Europe AI Irrelevance Risk: 5% Global Compute vs US 80% — Policy Thinkers Warn of 'Economic Backwater'

Europe has five percent of the worlds Artificial Intelligence computing power. The United States has eighty percent.

This one fact, which is in a report from Bruegel, a very respected economics research group in Europe shows how big the Artificial Intelligence problem is for Europe. The European Union has an investment plan worth two hundred billion euros, which is about two hundred and thirty billion dollars. However this plan is spread out over years. On the hand United States technology companies spent more than four hundred billion dollars on Artificial Intelligence infrastructure in the year two thousand and twenty-five alone.

The people who wrote the report are very clear about what they think. They imagine a scenario, which they call "Europe two thousand and thirty-one" where the European Unions rules are too strict and its goals are not ambitious enough. As a result the European Union will not be able to keep up with the United States in areas like defense, cybersecurity and international politics. The writers of the report think that Europe might become less important economically.

The Main Problem With The Artificial Intelligence Act

The problem is not just that the European Union is not investing money. The problem is also the rules themselves.

The European Unions Artificial Intelligence Act was created to make sure that Artificial Intelligence systems are safe before they are used. However Artificial Intelligence is not like products. It can work in different situations and do things that were not planned.

The writers of the report say that because Artificial Intelligence is so unpredictable it is not possible to protect against all problems before the systems are used.

The report also compares the Artificial Intelligence Act to the General Data Protection Regulation, which has made it harder for small companies to compete. The writers of the report think that the Artificial Intelligence Act might have the effect.

The Solution: Changing The Approach

The report suggests that the European Union should change its approach. Of focusing only on making sure that Artificial Intelligence systems are safe before they are used the European Union should also have rules for after the systems are used.

The report makes key recommendations, including:

1. Different rules for types of companies depending on their size

Small companies, like startups that make less than fifty million euros per year should have simpler rules to follow

Medium-sized companies, which make between fifty million and one hundred and fifty million euros per year should follow the current rules

Large companies, which make more than one hundred and fifty million euros per year should have to get their Artificial Intelligence systems checked by a third party

2. A new system for dealing with problems caused by Artificial Intelligence

The companies that make Artificial Intelligence systems should be responsible for any problems they cause

If an Artificial Intelligence system causes a problem the company should be held strictly liable

If an Artificial Intelligence system causes a less serious problem the company should have to prove that the system is not defective

3. More transparency about how Artificial Intelligence systems work

Researchers and auditors should have access to information about how Artificial Intelligence systems work

Companies should be encouraged to report any problems with their Artificial Intelligence systems without fear of punishment

The European Union should have a registry of incidents involving Artificial Intelligence systems

Three Important Facts

The report includes a model that shows when it is better to have rules before or after Artificial Intelligence systems are used:

Figure one shows that if an Artificial Intelligence system causes a small problem, like a delayed train it is better to deal with it after it happens. However if an Artificial Intelligence system causes a problem like someone dying because of misinformation from a chatbot it is better to have rules in place before it happens.

Figure two shows that if regulators can predict what might go wrong with an Artificial Intelligence system it is better to have rules in place before it is used. However if an Artificial Intelligence system causes a problem that was not predicted it is better to deal with it after it happens.

Figure three shows that if the companies that make Artificial Intelligence systems are too confident about their safety it is better to have rules in place before they are used.

Time Is Running Out

The European Commission has to evaluate the Artificial Intelligence Act in August two thousand and twenty-eight and can make changes by August two thousand and thirty-one. The report says that the European Union does not have time to wait and see if the Artificial Intelligence Act works. It needs to make changes

The report says that the European Union should not wait until two thousand and thirty-one to propose changes to the Artificial Intelligence Act. It should start making changes

With United States technology companies spending four hundred billion dollars per year on Artificial Intelligence infrastructure while Europe is struggling to spend two hundred billion euros over years the gap between the two is getting bigger and bigger. The scenario described in the report as "Europe two thousand and thirty-one" is imaginary. If the European Union does not take action it might become a reality. Europe and Artificial Intelligence are at a point. The European Union needs to take action to make sure that it does not fall behind the United States in Artificial Intelligence. Artificial Intelligence is an area, for the European Union and it needs to get it right.