Google is on the brink of losing two high‑profile AI scientists, as Gemini contributors Jonas Adler and Alexander Pritzel are set to move to competitor Anthropic.
Bloomberg reports that both figures are regarded as key players in Google’s flagship Gemini lineup. Adler has spearheaded the company’s AI coding initiatives, whereas Pritzel has been integral to the training pipelines that underpin new models. Neither side has yet issued a public statement, nor have they disclosed the exact dates of departure or the roles they will assume at Anthropic.
Building a large‑scale AI model demands critical choices around architecture, data curation, distributed computing, and post‑training fine‑tuning—all of which shape the system’s ultimate behavior. Talent with hands‑on experience at Gemini’s scale is therefore hard to replace on a short notice.
Both Adler and Pritzel also have a history of contributions to DeepMind’s research, notably co‑authoring the team’s work that extended AlphaFold’s protein‑structure predictions across entire proteomes, alongside luminaries such as John Jumper.
Google’s Gemini division is already feeling the tremors of a major personnel shift: Noam Shazeer, one of the co‑authors of the seminal Transformer paper that underpins today’s most powerful language models, has departed the company to join OpenAI. Shazeer had recently returned to Google in 2024 through a partnership with Character.AI, but the move to OpenAI marks a significant realignment of expertise at the forefront of AI research.
In a parallel development, Anthropic is bolstering its research ranks by attracting high‑profile figures from rival labs. Andrej Karpathy, the former Tesla AI director and co‑founder of OpenAI, has joined Anthropic’s pre‑training team, and the company has reportedly recruited several Google researchers. This influx of seasoned talent signals Anthropic’s intent to deepen its focus on the core capabilities that will drive future Claude models, rather than concentrating solely on product and enterprise sales.
The competitive landscape is further complicated by the intertwined commercial relationships between these giants. Anthropic directly competes with Google’s Gemini models, yet it also relies on Google as a critical infrastructure partner. In April, Anthropic announced an expanded collaboration with Google and Broadcom that spans multiple gigawatts of next‑generation Tensor Processing Unit (TPU) capacity—Google‑designed accelerators essential for training and running large AI models.
❓ Frequently Asked Questions (FAQ)
Who are Jonas Adler and Alexander Pritzel, and why are they significant to Google’s Gemini project?
Jonas Adler and Alexander Pritzel are senior AI researchers who have played key roles in Google’s Gemini initiative. Adler led the company’s AI coding efforts, while Pritzel was integral to the training pipelines that underpin new Gemini models. Both have also contributed to DeepMind research, including work on AlphaFold.
What impact could their departure to Anthropic have on Google’s AI development and the broader AI industry?
Their move could accelerate Anthropic’s model development by bringing in expertise on large‑scale architecture, data curation, distributed computing, and fine‑tuning. For Google, it creates a talent gap in Gemini’s pipeline, potentially slowing progress and prompting the company to seek replacements or restructure its research teams.
Are there any known details about their new roles or timelines at Anthropic?
No public statements have been issued by either party, and the exact dates of departure or the specific roles they will assume at Anthropic remain undisclosed.
News Source: Neowin
Comments
Be the first to comment.