Sarah Hunter | full interview
If you read our first newsletter, read on for the full interview with Sarah Hunter
MILLTOWN PARTNERS: How do you define “deep tech”?
SARAH HUNTER: MIT’s Engine Fund calls it “tough tech” and I think there’s a lot of truth in that! I hear two definitions. The first is technology that is rooted in basic science and at an early stage of development.
The second - and in my opinion more rigorous - definition is that it involves the application of technology to the physical world, as well as software. Some instances of machine learning would not fall within this definition, but the supercharging of hardware with AI - from climate tech to biotech - is a perfect example of deep tech.
From the Inflation Reduction Act and Chips Act in the US, to the UK’s Science and Technology Framework, governments are starting to use policy to support technology markets (dare we say it: “industrial policy”!). What in your view are the biggest factors driving that?
The US has been the most explicit about its rationale: the Inflation Reduction Act is a response to decades of hollowing out of American manufacturing industries. This line of thinking is being supercharged by geopolitical pressures - most pressingly the need to compete with China.
But it was the radical actions of rule breakers that opened the door to this type of interventionist policy. Trump instituted strong anti-China policies, which enabled Biden to in turn carry on down that path
Arguably, Brexit provides the same opportunity for a future government in the UK.
The rulebook has been thrown out and we're left with a blank sheet of paper where industrial policy has to be part of the UK’s answer. But we haven't acted on it in such a radical way - yet.
The UK’s Science and Technology Framework identified five critical technologies that will make the UK a science and technology superpower by 2030 (AI, quantum, biotech, semiconductors and next gen communications infrastructure). Do you see any benefit in grouping deep tech areas in this way? Is there any read-across for policymakers?
Those five technologies have very different use cases, so from a regulation perspective I don’t think it's particularly helpful to group them. But the interesting thing about all of these technologies is that they have huge potential to transform the world for the better, and there is some read-across in terms of the types of policy that can enable their continued development.
Sandboxing is a great example. Ten years ago, in what was arguably an early example of sandboxing, the UK Government created the self-driving car voluntary framework. This enabled self-driving car companies to develop their technology in the UK in the knowledge that the government had blessed their operations. More recently, Sir Patrick Vallance has recommended creation of an AI sandbox - a time-limited relaxation of the rules to enable breakthroughs that can benefit society. There are regulatory solutions that can encourage deep tech innovation if governments are smart enough to use them.
For governments creating policy, it’s also useful to understand how these technologies intersect and how, in the future, a technology developed in one sphere could have transformative potential in another. A good example is how CRISPR, developed for the human body, is now being applied to modifying the genome of plants to sequester more carbon and grow in more climate challenged conditions. This is a deep tech application of genetic medicine to agronomics.
One of the things that characterises deep tech companies is the long time horizons on which their technologies are being developed, whereas politicians typically focus on short term wins. How can policy professionals keep policymakers’ minds focused and minimise the impact of change of governments?
One of the biggest challenges of deep tech is the fact that changing physical systems and infrastructure is hard. It takes time, people and partnerships. Software industries are not good at this, and venture funding cycles don't suit it.
We are entering an era in which deep tech companies are increasingly going to need to seek government help. This is a big shift from the time of software companies telling governments to back off and let them innovate. A different relationship is needed between deep tech companies and governments: one that involves much more collaboration.
So far (apart from China) I think the US has done this best. The Inflation Reduction Act, The Chips Act and The Infrastructure Act all offer government partnership to the technology sector in numerous ways, from subsidies and tax breaks (the IRA) to investing in super early stage Research & Development with the National Labs (the Chips Act). The US Environmental Protection Agency (EPA) could take further action to update planning regulations.
In terms of how to convince governments to listen, in the current economic climate it's all about jobs. The Inflation Reduction act was well received because it created jobs, particularly in Republican and swing states.
Local and regional governments can also be incredibly important to technologists. If you are a deep tech innovator in the climate space building an innovative battery it will be the regional and local politicians who have the power to open up new land for you to build on planning permission and provide local tax breaks. For local politicians, economic growth and opportunity is the number one priority.
Given the tough economic environment in the UK and US, are there more innovative ways that governments can think about supporting deep tech companies that don’t involve huge budgets? How can policy professionals help them to think about alternative solutions?
Number one: speed up the planning process. This is essential - it currently takes too long to build large projects, and often they are stopped at the last minute by landowners or by local residents as a result of misinformation.
Number two: improve collaboration with universities. Making universities much more porous to startups and commercial operators would be powerful for deep tech innovators who rely on integration with other technologies being developed in universities, or need talented engineers. We also need to address our spin-out problem. Universities are often slow and greedy when it comes to IP demands on spin outs, so that should be standardised and simplified.
Number three: use procurement to support innovation. Government agencies spend billions of dollars acquiring technologies and it is an opaque and cumbersome process. They have the power to sort this out and save money in the process. Biotech is one of the most world-changing applications of AI and new forms of data and technology. How amazing would it be if the NHS had the power to procure all of those early stage technologies and give innovators opportunities to access their contracts?
I think we might see some progress in the procurement of climate tech solutions by US states and cities, which is where a lot of IRA money is going to be spent. There’s a great startup called Atlas which provides a database of technologies that have been proven, procured and applied by individual cities. So if you are a city and you want to buy an innovative wastewater solution, you can see what other cities have done. Not just deep tech but high tech!
Many governments are still focused on digital policymaking in the areas of data, safety and competition, as well as increasingly on AI. What are they missing that they need to get ahead of?
For policymakers, Generative AI is clearly the topic that generating the most buzz. From Biden hauling the CEOs of the big AI companies in for meetings to every politician in the world using our ChatGPT to write their speeches - it’s clearly on the radar of policymakers.This is a really good thing compared to where we were 20 years ago with the advent of web 2.0, where it took policymakers a long time to even notice what was happening let alone consider public policy solutions.
I think policymakers have learned and evolved, but they still risk relying too heavily on implementation of existing law and regulation in areas like data and competition. Governments are pushing the responsibility to the regulators with a mantra of “we’ve got the powers to regulate AI and we just have to use them”. But no regulator I've come across has anywhere near enough technologists working in there to understand what to do about AI. The UK’s Competition and Markets Authority has kicked off a review of AI, but this is not just about competition.
I also think copyright is an interesting space in which we’ll see policy development for generative AI. I would advise anyone working in AI to pay close attention to what’s happening with the screenwriters strike in the US. Artists and writers are raising the flag of copyright infringement with generative AI, which is exactly what happened 20 years ago when the web started. The copyright lobby is the most powerful lobby out there and policymakers are much less inclined to support technologists because of monopoly concerns they see.
One of the consequences of the “move fast and break things” approach of the current Big Tech companies has been a government backlash and a focus on regulating to prevent harm. Are there any lessons governments can draw from this experience? Is there a risk that fear of repetition leads to policy that stifles innovation?
I think it's inevitable, if you look again at the arc of internet policy making of the last 20 years. The case for open innovation was made very strongly and very successfully for the first 10 years of the internet’s existence. Then examples of harms began to pile up, from monopoly concerns to teenage mental health problems. We’re still in the phase of dealing with the fallout, so of course policymakers are approaching AI regulation from a more cautious and risk averse standpoint.
But one of the most frustrating things about the generative AI debate is that other AI models will be able to address climate change and discover new forms of chemistry - it won’t be ChatGPT. We have to guard against a rush to regulation that shuts down new models under development because we’re scared of disinformation from ChatGPT.
The UK Government wants the country to be a “science and technology superpower” (note the science comes first). How do you draw the distinction between science and technology, and how do policy considerations differ between the two areas)?
They are very separate. Most science - with some notable exceptions like pharma - happens in universities in the UK. And most technology happens in the startup and the private sector. There are different policies required and different things holding those different sectors back.
I think there's a real opportunity for the UK to become the hub of science entrepreneurship. Our policy aspirations for the UK being a science superpower are all based on the fact that our universities are in the top five in the world and research in all the different scientific fields. That's where this expertise is, this is where our reputation is founded.
However, they are not making it out of the labs and into commercial settings. Nor are scientists staying in the UK to the extent they should be. We need to be doing more to unlock and keep science talent in the private sector.
You held a global role at X - did you see any examples of governments doing tech policy particularly well?
The world is fundamentally changed. We all work, play, socialise and educate in totally different ways, and governments haven’t caught up - they are still employing processes they had 30 years ago. So I think the great opportunity for any government is to radically transform how they do their business - to become Amazon, not Vodafone.
Taiwan is the government that I would cite that has done that. It has very unique geopolitical characteristics and lots of lots of coders.
I think the opportunity is for a Western government to change itself, too. We should all be technologists. We should all be confident about how we use technology, and owners of the tools rather than just recipients.
So if you were running one of those governments, where would you start?
It’s all about talent. It’s no accident that Taiwan is a case study of effective tech policy and Audrey Tang is the digital minister.
You see case studies of transformation projects being farmed out to consultants and failing, because ultimately they are not the client. Governments - the clients - need to own the capabilities themselves. A good example is the Obama Presidential Innovation Fellows programme. They brought technology innovators into federal government departments for 12 month periods to solve a specific problem using technological tools.
If you can get the people at the top to hire the right people then the policy they make, and the way they interact with the public, will be transformed.
What did you learn during your time at Google X that might be useful to other tech policy professionals?
Engineers have much more credibility than policy people - make sure you bring them into meetings.
Focus all your conversations on what it is that the government wants, not what you want. Reframe your pitch alongside their goals and not yours.
Do more listening than talking - don’t go into a room and just talk at someone, go into those meetings to ask questions. Sundar [Pichai] is a master at this.
Timing is everything - don't talk to the government unless you have a tangible and concrete idea of what you need from them, But also don't leave too late so your competitors have already got in the door.
You recently became a board member at the Advanced Research and Invention Agency - the UK’s version of the US’s ARPA agencies - which will no doubt be investing in deep tech. What can you tell us about your initial priorities?
Our first priority is to hire programme directors. We are looking for a small number of people with a science or technology background, from Nobel Prize winning scientists to start-up leaders. These individuals will define which missions we’ll focus on and where we put the R&D money. Once those program directors have identified their missions, tech companies and corporates should come and talk to us, because I think there's a real opportunity to experiment