What Discovery Loop is setting out to do
Discovery Loop is the new company Jeff Dean announced from his own account. It is a Public Benefit Corporation — a US company form that operates for profit while carrying a public benefit purpose in its charter.
The stated mission is to automate machine learning, science and engineering, accelerating discoveries and progress. That sounds abstract, but the method is put more concretely: automate the experimental loop, an approach the founders expect to apply broadly across many fields of science and engineering.
The order of work is spelled out too. Start by automating machine learning research and engineering; use the resulting capability first to optimise their own technology stack, then expand into other domains — making themselves the first customer. This is not using AI as a tool so much as handing the iteration of research itself to machines.
The immediate order of business, Dean says, is finding office space and hiring a founding team over the next few weeks.
Announcing Discovery Loop! ／ … we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. ／ Our general approach is to automate the experimental loop. ／ We think this approach is broadly applicable across many different fields of science and engineering. ／ Our immediate order of business is to find office space, and hire an amazing founding team over the next few weeks. — From the passages on the company name and form, the mission, the methodological approach, the expected breadth of application, and the immediate practical steps
At Discovery Loop, we are building systems to automate these entire experimental loops. ／ By utilizing frontier AI models and large-scale computational infrastructure, our systems will be able to rapidly propose, run, and learn from evaluations. ／ This approach allows for the parallel execution of thousands of experiments, drastically compressing iteration time and driving up the quantity and quality of scientific and engineering output. ／ We will initially focus on automating the process of machine learning research and engineering. ／ We will use these automated ML capabilities to rapidly optimize our own technology stack before expanding to other domains. ／ Ultimately, we are building systems capable of taking on National Academy of Engineering (NAE) Grand Challenges—such as engineering better medicines, advancing health informatics, making solar energy economical, providing access to clean water, securing cyberspace, and engineering the tools of scientific discovery. — From the passages on what is being automated and with what resources, the effect of parallel execution, the first domain of focus, the plan to apply it internally first, and the eventual target problems
The four founders
This is the heaviest part of the announcement. The founders are Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le, who Dean says have worked together for 14 to 30 years.
One thing to watch: Google's internal message names only two of them, Dean and Ghemawat. The figure of four comes from Dean's own announcement. Since the sources differ on the count, keep them straight when citing.
Pichai's message covers Dean's 27-year tenure and the launch of an independent Public Benefit Corporation with Ghemawat. Google frames it as a moment when Dean "wants to try something new" and does not use the word "resign". It adds that the two of them drove some of the company's most significant technology transitions, from the early search infrastructure to the neural networks behind the modern AI era.
I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix ／ The four of us have worked together for 14 to 30 years, and have helped build some of the world's most used products, infrastructure and AI models, and we're excited to turn our attention to this ambitious endeavor. — From the passages naming the co-founders, the years of collaboration, and the areas they have worked in
Lastly, after an incredible 27-year run, Jeff Dean is at a moment where he wants to try something new, and we're excited to support him in that. ／ Jeff and Google Senior Fellow Sanjay Ghemawat are launching an independent public benefit corporation to accelerate discoveries in ML, science, and engineering. ／ Jeff and Sanjay helped to drive some of the most significant technology transitions, from our early search infrastructure to the neural networks that helped create the modern AI era. — From Google's account of the tenure and departure, the founding of the new corporation, and the technology transitions the two drove
The relationship with Google continues
A departure, but not a severed tie. Google states that it will continue as a founding investor and a Cloud partner, and will collaborate on a research framework for ML systems and related infrastructure advances.
In a field this compute-hungry, having a cloud partnership secured from day one is not a small thing. The company that left also becomes a customer of the company it left.
The announcement came inside the same message as the Google DeepMind leadership change — the day Hassabis handed over daily operations and Google's research structure shifted.
We'll continue to work with them as a founding investor and Cloud partner, and collaborate on a research framework for ML systems and related infrastructure advances. ／ On a personal note, it's been a privilege to work alongside Jeff and Sanjay, and I wish them all the best! — From the passages on the form of Google's investment and partnership, and the research area of collaboration
How to read Discovery Loop
The centre of the new company is the automation of research. Not having AI solve the problem, but handing the work of running experiments to machines. It reads as a company betting on that single point.
The website sets out the mission, the method, the first domain, the plan to apply it internally first, and the eventual target problems. What it does not set out is a product or a date. This is not yet a company that can be judged on results. Even so, the fact that people who built both the infrastructure and the foundation models have moved together toward automating research conveys how much weight the field is being given.
