I spent half a day listening into an FT Live event (end January 2024) in Brussels on “A Collaborative Approach to AI implementation and regulation”. The speakers ranged from MEPs and parliamentarians to representatives from citizens groups and the tech world. Below were my 10 key takeaways of the mood music around AI in Europe:
1. The AI act is here. After painful trilogue talks between the EU Commission, Parliament and Council, the member states now have a text to consider. [Update: The member states unanimously ratified it on February 2nd].
2. But it’s only the beginning. Even once the member states ratify it, the provisions then need to be embedded by all the actors in society; the professional bodies, the consultants, the organisations. Standards and common practices will need to be defined and evolve. Brussels is only the beginning.
3. “European values”. Pervading the proceedings, there was a strong and idealistic sense of Europe needing to protect its citizens and hopefully set a standard that the rest of the world might emulate – it remains to be seen whether the so-called “Brussels effect” will permeate the world of AI, in the way that it impacted GDPR standards.
4. “A beautiful gift”. There was recognition from the speakers in the room of the possibilities of AI for humankind.
5. “Forks don’t kill people”. In addition, there was a robust defence by tech companies of the good the tech is already doing, plus how it’s up to us to manage the bad but not to stifle innovation.
6. The costs… of compliance will be significant, especially for smaller organisations. One speaker mentioned a forecasted cost of €300k, for even a small company and asked where the support was coming from. What seemed to be forgotten was that, of course, deploying AI is entirely optional, and the overall costs of implementing AI will greatly exceed those of compliance.
7. Promoting AI. There was recognition, at least in words (deeds to follow) that along with regulation, there is a twin responsibility, and industry would say an urgency, to support AI development, through supercomputing power, education, grants and assistance. Europe is in a race with the US and China to build the thriving ecosystem that will be needed to win in this space.
8. Talent. People are crying out for education, but it’s moving so fast. The Google head of AI in Financial Services said she is attending two lectures per week on the topic. There is a huge need to educate people, and perhaps less in the obvious areas of coding and more in topics such as ethics, fraud awareness and connected thinking.
9. Positive uses of AI. The tech speakers provided some mind-blowing examples of the power and potential of AI benefits already taking place, such as with genome sequencing, medical protein modelling – “a billion years of work condensed into a few months”, weather forecasting and translation.
10. EU investment in AI. The EU has already pledged to spend almost half a billion euro in digital technologies, including AI factories – supercomputing facilities with the processing power to facilitate the development of AI capability. It is also establishing an AI office, right beside EU Connect, to co-ordinate development.
In the race against the US and Chinese tech giants, there are causes for optimism. Take for example the Finnish company, Silo, that developed a language model using an EU-funded language database boasting 7,000 terabytes of language data across 80 languages (by way of comparison, ChatGPT-3.5 was trained on “just” 45 terabytes).
The EU has proven itself exceptional at legislating across 27 countries. Will it prove to be as capable in growing and fostering innovation? “Europe is only the front-runner in AI when it comes to regulation” – said one commentator from a Brussels-based think tank – ouch. The EU has taken the lead in legislation – with certainty in place on the compliance front, can it now launch a winning formula for AI to safely flourish?
