It’s the new must-have accessory for any country that fears trailing in the dust of modernity.
Nations such as China, Finland, France, the UK and India, plus the European Union, have all released artificial intelligence strategies of sorts over the past year, hoping to ready their economies and societies for a wave of upheaval and opportunity afforded by increasingly smart machines.
Researchers and universities are seen as key: announcing France’s strategy in March, president Emmanuel Macron earmarked €400 million (£349 million) for AI research.
The strategies include a wealth of suggestions – and dilemmas – as policymakers grapple with how best to support research into AI, link it with other disciplines and feed the results through into the economy.
Perhaps the most pressing problem for AI research, particularly in Europe, is that universities cannot match the lucrative salaries on offer at tech giants in Silicon Valley.
The French approach, set out in a report called For a Meaningful Artificial Intelligence, suggests doubling the salaries of AI graduates who work in public research institutions, otherwise the flow into universities could “dry up completely”.
Marc Schoenauer, a former president of the French Association for Artificial Intelligence and part of the team that drew up the report, admitted that this was unrealistic. “We know it will not happen and it will not be enough anyway,” he said.
“We just can’t compete on funding,” said Stefan Heumann, co-director of Stiftung Neue Verantwortung, a German technology thinktank that has set out recommendations in advance of a German AI strategy expected in the summer. Instead, Europe can offer researchers a more ethical approach to AI, a good lifestyle and a democratic society to live in, he said. With the US under Donald Trump potentially becoming more hostile to Chinese AI graduates coming out of American universities, Europe had a “real opportunity” to attract them, he added.
The US, meanwhile, has no national AI strategy – something that has drawn criticism from both US politicians and the chief executive of chipmaker Intel. Dr Heumann argued that the US has in effect delegated its AI direction to big technology firms.
Another question for countries is whether to invest in basic or applied AI research. This question is causing a “big debate” in Germany at the moment, where basic research has traditionally been strong, said Dr Heumann, but Germany has been “poor” at turning it into world-beating technology companies, “the stuff Silicon Valley is good at”.
The French report stresses how crucial it is that public institutions keep researching areas of AI that are not currently fashionable – or profitable – in the hope that they will eventually bear fruit, “catching these AI giants with all their immense resources by surprise”. Big technology firms tend to have to chase hyped areas of research – such as deep learning – because they have to justify their research with a commercial benefit, said Dr Schoenauer. Public research can take a longer view.
Computer science is not the only discipline that could benefit from an international push on AI research. Mr Macron has stressed he wants philosophers and social scientists to be involved from the very beginning to explore the wider implications of AI advances. “It has to be interdisciplinary. It’s mandatory. AI will be everywhere,” said Dr Schoenauer.
The UK’s strategy, Growing the Artificial Intelligence Industry in the UK, meanwhile, is far more focused on providing businesses with workers skilled enough to work with AI in the future. It suggests a raft of new master’s and PhD programmes.
But the real elephant in the room is money. Mr Macron’s promise of €1.5 billion may sound significant, but it was actually a re-announcement of funds pledged by his predecessor, explained Dr Schoenauer.
And the contribution is tiny compared with the money being poured into AI research by industry: the consultancy McKinsey has estimated that the world’s biggest US and Chinese tech companies collectively invested between $20 billion (£15 billion) and $30 billion in 2016.
The sheer amount of funding now spent privately is “a challenge to the European approach” – where the state drives higher education and shapes the research agenda – warned Dr Heumann.
Eyes on the prize: what national AI strategies say
- Simplify the country’s myriad related research centres into two bodies: one focusing on basic research, and another on deploying it
- Lead the development of a “Cern for AI”, an international research effort that shares results with the entire world.
- Industry to sponsor a “major cohort” of master’s-level courses in AI; government and universities should create at least 200 PhD programmes
- Universities to look into creating one-year conversion master’s degrees in AI for graduates who have not studied computing and data sciences.
- €1.5 billion in extra AI funding by 2022, including €400 million on competitive research calls
- Four to six interdisciplinary institutes for AI.
- Become a leading player in AI research and development by 2025
- Lead the world in the field by 2030.