Dear Editor,

Our recent past has been defined by ever-accelerating technological progress.

Digital technologies have advanced rapidly, diffused across societies, and ushered in a new era by providing the modern infrastructure required to collect data at scale. In the coming decades, artificial intelligence (AI) will have consequences that compare to the Industrial Revolution. Twenty-first century leaders face a strategic choice that will define their countries’ future.

At present this is a choice for which the potential price of change may be high. But the reality is that in time this will be the only way to govern effectively. The nature of the challenges we face means we don’t have an alternative. Our institutions need updating, and our public services need improving. In an age in which people can access nearly unlimited information at their fingertips, a failure to act will widen the disconnect between what they enjoy in some parts of their lives and what they experience with Government. The risk? Lower trust that the Government can deliver.

Added to this are problems we face as a species that does not lend itself to simple solutions. Progress is not being made as quickly as we would like as the number of “disruptive” scientific discoveries, that is, those that take a field into an entirely new direction, begins to plateau. Climate change, medical breakthroughs, energy security, and other areas require us to think beyond business as usual.

In response to multifaceted challenges, we need to use technology, incentives, and systems that can contend with complexity.

Google DeepMind’s AlphaFold system is an example of how AI represents an opportunity to make progress on incredibly complex problems. Before AlphaFold it took a PhD and around US$100,000 to determine the shape of a single protein, the fundamental building block of life that can unlock new drug development. Using machine learning targeted at a specific challenge, AlphaFold was able to discover the shapes of all 200 million known proteins — equating to around one billion years of research. We don’t have a billion years to solve these problems, so doing science at digital speed is a necessity.

The recent AlphaMissense model builds on AlphaFold to predict the harm potential of 71 million “missense” variants: genetic mutations that can sometimes cause diseases like cystic fibrosis or even cancer. Both databases have been made freely available so researchers can use these discoveries across multiple disciplines.

These are just two examples that could unlock once-in-a-generation discoveries in biology. But if the UK is to stand the greatest possible chance of making the potential of these discoveries a reality, there are actions the Government should take right now. With its AI Safety Summit approaching, there is no better time for the UK to move to secure its future.

*Invest more in digital infrastructure, including the compute and health sectors. The benefits of investment could be significant

*Treat data as a competitive asset. Many of the giants of the modern economy have built their business on data, and countries need to capitalise on what they can create.

*Continue to improve private-public collaboration.

In all of this, there can be no trade-off between AI safety and progress. We need both.

This is an important conversation to get right, not just for the UK, but for countries everywhere. Getting it wrong risks missing out on the upsides of this new wave of innovation. For the UK, this technology presents a major opportunity for a country that is already towards the front of the pack. By grasping the chance now, the UK can seize the opportunity to reimagine the State, stimulate scientific discovery, and solve some of the challenges it faces. This window won’t last long, so we need to move now.

Tony Blair Institute for Global Change