s position in the AI race, even as the company retains important advantages in chips, cloud infrastructure and large-scale consumer products.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
