Golden opportunity or massive threat?

You can hardly read the news, listen to the radio or scan your preferred social media without hearing about AI. Or experiencing it in practice, whether you are aware of it or not.

On the one hand, it is seen as offering huge potential to transform business and other organisations, reducing costs and creating entirely new capabilities.

On the downside we hear of threats to democracy with a surge in fake videos and information; the potential for mass job losses as AI systems replace employees to reduce costs; at the extreme, dire tales of AI systems taking over humanity altogether.

One important concern is the concentration of AI development in too few powerful hands and the struggle of governments and international bodies to regulate them.

In this podcast we try to provide a balanced view and to give a sense of what you need to know, and perhaps what you do and do not need to be concerned about. That includes what to be looking for both from governments and business, and how to balance the need for regulation with desirable innovation.

Our three guests are very well qualified to provide the full range of perspectives, from the technical and innovation aspects of AI through to its impact and the challenges of regulation.

Leonie Mueck is currently VP of Products at nPlan, a London-based scale-up, which uses AI to forecast construction project outcomes making construction risk management easier, faster, and more reliable. She is also a speaker and mentor on deep tech product management, a member of the technical advisory group of the UK’s National Quantum Computing Centre and an assessor for Innovate UK projects. Leonie sits on the party’s policy working group on science, innovation and technology.

Leo Ringer founded and runs Form Ventures, a VC fund investing in early stage technology companies with a focus on regulated markets like fintech and healthcare. He spent the first decade of his career in business and economic policy, including as an adviser to the UK Business Secretary. He’s a firm believer that policymakers, entrepreneurs and investors need to spend more time talking to and understanding one another.

And chairing the session we have Lord (Tim) Clement-Jones CBE – Tim is the Liberal Democrats House of Lords spokesperson for Science, Innovation and Technology, and former chair of the Lords Select Committee on AI. He is also the Vice-Chair of the APPG for IP and APPG for Digital Regulation and Responsibility. Outside of Westminster, he is a founding member of the OECD’s Parliamentary group on AI and is a Consultant on AI Policy and Regulation to the Global Law Firm DLA Piper, where he was London managing partner.

Synopsis and analysis

The podcast kicked off by discussing what is meant by AI and how it has developed, with examples from different sectors, before moving on to more recent developments that we have been hearing about – Large Language Models such as ChatGPT. Leonie explained how these take us into a different world, contrasting the ‘deep learning’ on specific tasks that we have had before, with the ability of today’s LLMs to do things that they were not trained on.

Leo brought a different perspective, seeing it from the point of view of an investor. For him, AI really is at the frontier, not just on its own but more often as part of a wider business proposition. Yes, there is a lot of hype out there, with people jumping on the AI bandwagon, even if it is not necessary. Plenty of innovation is possible without AI, with the best value perhaps being where there is a mix of AI and human judgement.

Another big shift, with tools such as ChatGPT, is that they have become available to everyone, without any technical knowledge. It also means that new products can be developed much more quickly, creating huge opportunities. And it is there already, sitting on our smart phones and embedded in the apps that we use, deciding what we see on social media whether we realise it or not. Increasingly it is possible to train models on specific tasks, much more quickly and cheaply, in contrast with ChatGPT, which was trained on the whole Internet, with a trillion parameters.

Leonie’s business is focused on how big construction projects, notorious for massive over-runs in costs and time, can be managed better – a hot topic in the UK with many examples such as HS2 or the Hinkley Point power station. For businesses like this, it helps that even small organisations and start-ups have access to massive computing power through the Cloud, that previously was only available to large organisations. This fits with what Leo is seeing in the investor world, with the start-ups coming to him for investment. Most of them are building on the infrastructure and tools that have been developed by large firms like Microsoft and Google. However, that does beg the question of ever-increasing power in the hands of those big firms.

Next, on opportunities for the UK, the panel was optimistic given the expertise and capital available, increasingly not just in London but in other major cities. However, it is very much a global race with much of the underlying technology being built in the USA. Not to be underestimated is the potential for less glamorous use of AI, transforming the workflows in existing businesses with big potential for improving productivity.

The UK is still more risk averse than the USA, and there is the problem of successful start-ups being taken over as they scale up, as happened to Deep Mind which started in the UK but was then bought by Google. The big money does tend to be global. Leonie talked about the potentially positive role of the government, using the example of quantum computing, where there was an early strategy and supporting ecosystem. Along with the government’s role as an important customer, it provides a possible lesson for the development of AI.

Leonie also highlighted the challenge of change when introducing AI, as cutting-edge technology collides with the realities of organisational inertia. Technologists need to make what they have to offer much more understandable and less scary.

In talking about the risks, there was no doubt on the panel that jobs, and especially white-collar jobs will change. That makes the Liberal Democrats’ lifelong learning policies even more relevant. There are also risks to democracy where previous generations of AI tools have undoubtedly played a role in recent elections, with AI being used to determine what we get to see when we use social media. The UK government’s responses have so far been disappointing, though the panel’s biggest concerns were perhaps about the impact on young children as they grow up and get their smart phones.

Conclusions

Regulators and policy makers have a very important role to play, and this needs high levels of skills both in the technologies and the sectors being affected, cutting across many departments. AI is best regulated in the sectors where it is being applied, rather than just one regulator across all sectors, social media being a sector in itself. More informed and capable regulators are as important as more regulations. Furthermore, as this is cross border, it will need supra-national regulation – the UK cannot go it alone.

Interested in reading more about AI? A few suggestions include:
Cathy O’Neill – Weapons of Maths Destruction
Hannah Fry – Hello World
Richard and Daniel Susskind – The Future of Professions
Anthony Seldon – The Fourth Education Revolution
Yuval Harari – Artificial Intelligence and the Human Mind
Melanie Mitchell – Artificial Intelligence: a Guide for Thinking Humans