Skip to main content

EU on the right path with AI policy – But the US & China remain ahead

(Image credit: Image Credit: John Williams RUS / Shutterstock)

The artificial intelligence (AI) policy white paper from the European Commission was a late but necessary step to differentiate the AI policy of the European Union (EU) from those of the US and China, both of whom they lag behind in terms of almost all metrics of AI readiness. The paper recognises that the EU needs to focus on the battle for industrial data and scaling AI across established businesses, where its relative shortfall in areas such as AI talent and the commercialisation of research are potentially not as important as in the consumer sector. However, its plan for an ecosystem of excellence currently does not give enough detail on how they will address the legitimate problems that businesses are having with AI adoption.

In particular, the focus on trust and regulatory frameworks is only discussed in terms of reducing the negative consequences of AI (and has thus been framed by many opinion pieces as a potential innovation killer). Yet, such transparency would help overcome one of the key barriers that many existing businesses have with new technologies now – communication.

The ecosystem of excellence gives the EU the chance to address the fact that the adoption rate of AI by European businesses is lower than that of China and the US. This has created a level of scepticism around AI within many businesses, where it is often viewed as unproven and overhyped. In general, it has been harder than expected for established companies to scale innovative solutions across their business, meaning that successes have tended to be on smaller, tactical problems with less demonstrable value than strategic ones. This is in contrast to the business-to-consumer (B2C) sector, where pockets of positive AI experiences are much more noticeable and embedded into our daily lives.

Accessible to non-technicals

For companies to succeed with AI, they cannot simply overlay it onto their current operating models. They will require strategic changes to their business processes. The tools they need are likely already out there – there’s no need to wait for algorithms or technological advances anymore. It’s about hiring the right people, investing in the right training for them and crucially finding ways to tie the new technologies together. It’s not the AI itself which causes bottlenecks, but the capacity to successfully integrate it into a business.

Rather than desperately trying to catch up to the US and China in terms of AI skills and innovation, the EU needs more focus on the right sorts of skills, tools and communication mechanisms for its ecosystem strategy. If the EU simply tries to fast-follow the US and China on AI policy, it’s going to be left in the dust.

The white paper outlines how the EU aims to increase the general level of technical competency through its forthcoming Skills Agenda. However, experience has already shown many businesses that taking a host of technical data scientists and expecting them to solve industry problems without domain knowledge is a recipe for failure. The most acute shortage is in crossover skills that can combine innovative technical approaches with domain knowledge to solve industry problems. It’s not much different than universities designing science, technology engineering and mathematics (STEM) curriculums without sufficient humanities incorporated to help provide a comprehensive education. Developing these skills needs to be a key part of the agenda for industrial collaboration if businesses are to move away from their current siloed approach to AI implementation.

In addition to upskilling and reskilling the workforce, there needs to be a focus on making AI more accessible to non-technical users. For many problems in industry, algorithmic innovation is often miles ahead of the status quo, and applying the latest deep learning approach is like using a sledgehammer to crack a nut. Instead, it’s better to build around approaches that are already well established and commoditised so that AI is more approachable for less technical users.

Important first step

However, this is not to say the EU isn’t leading in certain aspects when it comes to AI policy and regulatory approach. The EU recognises the importance of data pooling and collaboration through their parallel European strategy for data, and it’s a framework we’re seeing the US and China increasingly discussing at the state and federal levels of government. This is an extremely important consideration. For many enterprises, the ability to develop new products and services will depend on how well they can utilise their proprietary data. Tooling that does not rely on a heavy investment in technical specialists or on third-party models that do not properly represent their business will give them the best possible opportunity.

Finally, we come back to communication and trust, and how AI can win the hearts and minds of business stakeholders and overcome their scepticism. Although it rightly focusses on “high risk” use cases, the white paper surprisingly does not discuss how increased transparency can educate and convince workers of the benefits of AI. A recent survey highlighted that industry leaders see AI playing a key role in allowing staff to do their job more effectively, rather than replacing them through full automation, so these staff need to have full confidence in how any AI is operating.

This encompasses everything from transparency about what data is being used by the algorithms, through to the delivery of output in a form that is intuitive, understandable and can augment human decision making.  In addition, transparent output will allow staff to be able to communicate more effectively with other stakeholders, demonstrating business value without needing to get bogged down in technical detail. The acknowledgement in the paper that there is a need to build bridges between academic disciplines in order to enhance AI explainability is an important step in this regard.

Overall, the white paper is an important first step to increasing the business adoption of AI in the EU. And the focus on transparency has wider potential benefits than officials seem to anticipate. However, the EU still needs to understand more about what businesses truly need to make AI work for them if it wants the ecosystem of excellence to be successful. If it can’t develop its understanding further, it’s doomed to lag the US and China in terms of AI adoption and innovation for the foreseeable future.

Dr Bob De Caux,  vice president of AI and Intelligent Automation, IFS