When the German government announced its national strategy for artificial intelligence in 2018, it had big plans. People spoke of a “key to the world of tomorrow” and of charting a path to becoming the “world leader”. The message was clear enough: it was about nothing less than securing “Germany’s future as an industrial location”. The strategy defines twelve fields of action for achieving the stated goals, research being chief among them. To this end, it calls for establishing “a strong, dynamic, flexible, wide-ranging and interdisciplinary AI ecosystem that is internationally competitive.” This ecosystem has now taken shape in a network of six flagship research institutes located all across Germany (see overview on page 16). Having been set up over the past three years via project funding, these AI competence centers will be permanently established through institutional funding as of July 1.
Ready, Set, Research!
The German AI competence centers will be receiving institutional support starting this summer in accordance with the AI strategy set by the previous federal government. If the country is really to be a world leader in artificial intelligence, however, this should only be the beginning – and that’s something the new government knows, as well.
By Martin Schmitz-Kuhl
But does that mean Germany really is on its way to becoming the world leader in AI? Not even the competence centers themselves answer that question with a definitive “yes”. Prof. Antonio Krüger, managing director of the German Research Center for Artificial Intelligence (DFKI), points to “some top-class people – internationally recognized! – who are involved at the highest level”. Prof. Katharina Morik, spokesperson of the competence center in Dortmund/Bonn, reports that “outstanding individual achievements” have been made. That said, none of the new centers can claim that the German AI ecosystem has crystallized into a stable structure. And Prof. Klaus-Robert Müller, co-director of the Berlin Institute for the Foundations of Learning and Data (BIFOLD), stresses that while German AI research can hold its own globally, it can only be described as good in proportion to the resources at its disposal. After all, it’s not only the Chinese and Americans who are investing much more in research. “My colleagues in Montreal receive at least ten times as much funding as we do here in Berlin,” Müller laments. He wants this to be clear to all the people in charge who are boasting about Germany becoming a world leader.
Only 15 of 100 AI professorships filled
Krüger, Morik and Müller agree that there is a shortage of professors in Germany who not only do research on artificial intelligence, but can teach it, as well. The demand for skilled workers is huge. But where are they supposed to come from when even the new competence centers often have no more than a handful of professors? At the beginning of May, it was announced that the AI strategy’s program to fill 100 additional professorships had been successfully completed. It’s unclear, however, whether this refers specifically to the professors so urgently needed in the field of AI. "We don’t need ‘hyphenated’ AI professors – people from other disciplines who only use AI – but people who work in AI themselves, meaning in developing and advancing new methods," Morik stresses.
Another problem is that it wasn’t possible to advertise open positions until the institutional funding started flowing. As a result, only 15 of the 100 professorships have been established in the university competence centers, mainly in Munich and Tübingen. Only the financially strong states of Bavaria and Baden-Württemberg had the resources and the political will to push ahead and invest in expanding the personnel in their competence centers. For everyone else, the “hunt” (as Morik puts it) for the few experts available has only just begun. And the centers are competing not only with each other, but also with other locations all over the world – no easy task when people can choose whether they want to go to Dortmund or Dresden or to Cambridge or Granada. “It really is a question of how we can get new professors, because a normal job posting is definitely not going to cut it,” Morik concludes. To acquire good people, you have to be attractive. To be attractive as a location, however, you have to have good people already. “It’s a real dilemma,” Morik admits.
Is it competition that stimulates business, or cooperation?
This competition among the German AI centers does not serve the mission for which they were set up. Former Minister of Education and Research Anja Karliczek (CDU) introduced a coordination unit that was supposed to facilitate cooperation within the network in the name of a joint national strategy. However, Karliczek failed to provide this office with the appropriate funds, as Morik notes with some annoyance – after all, it was she herself whom the minister entrusted with the coordination effort. Prof. Müller pushes back here a bit, stressing that he has always exchanged information with colleagues anyway, that research should know no borders, and that a certain level of competition among the centers could surely stimulate their business and research (see interview at https://dialog.vde.com/en). On the whole, though, he agrees with Morik and Krüger that a coordinated approach and a common public image would be useful. “Especially since we want to operate AI at scale in Germany. The network of personal relationships between individual professors is simply not going to get us there,” Morik points out.
This brings us back to the €50 million that are now to be spent on the competence centers each year. And back to the AI strategy, which calls for an amount 100 times larger to be allocated for artificial intelligence research from 2019 to 2025. This has people at the competence centers wondering why their institutions – the “flagships of German AI research” – are getting such a relatively small piece of the €5 billion pie (Morik: “I’d like to know where all the money is going”). Will the money actually be used to make Germany a world leader in AI? The latter question also piqued the interest of German parliament member Dr. Anna Christmann (The Greens), who requested corresponding information from the German government last September. The answer: of the €5 billion allocated for the AI strategy, almost €3.5 billion have been included in the budgets of the federal ministries to date. Of that amount, however, just €346.5 million – less than 10 percent – have been spent. In Christmann’s unvarnished opinion, the federal government’s “supposed modernization offensive” thus stands in “stark contrast to its miserable record on future-oriented technologies”. She herself is no longer a member of the parliamentary opposition, by the way, but rather the coordinator for the digital economy and startups in the Ministry for Economic Affairs. This makes her partially responsible for ensuring that the money is spent and received, and nearly a quarter of the total AI funding (just under €1 billion) is contained in her own ministry’s budget.
The change of government at the beginning of December would have been a good time to get things going in other ministries, too, especially in the Ministry of Education and Research. While Angela Merkel’s grand coalition already put in place a national AI strategy in its first six months in office, the new coalition of the SPD, the Greens, and the FDP has not done much on AI in the same period. After multiple requests for comment, a spokesperson for the Ministry of Education and Research responded in late May that it was hard at work implementing the AI strategy and promoting the ongoing development of the competence centers. “Further measures are planned and will be announced in a timely manner,” she stated.
German government relatively silent on AI thus far
At the time of this writing, the exact nature of these measures remains unclear. Prof. Krüger, the head of the DFKI, nevertheless suspects that technology transfer in particular could be accelerated under the new liberal leadership at the Ministry of Education and Research. It was just last year that Dr. Thomas Sattelberger, then spokesperson for the FDP parliamentary group for research and innovation, caused a stir when he called the DFKI a dinosaur, “sluggish and hardly innovative”. And this is meant to be the business-oriented AI competence center that, unlike the university-affiliated centers, has devoted itself entirely to applied research and technology transfer. If this center focuses too much on developing prototypes instead of actually bringing them to market, as Sattelberger criticized at the time, what could be said about the other five – alleged or actual – “ivory towers”?
Germany has long had an indisputable reputation for good (foundational) research, but the country falls short when it comes to “cashing in” on its own innovations in the form of products. While this problem is by no means limited to artificial intelligence, Antonio Krüger, Katharina Morik and Klaus-Robert Müller openly admit that it does affect their field. The reasons for this are manifold, however, and it wouldn’t be fair to blame the research centers. Morik stresses that it wouldn’t be possible to train enough skilled workers to transfer AI expertise to industry and business. Müller, meanwhile, points out that venture capitalists are too risk-averse, which leaves startups with too little money to make big innovations happen. And Krüger laments the apparent lack of entrepreneurial spirit among young people, along with the general absence of a startup culture in German society. “There’s certainly still room for improvement in that respect,” he says.
At the same time, all of them insist that Germany is heading in the right direction. The DFKI is already implementing 90 percent of its projects in close cooperation with industrial partners. The city of Berlin has been focusing on becoming a more attractive place for startups for years now, having long since recognized its particular advantages in this area. Other places, including the Rhine-Ruhr region, are also making every effort to arrange strategic partnerships with industrial firms. In addition, there are outside initiatives such as the KI Park network, of which VDE is a founding member and a permanent member of the administrative board.
However, the future of AI research will depend in no small part on how the government handles the topic going forward. Will the industry gain momentum, as many hope, and command more attention? Or could current challenges and financial burdens such as the COVID-19 pandemic and the war in Ukraine halt its progress, as others fear? The resignation of one Thomas Sattelberger could be regarded as a bad sign in this context. After the recent change of government, the once critical FDP lawmaker saw himself moving up in the world. He became not only Parliamentary State Secretary in the Federal Ministry of Education and Research, but also the commissioner in charge of science transfer and scientific startups – a position that was specially created for him. Among other things, he planned to set up the innovation agency DATI, but it didn’t quite get off the ground because the Budget Committee refused to allocate the necessary resources. Sattelberger stated that his departure at the end of May was due to personal and health reasons. However, his decision could also be interpreted as a “veiled political protest” (as posited by the Tagesspiegel newspaper). As of this issue’s editorial deadline, Sattelberger had at least not gone as far as to refute this theory.
Martin Schmitz-Kuhl is a freelance author from Frankfurt am Main, Germany, and an editor of VDE dialog.