Google's AI research agent ERA grew from Kaggle automation ambitions
Google Fellow John Platt reveals how the company's Empirical Research Assistance system evolved from an effort to automate competitive machine learning into a production AI research agent combining language models with tree search. ERA successfully tackled a long-stalled climate modeling challenge, signaling a shift toward AI systems that augment rather than replace scientific reasoning. The conversation surfaces critical tensions in agent design: how to structure optimization problems so models can meaningfully contribute, the reward hacking risks inherent in automated research, and why domain expertise remains essential for distinguishing genuine discovery from statistical artifacts.
Modelwire context
ExplainerThe Kaggle origin is more than a fun backstory: it explains why ERA is structured around well-posed optimization problems with clear reward signals, which is precisely what makes it tractable and also what limits where it can be applied. Most real scientific questions don't arrive pre-formatted with a leaderboard metric.
This is largely disconnected from recent activity in our archive, so it belongs to a broader conversation happening across the research community about what 'AI for science' actually means in practice. The distinction Platt draws, between augmenting scientific reasoning and automating it, is the same fault line running through debates over AlphaFold successors and autonomous lab systems. ERA's climate modeling result is notable precisely because it came from a domain where the problem structure could be made legible to the agent, not because the agent reasoned freely.
Watch whether Google publishes ERA's climate modeling result in a peer-reviewed venue with full methodology. If the finding survives independent replication on held-out data, it validates the augmentation approach; if it quietly disappears from public discussion, reward hacking is the more likely explanation.
This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.
MentionsGoogle · John Platt · Empirical Research Assistance · Gemini 2.0 · Gemini 2.5 · Latent Space
Modelwire Editorial
This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.
Modelwire summarizes, we don’t republish. Latent Space originally reported this story as “🔬 Google's AI Scientist Started as an Attempt to Automate Kaggle , John Platt, Google Fellow”. The full content lives on youtube.com. If you’re a publisher and want a different summarization policy for your work, see our takedown page.