AI revolution is revealing gaps in trust and regulation

Is AI (artificial intelligence) really worth the $3 trillion that was added to the valuation of the big global tech companies? Investors were betting that the promised improvements to work and life will be very profitable for the suppliers.

But some economists are more sceptical of the likely productivity boost, with many businesses struggling to see exactly how they should harness the new technology. And there is a risk that the social impact of the technology will be too important to leave to businesses and the tech giants.

Already there are fears about biases and misuse of data. Some AI models have been found to ‘hallucinate,’ creating content that is untruthful or does not match the facts provided. There are gaps in trust and regulation; the field has developed faster than effective control. Last November the UK hosted an AI safety summit but it is a tough ask to translate promises by almost 30 governments and global technology companies into action. A bi-polar world is not the best setting to create international cohesion and new institutions.

READ MORE: Are critics too quick to judge Scotland's investment bank?

Some of the challenges are hard to quantify but nevertheless look serious. AI may initially compete with the cheapest available workers, often trainees and interns. Investment firms, for example, have found that attending company results meetings and summarising what is interesting can be done by AI remotely, condensing the meeting transcripts.

It looks like a productivity boost but means less work for investment analysts who traditionally would have started their careers in that role. Certainly, it might allow everyone to move up to work of higher added value - more judgement, less data collection – but it is easy to see the potential for widening inequality.

Entry-level jobs could be squeezed and we should not assume that AI will solve the problem of decline in the working age population. It means that economic policy must emphasise re-skiling and continuous education.

AI has the potential to create immersive learning and help citizens navigate complex areas requiring personalised advice, like healthcare. It can also evaluate medical scans, with one US study finding superior performance in diagnosis when compared with clinicians.

READ MORE: Popular crisps win first supermarket deal outside Scotland

Financial services will also harness AI to deliver lower cost products and better match those to customer needs and risk tolerance. In the last twelve months many users of public and private services will have found chatbots are the only entry into accessing help, and the boundary between AI and human engagement is increasingly blurred.

But public policy is needed to ensure this does not create a barrier of digital exclusion for those less able to use the technology. Some service organisations have been quick to withdraw phone contact or use no-reply emails. New interfaces with customers may make service provision manageable in a world of working from home but standards of practice should be set to ensure equality in access. In the early years of AI adoption this looks like the issue that poses the biggest social risk.

Innovative technology often means rational thought goes out the window. There is a fascination with shiny new things that can lead to wasteful business investment and stock market bubbles.

Already AI has been embedded in products ranging from toothbrushes to prams, without apparent customer benefit. In the US there are 1,400 start-up firms claiming AI innovation, but most are likely to fail as actual real-world applications become clearer. We may be seeing a parallel with the dot-com bubble. In a more recent example of overblown expectations, Bitcoin has given stockmarket investors big gains, but the promised uses for blockchain technology have been slow to materialise.

The best uses for AI may take time to develop, but already it can gather data effectively and personalise service experience. This can improve work and lives, even if it does not show up in economic growth and productivity statistics. But there is a role for standards and public policy to ensure AI reflects society’s values and the benefits are fairly distributed.

Colin McLean is a director of SVM Asset Management Holdings