Where’s the European Union when it comes to AI?
It has the world’s second-largest GDP — $18.4 trillion in 2023. Its 448 million population is more than 100 million greater than the U.S. It is home to many top educational institutions with some of the world’s leading scientists and regularly cranks out more. Put the United Kingdom back in and it is an even more impressive economy.
Yet, as the AI revolution takes off, the EU and the U.K. do not have a single major company playing an important role. That continues a long record of failing to spawn global digital players or even pan-European ones, with few exceptions.
Artificial intelligence, however, is a real wake-up call — like your carbon monoxide detector going off in the middle of the night.
The numbers are stark.
As of 2022, the EU and the U.K. had 8.9 percent as many granted AI patents as the U.S. and 3.7 percent as many as China. Patent applications, which measure more recent efforts, are not much better. In 2022, the EU and U.K. had 11.3 percent as many as the U.S. and 2.8 percent as many as China.
Patent grants and application counts are an imperfect measure of innovation. The EU and EU+U.K. shares are so tiny relative to the US and China, however, that they point to far less innovative effort going into AI.
Private investment in AI looks better but still portrays the EU as the tortoise to the U.S. hare. Between 2018 and 2022, private investment in AI was $206.3 billion in the U.S., compared to $82.3 billion for China and $37.5 billion for the EU plus the U.K. EU+U.K. private investment was less than a fifth (18.2 percent) than in the U.S. (Patent data is sourced from the Center for Security and Emergency Technology and investments based on the 2023 NetBase Quid via AI Index Report.)
The share of AI startups deals does not portend well either. Europe accounted for 21 percent of AI startups globally between 2019Q1 and 2023Q3 compared to approximately 40 percent for the U.S. (based on the average quarterly shares). (CBINSIGHTS Q3 2023 State of AI report. Europe is not defined but appears broader than the EU and U.K.)
Artificial intelligence investments could have been a bust. For decades, the field had moments of great excitement followed by long periods of disappointment as scientists found how hard it was to create something that might one day be called intelligence.
It wasn’t that long ago that many had written it off. A recent New Yorker profile of Jensen Huang, the CEO of Nvidia, summarizes the situation a decade ago:
“At the beginning of the twenty-tens, A.I. was a neglected discipline. Progress in basic tasks such as image recognition and speech recognition had seen only halting progress. Within this unpopular academic field, an even less popular subfield solved problems using ‘neural networks’ — computing structures inspired by the human brain. Many computer scientists considered neural networks to be discredited.”
Not long after the start of the decade, though, the neural networks proved their stuff. Nvidia double-downed its focus on AI chips and supercomputers, which became essential for computational-intensive foundational AI models. Venture Capitalists ratcheted up startup funding. In 2017, Google researchers invented the pathbreaking transformer architecture. Between 2013 and 2017, private investment in AI in the U.S., EU, U.K. and China almost quintupled from $5.5 billion to $25.6 billion. (2023 NetBase Quid via AI Index Report.)
Even then, it wasn’t for sure that AI would truly take off outside of some specific applications, deliver a payday that made all those large, risky investments worth it, and have the possibility of fundamentally changing the world.
The release on Nov. 30, 2022, of ChatGPT by OpenAI, its massive adoption by consumers, and ChatGPT’s rapid improvements in the months to come demonstrated the power and potential of AI.
The renewed promise of AI, backed up by an increasing number of practical applications, did not go unnoticed in Europe in the mid-2010s.
On April 25, 2018, the European Commission presented “a series of measures to put [AI] at the service of Europeans and boost Europe’s competitiveness in this field.” It proposed a three-pronged approach based on investing in AI, preparing for socio-economic changes and ensuring an ethical framework.
Later that year, the Commission announced that the Member States and the Commission would “work together to boost artificial intelligence ‘made in Europe.’” Mariya Gabriel, the Commissioner for Digital Economy and Society, said, “The coordinated action plan will ensure that Europe reaps the benefits of AI for citizens and businesses and competes globally while safeguarding trust and respecting ethical values.”
More than five years later, it appears that the EU has little to show for its efforts. There is not much in the way of AI products being “made in Europe,” and Europe does not “compete globally.”
Politico wrote a scathing summary of the EU’s approach on the day of its April announcement: “Europe’s secret weapon in the race against the U.S. and China on artificial intelligence is … ethics,” where the ellipsis implies breathless anticipation not missing text. The article claims that the block believes that preserving fundamental rights “will boost consumer trust in European AI applications and help the Continent —which lags far behind the U.S. and China in building a state-of-the art AI industry — catch up with competitors.”
That’s too harsh. The EU had sound reasons to consider ethics. That shouldn’t have prevented success in AI. In fact, OpenAI was founded on the premise that it had a “fiduciary duty to serve humanity.”
Yet, EU, here we are.
More worrisome is the fact that the Commission seriously thought that regulations were a core component of closing the long lead held by the U.S. and China in AI. And that a focus on ethics could lead to a different outcome for EU than the previous two decades of digital tech.
At the end of 2023, to the extent the EU has a seat at the table, it comes from its progress in developing AI regulation. The European Parliament passed the comprehensive Artificial Intelligence Act in June, and the EU Council, Parliament and the Commission are negotiating its final terms.
It could become another notch in the belt for the Brussels Effect. That’s the term Professor Anu Bradford coined for EU regulation becoming the global standard, which companies comply with because they don’t want to lose the EU market or because other jurisdictions adopt the same or similar rules.
That seems like a poor consolation prize, whatever you think about the AI Act.
There’s an old saying that success has many fathers, but failure is an orphan. The problem in the EU, when it comes to digital successes, is that it seems that failure has many fathers, but no one knows for sure who they are.
Regulation, for example, could be a cause — making it harder for EU businesses and their investors to innovate. Or a correlate — engrained risk aversion that embraces the certainty of regulation and abhors innovation risk. Perhaps a bit of both.
Artificial intelligence may be the shock that forces the EU, or the Member States, to become laser-focused on coming up with real solutions for driving innovation in AI and other cutting-edge technologies.
In fact, the latest news on the EU’s AI Act is that France, Germany and Italy are pushing back some of the stringent regulations of the Act. That is partly from pressure from AI startups in those countries.
According to Tech Policy Press, “German company Aleph Alpha and French start-up Mistral … fear that excessive regulation of foundation models in the EU could put them at a massive disadvantage compared to their American and Chinese counterparts.” Those countries are now reportedly pushing for “self-regulation based on a code of conduct for foundational models” in the trialogue negotiations. We will know soon.
Peter Thiel famously said about digital innovation, “We wanted flying cars. Instead we got 140 characters.” Long term, it probably doesn’t matter that the E.U. doesn’t have its own global social media business or other ad-supported platform.
AI is different, though, because it will touch and have the ability to dramatically improve so much of the physical economy, from healthcare to cars to agriculture. Europe will need global AI players, which need not be foundational models, to drive its economy. And it will have to be sure that its businesses can incorporate the best of AI, wherever it comes from, in their products and processes.
David S. Evans is an economist who has published several books and many articles on technology businesses, including digital and multisided platforms, including the award-winning Matchmakers: The New Economics of Multisided Platforms. He is currently the Global Leader for Digital Economy and Platform Markets at Berkeley Research Group and Chairman, Market Platform Dynamics. For more details on him, go to davidsevans.org.