New ways to make scientific advancements?
Thinking through open science, non-profit science and interdisciplinary science
The technological progress we’ve seen over the last few years has been immense. It’s unlike anything we have ever seen before and that has got a lot to do with technologies building on top of other technologies and different major waves of change blending together (e.g., biological sequencing and AI). And that progress is coming from everywhere. It’s not just coming from companies or academia, but from both and everything in between. It got me thinking about how this landscape (of changemaking) is evolving and what (if anything) it can become. This is what this blog post is about. And I thought, maybe we can learn something by looking back at the early days of technological progress.
Looking back: Bell Labs
It was 1924 when Western Electric and AT&T had the idea to bring together all of their R&D activities in the communications field under one roof. Named after Alexander Graham Bell and his important work on the invention of the telephone, Bell Telephone Laboratories was born and around 3600 scientists, engineers and support staff set to work in their building in New York. Bell Labs was a first important lab that emphasized the importance of science in discovery and innovation. Over the years, many revolutionary inventions were credited to the teams at Bell Labs, including the transistor, the laser, information theory, Unix, programming languages C and C++ and many others. It was a fabulous experiment to find out what would happen when you put brilliant scientists together and give them enough resources to work on whatever they find most important.
By the 1950s, William Shockley grew tired of Bell Labs’ management and moved to California, eventually to start his own company Shockley Semiconductor Laboratory. While the company itself was not a very big success, it planted the seed of what would grow into the iconic Silicon Valley, arguably the biggest center of innovation in the world. Given this backstory and what drastically positive impact a structure like Bell Labs has had on the world, I believe we need more present-day Bell Labs. And there are a few examples in today’s world!
The industrial R&D labs
Indeed, there are a few present-day Bell Labs across the world. Probably the most known ones are the big industrial AI labs at Google, Meta and Microsoft. Google DeepMind, OpenAI and Meta AI are three (as far as I can tell) quite independently run organizations that definitely qualify as ‘research labs‘. Two of them even have the lofty goal of creating artificial general intelligence (AGI), which definitely requires a lot of research. OpenAI is probably the most unique of the three in terms of its history and current structure. The lab started out as a non-profit with the goal of creating safe AGI to benefit the world, and has since transitioned to a capped for-profit because it struggled to raise enough money to adequately expand the scale of their efforts. It’s not like these labs have unlimited budgets, but compared to typical academic labs you would consider them very well funded. And in the AI space, that’s a big advantage to make progress. I would say the same applies to biological R&D: it’s expensive and time-consuming to work with biological entities. And while more funding does not solve biological complexity, it can give you access to better infrastructure, more people etc; which allows a well-funded lab to make progress faster.
So does large scale funding clash with an academic or non-profit focus? I wouldn’t necessarily say so…
Open science and non-profit orgs
Indeed, the industrial AI labs are not the only kids on the block. One other great example is the Chan Zuckerberg Biohub Network. Founded as part of the Chan Zuckerberg Initiative, the Biohub network is a group of nonprofit research institutes that pursue grand scientific challenges over 10-15 year time periods. Two more examples that were only very recently announced are FutureHouse in San Francisco and kyutai in Paris. FutureHouse is a moonshot project aiming to build semi-autonomous AI systems to accelerate scientific research and discovery over a 10 year timeframe. Kyutai is Europe’s first independent research lab dedicated to open AI research. Both FutureHouse and kyutai are organizations that fit within a new concept called Focused Research Organizations, and both are (co-)funded by Schmidt Futures.
What these non-profit orgs additionally have in common is their dedication to open science, an important trend we’ve seen emerging in the last few years. I believe fundamentally what we’re seeing is that scientists and researchers are starting to realize that doing open science makes a lot of sense in terms of (1) building on top of other labs’ advancements, (2) avoiding mistakes that other teams have already figured out and (3) just communicating openly to the scientific community, which could lead to increased understanding and more collaborative efforts. A tweet by Hugging Face CEO Clément Delangue has a similar take on it:
Secrecy is so detrimental to the AI field and to the world right now. It fuels misinformation, sends tons of researchers and ressources to the wrong directions and leads to dangerous mistakes like bad regulation because of public opinion fears. I wish leaders of the field would take this issue more seriously and foster more transparency and openness!
Conversely, I think the biggest fear people have with open science is that bad science (that might seem like good science) can also be shared and thus could still lead to misinformation and wasting of resources. This is definitely a justified concern to have, which is why we still need good peer review (ideally also FAST peer review once a manuscript has been shared). But overall I think open science is a net positive, and that this movement will continue to grow.
My overarching point here is that we clearly see money flowing into research and development beyond academic research, at a larger scale than academic research yet typically without a direct or short-term goal of commercialization or profit (but rather having an open science goal). And I want to make the case that we need even more of those.
Back to the triple helix for science & innovation
We need more of these initiatives because (1) the most cutting-edge science is becoming increasingly interdisciplinary, requiring larger teams, better infrastructure and longer time horizons, and (2) science and technology are increasingly important to society, which requires open communication and a tight feedback loop with regulatory entities. Teams on a non-profit, non-academic and open science mission are ideally suited to take on these larger, bolder scientific missions without neither of the downsides of academia nor for-profit companies. But on the other hand, we do still need both these entity structures!
In my opinion, what we ideally should strive for is the triple helix model of innovation (first theorized by Etzkowitz and Leydesdorff in the 1990s), focusing on interacting between universities, industries and governments.
Academia is where research typically begins at the most fundamental level up to strategic basic research. If academics work closely with industry, non-profits and regulators; research that is ripe for further development and commercialization can make its way faster towards reality. But new research ideas can flow in the other direction as well: from industry or government to academia. Both these entities can also provide the necessary funding to academia (or non-profits for that matter) to enable new research ideas to come to reality. Industries and governments also benefit from collaborating and communicating openly with one another, as companies should adhere to existing regulatory frameworks. But at the same time, they can work with governments to help reshape and improve outdated regulations. Non-profit institutions come in as a hybrid between academia and industry, playing an important role in open science and upscaling research, which, at a later stage, can still be commercialized if the interest is there.
The key is open communication across all these entities. There’s definitely some of that today, but we can always improve. And as technological progress increases in speed, we have to improve! This requires trust and good intent. It requires us to overcome some of the fears we have for keeping things secret and being hesitant to collaborate. But I think that’s a worthy tradeoff if we want to make better progress across science and technology that ultimately benefits humanity!