Dear Boris

Dear Prime Minister Johnson 

Congratulations on your new appointment. Now that you’re PM, I’m sure lots of people are going to be badgering you to develop new policies and address societal issues, and here I am to join them. Leaving the EU on 31 October is your immediate priority, but there are many large challenges facing this country; not just Brexit.

The way people work and the jobs that they have is about to change as artificial intelligence (AI) becomes more sophisticated and pervasive. The OECD estimated that over 40% of people will see their jobs change. 

% jobs changed by AI across the OECD.png

While we’re still not at the point where computers can compete with people in terms of breadth of expertise, we are at the point where they can do some tasks faster, cheaper or more accurately. This ability to harness computers as ‘the ultimate intern’ - one that can be trained to perform specific tasks tirelessly, to an ever improving quality - is transforming professions and altering traditional career paths.  In a public service context you might say we are about level 1.5 in some fields.


This is not a tale of doom. Like every other technological change it will bring big economic benefits (PWC estimate it might add 26% to GDP by 2030); some people will win and some will lose. The job of government is to steward this change through and make sure that the UK realises the undoubted economic benefits while minimising the social ills. We should not replicate the response to automation in the 1980s where a more laissez-faire approach has left pools of unemployment as old jobs disappeared and people didn’t adapt into new roles.

To hone in one sector, one of the most exciting fields at the moment is in the clinical setting. AI is getting very good at two things which have traditionally relied on medical training: 

  • Working out what the problem might be and proposing a course of action - Google Deepmind are just one company creating tools that support doctors by mining the patient’s full history and comparing them to lots of others who had similar symptoms;

  • Anomaly detection and image recognition - Many diagnostic tests take an image or sample of the body and rely on specialists like histopathologists or radiographers to identify problems e.g. cancerous cells. Companies like are now reaching near clinical levels of accuracy for cancer detection.

But the key thing with medicine is that you are dealing with people. When you speak with clinicians they often flag how little time they are able to spend with each patient (particularly in primary care). This is an area where people still have AI licked. We can reinvest the time freed up above to provide deeper, holistic, care for each patient. While clinicians are excited by the potential for AI to be used as a diagnostic tool, at least 35% of patients would refuse to integrate at least one existing or soon-to-be available intervention. 

This highlights the challenge we face as a country - to develop and adopt useful AI tools; tools that get rolled out to improve services; and ones that respect people’s boundaries. 

The UK is well placed to be the country of choice for the development of ethical AI. With skill and ambition the UK as a whole can be a one of the big winners from this change. We have leading Universities like Oxford and Southampton attracting the best domestic and international minds, vibrant startup scenes in Manchester, Leeds, Bristol and of course London. We have a strong industrial base to work with.For example, when we looked at countries around the world the UK ranked #2 in the world (after Singapore) at how easy it would be to deploy AI in the public sector.

The UK AI sector deal is a great start and you are blessed with a very talented team in the Office for AI. You have new institutions like the Centre for Data Ethics and Innovation. However, placed alongside the investment of Korea or France, or the ambition of China or USA it looks modest. The pace and scale of implementation should be picked up. 

commitment to AI in dollar.png
Commitment to AI as  %.png

Prime Minister, to reap the full benefits of these changes, here are some suggestions:

  • Take the global stage - at the moment the running is being made by the USA and China in the private sector while France and Canada are leading international efforts to create an ethical framework for AI. As an independent global power with ambition, Britain should be leading the debate, and also supporting the adoption in aspiring economies.

  • It’s about more than STEM - As the medical example illustrates, the adoption of AI doesn’t just create a demand for more technologists. It places an emphasis on human skills: creativity; empathy; and connection - what we call, human specialties. The curriculum and adult education systems should be reviewed to make sure that they are “AI Ready” and preparing people appropriately.

  • Create a solid analytical base - AI is having a rapid and poorly understood impact on society. You should establish rapid and responsive baseline research to e.g. quantify the impact of AI on the labour market in real time so decisions can be taken on the basis of facts not assumptions, to track adoption in different sectors so support can be tailored, the impact of automated decisions on regulated markets so intervention is timely, etc.  

  • Create legal certainty - The Centre for Data Ethics and Innovation is doing a good job of investigating some of the legal and societal issues that occur from techniques like micro-targeting or in the sources of biased decision making. What we need to do is create a legal framework which gives innovators clarity about what is ok and what is not. A framework which can adapt quickly as technology evolves. 

  • Invest in implementing AI across the economy - AI is a transformative technology which will touch every sector. While there are undoubted benefits to countries who are producers of the technology, the UK should not ignore its opportunity to be the best at implementing the technology. This means giving people the skills to be good clients, to understand how and when to roll out AI. To encourage this investment you may want to clarify the rules around R&D tax credit and capital allowances to show that AI investment is in scope. 

  • Focus on the National Data Strategy - Data is the lifeblood that runs through AI. With rich, timely data then the systems can make accurate recommendations, without it then bias and inaccuracy creep in. The National Data Strategy is an opportunity to put in place the right data infrastructure to support economic growth. As part of that consider: 

    • Investing in training data - many of the training data sets are created in the USA or China. The UK should lead the way in creating truly representative data sets to support domestic industry in addressing a global market.

    • Supporting collaboration - Lyft have opened up a massive data set of data about roads, signs, etc to accelerate the development of self-driving cars. This style of pre-competitive sharing should be encouraged and the development of data trusts supported.

    • Reference data - Often the best insights come from linking different data sets together through shared identifiers. The previous administration made steps to make this simpler through opening up ‘core reference data’. This work should be completed.

    • Safety - Creating complex systems built on data is hard and it can be easy to e.g. remove anonymisation or overlook causes of bias. Continue to support efforts by the CDEI and ICO to create a safe environment for data innovation.

  • Ease the links between academic research and industry - There is plenty of support through Innovate UK for joint private sector/academic research. However, it is still rare in the UK to find a professor who is also CTO of a startup or a head of R&D who also lectures part-time. These joint appointments help remove the sense that academic and commercial research are inherently different disciplines.

We would love to talk these over in more detail so please do get in touch.

Yours sincerely

The team at Oxford Insights

Richard Stirling