In a structured scenario process, you have spoken with experts about the future of artificial intelligence in Europe. The result of your scenario sprints are five scenarios. In three of those scenarios Europe succeeds in catching up with AI. What are the requirements for that?

What all three successful scenarios have in common is that they do not happen by themselves but are based on considerable efforts. This involves either targeted research funding for new technical approaches, massive investments in security and credibility of AI, following a “wake-up call,” or the large-scale expansion of infrastructure and training.

Let’s start with the most ambitious scenario in which Europe does not only catch up but also takes on the global leading role: ”Europe Leads a unique AI Revolution”.

This scenario describes a world in the year 2040 in which the EU undertakings are more innovative than other world regions. They have succeeded in overtaking the large AI centers USA and China with new technologies and more efficient approaches to train and run AI models

Reason for that: the tech giants have until now completely concentrated on generative AI and invested billions in their Large Language Models. Therefore, they are stuck on this technology path. While Europe in this scenario has developed through targeted research funding, creative researchers have explored completely new technology approaches which are not completely based on probabilities.

Generative AI is not the only AI technology, there are many more, for example predictive AI, hybrid AI, embodied AI or State Models. In this scenario, it was not determined which AI technologies or method combinations are involved but the experts who were involved in the scenarios think that it is possible that a European technology shift is happening, which becomes an entrance ticket to the global top .

The second scenario, “The EU Safety First Payoff in the Long Run”, concentrates on a different aspect: Safety and trustworthiness of AI. What are the preconditions for this scenario?

The scenario begins with an AI accident. The details are deliberately kept open, but this incident serves as a wake-up call. Something serious has happened, AI is suddenly up for debate in its entirety. The public is therefore alerted and demands safety first - specifically, across the entire spectrum from the developers to the users.

This leads to massive investments in security and credibility of AI. That way AI can step by step be established in niches and beyond. The European approach pays off in the long run, therefore “Payoff in the Long Run”.

The vertical use of AI plays an important role in this scenario: AI is used for different applications, thematic fields or specified areas, in contrast to horizontal AI which is supposed to be able to do everything and is used in all areas of application.

The third scenario relies on continuous efforts: “A rearmed EU among the frontrunners in the global AI race”. What is this about?

In this scenario, Europe invests in military purposes and rises to the global top group. The success is based on the expection that EU member states have massively invested in infrastructure and training until 2040.

It is a scenario in which Europe catches up with the leaders in the field of military buildup through appropriate investments.

Two scenarios remain. The fourth scenario “EU attached to Transnational Global AGI and ASI Networks” appears to give up the aim of an own leadership and instead counts on international collaboration.

Exactly. This scenario concentrates on artificial general intelligence (AGI) and artificial superintelligence (ASI). Both terms describe a state in  which artificial intelligence can rival human intelligence and its ability to navigate the world. While AGI aims to reach human intelligence, ASI exceeds this by far. In this scenario nobody leads anymore but artificial intelligence decides partly by itself. It is owned, however, by large companies which profit from them.

The EU is only attached, profits from preliminary work which is done elsewhere. It uses AGI and ASI services to its own advantage and to develop its own attractive services, especially in the area of education. In this scenario the issue of AI leadership is no longer relevant, because AI that has taken on a life of its own, it ultimately doesn't matter where it comes from.

And the last; the negative scenario?

The fifth scenario is called “The Global AI Winter” In this scenario the AI-Boom is gone before 2040 because the high expectations of AI have not been fulfilled.

It is crucial here that it has not been successful to get AI hallucinations under control. However, business critical processes need robust systems and reliable results. In this scenario, AI tools and AI agencies could not be efficiently integrated into business processes, also sophisticated AI-data research centers were not able to eliminate fundamental flaws of generative AI.

Consequently, the investment bubble burst and a global stock exchange crash happened. The disappointment led to the fact that the AI research had to start from scratch again and funding money was invested in other areas.

Besides this global power shift the question of the innovation potential of AI is decisive for Europe’s future. Many say AI means in the long run the end for innovations, because only things that already exist are combined, AI cannot therefore be creative and nothing new can originally be created. What do you think?

This fear is very much present. The argument is: AI was trained on the basis of already existing information. When developing new models, it draws on these existing resources-some of which were even created by AI- that would spell the end of innovation.

There is also another way of looking at things: humans are limited in their capacities and cannot work through all possible combinations. AI, on the other hand, can generate an extremely large number of possibilities. This way, AI can present us with the approaches which we ourselves develop further, it can provide us as sparring partner with original input for innovations, it can give us ideas which otherwise we would not have had. We can even tell AI to produce unusual connections, to transfer concepts from certain areas or allow fuzziness-strategies which normally give us new ideas. AI doesn’t therefore have to directly generate innovation but can deliver starting points.

In certain areas of science, AI can contribute by its constant recombinations to accelerations of innovations. For example, researchers can tell AI in automated laboratories to run through many variations. Particularly in chemistry, in material sciences, in the medical area and in pharmacy, this can give us new findings.

However, the automated science has limits, especially when it is not about classical numbers, dates and facts in clearly definable areas, but about contexts which are between the lines as it is often the case in texts. AI doesn´t help us here.

When it comes to the question, how AI and innovations are related, it was clear in all our discussions: humans have to decide at the crucial points.

Thank you very much for your time.