Connecting the dots: Finding the missing links in the UK’s discovery problem
The UK has a long tradition of punching well above its weight in science. After all, it was a London lab which gave the world AlphaFold, and with it, one of the latest in a litany of British Nobel Prizes.
So how can the UK use AI to translate this discovery and research strength into commercial and societal impact? Our universities already produce research that the whole world cites, builds on and, too often, commercialises without us. The missing link is visibility. The UK cannot fully see its own research frontier, the talent behind it, or where patient capital should flow. So the connection between discovery and commercialisation breaks, and this is where AI can help.
So, the question for this week is the right one. Not whether the UK can discover (we know it can) but rather, whether it can turn discovery into impact, at pace, and keep the value here.
Knowledge is growing faster than anyone can read it
Every year brings millions of new papers, preprints, patents, trials and datasets. No researcher, investor or policymaker can hold more than a sliver of it. So, breakthroughs hide in plain sight, because the answer to one field's problem is sitting in another field's journal. For example, a materials finding that matters for energy or a supply chain question already answered in a chemistry paper nobody in logistics will ever read.
To compound this, the most valuable insights are often those right at the frontier or the fringe edges.
The first wave of AI for science was about prediction, enabling us to model climates and digitally fold proteins. This wave has had extraordinary impact and is now being followed by a second wave with just as much potential. This one is about connection: understanding how evidence links across disciplines and geographies, showing people where to look next.
What connection looks like in practice
We see this every day with our partners. In blood cancer research, Corpora.ai is deploying its research technology to enhance the national Acute Myeloid Leukaemia (AML) patient registry, in partnership with the University of Warwick and Accelerated Clinical Trials (ACT). The platform analyses clinical literature, biomarkers and treatment pathways at speed, with the aim of supporting earlier diagnosis models, better trial design and the identification of predictive biomarkers. ACT brings access to more than 27 NHS-embedded research centres, and the programme is funded by MSTART, the Medical and Sporting Technology Research Trust founded by our co-founder Mel Morris, CBE.
MSTART, the Medical and Sporting Technology Research Trust founded by our co-founder Mel Morris, CBE.
This is what a sovereign research technology play looks like in practice. The frontier research connects fragmented evidence, so researchers can see patterns that no single site or study could reveal alone. Warwick's new data science scholarships build the talent to use it, and philanthropic capital makes it possible. The value sits in those connections, and someone has to build the infrastructure to see them.
Three things the UK should do now
1. Treat research connectivity as national infrastructure. We invest in compute and research but far less in the layer that actually links them: interoperable data, shared registries, and tools that map evidence across silos. Without that layer, more compute simply means faster isolation.
2. Build sovereign capability and make it broad. If the systems that decide which evidence matters are built elsewhere, the UK imports someone else's priorities along with them. Sovereignty should also mean breadth. Systems that draw on non-Western research, data and overlooked journals not only the most-cited sources in the most familiar places.
3. Measure translation beyond publication. We count papers and citations with great care and often view frequency here as a measure of success. We should track, with the same care, the time from finding to first application, the number of cross-disciplinary partnerships formed, and the spinouts that scale here rather than abroad.
The human part
AI will not replace the scientist. It changes the job. The researcher of the next decade spends less time searching and more time judging which connections are real, which are noise, and which are worth a grant, a trial or a company investment.
That needs people who can think across boundaries. Personally, some of the best connective thinkers I have worked with are neurodivergent. They are pattern-spotters who see links others miss. Building research and innovation teams that value different kinds of minds will always be a competitive advantage, and in AI for science it may be a decisive one.
From birthplace to home
The UK has the science, institutions and talent. It even has sovereign technology in abundance. What it needs now is the visibility to connect them: a sovereign connective layer that turns isolated brilliance into compound progress, with gains for patients, public services and businesses alike.
Get that right and the UK stops being the place where breakthroughs are born and starts being the place where they grow up
I look forward to continuing this conversation at techUK's AI for Science event on 26 November. It's all connected. The job we have now is to prove it.
