Why the EU standard for AI is the best thing that we can do right now?
At the moment AI is rigged against European Union as we know it. The biggest companies come from the United States, and they are very interested in to see that the coming international standard is forming around their existing products. The US companies are very active on AI standardisation right now, and their interest is clear: creating an AI standard internationally, that support best the American industry for the expense of others.
China is also a major player in the field of an AI. They have the best AI solutions and machine learning algorithms today (even if you ask from Americans), and in China has also some extraordinary advantages that almost no other country has: very relaxed surveillance law and a very large data-pools in their disposal for the AI development… and AI needs lot of data in order to be excellent.
European Union has been active on thinking of the dangers of an AI and trying to modernise the industry, but in here, there really hasn’t been any major development on AI itself. Now there is a perfect opportunity to step up EU’s AI game and start to define how we are going to develop it in the future. Today EU version of the AI product may not be very attractive choice for a company CEO when he has to decide between the leading products. EU standard as such may be a compelling reason to look at also European alternative.
A Good standard would define AI as a modular entity where also modules communicate in the standard way. That would give to European industry the possibility to make only those AI modules that they are best at. It would be very cost efficient and result to faster development in this field than we have today.
A modular structure would give us also some interesting choices in a multicultural Europe. For example, Language module could be chosen by each area where the AI is used. Local laws may also differ in certain countries, so there should be a module for law also since we all want AI to be law obedient entity, don’t we?
Artificial intelligence does have an affect into our lives with such an impact, that we really cannot think AI standardisation as a technical rehearsal any more. We must adopt all the looming questions around it and form together so good model, that it includes also practices, laws, ethical questions and what we come up to in the future. Safety can be planned so that data access, logging, control and feedback can be reached only by the authorised personnel and data is anonymous for the GDPR. Neural models and quantum processing will bring the development to such a speed that the singularity is imminent, If we do not act now. This really is the 11th hour for us, AI and of course for European Union.
In a global scale European Union is the best party that can make a real AI standard to happen. Despite of all Our differences, we are Peace driven, multicultural, multilingual and we do respect human rights. European Standard could be easily omitted to other areas also, like Asia and Africa , and that would widen our AI market significantly.
I really hope that the European Union doesn’t let this opportunity to pass by and it can make to EU standardisation office an initative of this matter and oversee the process. There's a lot of good EU discussion ongoing already, so lets make the best of it.
Kai Salmela, AI Specialist Robocoast R&D