EU is lagging behind US and China in investments in artificial intelligence, says audit report

brussels times​​​

    The EU has so far had little success in developing Europe’s artificial intelligence ecosystem and failed to accelerate AI investment on a par with global leaders, according to a new audit report published on Wednesday by the European Court of Auditors (ECA).

    AI encompasses emerging technologies in fast-evolving areas including robotics, big data and cloud computing, high-performance computing, photonics, and neuroscience. The US has long been a front runner in AI, while China plans to become the global AI leader by 2030, with both countries relying on substantial private investment through their tech giants.

    Since 2018, the European Commission has taken multiple actions and worked on key building blocks to advance the EU’s AI “ecosystem”, such as regulation, infrastructure, research, and investment. In addition, the EU took early steps to explore AI risks, leading to the world’s first general rules on AI use.

    As regards regulation, the comprehensive AI act was approved by the Council last week. The ground-breaking legislation was hailed as the first of its kind in the world to regulate the use of AI. The legislation was outside the scope of the audit but was considered by the audit team. The cut-off date of the audit was end 2022 and the team thought that the legislation would not be ready in time.

    However, EU measures to boost research and investments in AI were not well coordinated with those of the member states, and monitoring of investment was not systematic, according to the auditors. Going forward, stronger governance and more – and better targeted – public and private investment in the EU will be paramount if the EU is to achieve its AI ambitions.

    The figures speak for themselves. Since 2015, venture capital investments have been lower than in the other AI-leading regions: the US and China. According to estimates, the overall AI investment gap between the US and the EU more than doubled between 2018 and 2020, with the EU trailing by more than €10 billion.

    Against this backdrop, the EU has gradually taken steps to develop a framework for coordinating AI across the bloc by scaling up investment and adapting regulation. In 2018 and 2021, the Commission and the EU’s member states agreed on measures to develop an AI ecosystem of excellence and trust, which would set the EU on the path of a global leader in cutting-edge, ethical, and secure AI.

    “Sizeable and focused AI investment is a game-changer in setting the speed of EU economic growth in the years to come,” said Mihails Kozlovs, the Latvian ECA member who led the audit, at a virtual press conference.

    “In the AI race, there is a risk that the winner takes it all. If the EU wishes to succeed in its ambition, the European Commission and the member states must join forces more effectively, pick up the pace, and unlock the EU’s potential to succeed in this ongoing major technological revolution,” he added. The AI policy is a shared responsibility of the EU and its member states.

    The EU’s AI targets for public and private investment were €20 billion over the 2018-2020 period, and €20 billion each year over the following decade. Actual investments are much less and the targets have not been updated. Nor was it clear how the member states would contribute to overall EU investment targets, meaning that there has been no EU overview.

    The audit was based on sample of R&D projects funded under Horizon 2020, the previous EU framework R&D programme. Research in AI during 2018 – 2020 increased but the programme lacked AI focus and did not trigger sufficient private co-financing. Its successor, Horizon Europe (2021 – 2027) with a budget of €95.5 billion, was delayed during a critical phase.

    Will AI research funded by Horizon Europe catch up? The auditors acknowledged the increase in AI research under Horizon Europe but did not see any breakthrough anytime soon. The recent suspension under pressure of student groups of on-going research projects with associated countries such as Israel is a political question, ECA member Kozlovs commented.

    The share of businesses in the EU using AI differs significantly between member states. France and Germany have announced the largest public AI investments, while four countries are still without any AI strategies at all. The EU has an ambitious target of reaching 75 % of firms using AI by 2030. In 2021, Europe and Central Asia were together responsible for only 4 % of AI patent filings worldwide.

    The concentration of the AI budget is probably inevitable but it should also be possible to identify and support promising companies in smaller countries, the audit team commented.

    More than five years after the first Commission plan, the framework for coordinating and regulating EU investment in AI is still a work in progress. Although the Commission generally managed to scale up EU budget spending on AI research projects, it did not significantly boost private co-financing. In AI, funding is key, Mihails Kozlovs stressed.

    The Commission also has to do more to ensure that the results of EU-funded research projects in AI are fully commercialized or exploited. The recent EU measures to achieve a single market for data are still at the inception phase, and cannot immediately boost AI investment.

    The European Commission responded that it has taken note of the audit report and referred to the new AI act as proof that the EU has demonstrated its commitment to fostering the development of “trustworthy AI systems and responsible innovation” in Europe.

    As regards funding, the Commission referred to the AI Innovation package in January 2024 which includes an increase of €4 billion in investments in generative AI. The EU is already investing more than €1 billion a year in AI research projects under the Digital Europe and Horizon Europe programme, with the aim of attracting more than €20 billion in annual investment in AI in the EU over the decade.

Users of Guests are not allowed to comment this publication.