2023-07-15 12:15:40
Artificial intelligence (AI) is on everyone’s lips and many are wondering how AI could one day change our everyday lives. This process is already in full swing: AI is making weather forecasts better, the analysis of X-ray images is more precise and traffic flows can be managed more efficiently. Better, more precise, more efficient. AI offers us opportunities to tackle key societal challenges.
In order for Germany and Europe to play a leading role here, we must put these opportunities at the center of political decisions. The EU started working on a legal framework a few years ago. The AI Act, with its risk-based approach, is fundamentally the right approach. The requirements for users and companies should increase with the risk of the respective application. Different rules must apply to AI systems that are used in critical infrastructure (energy supply, traffic management) than to service chatbots that provide advice on holiday planning.
Europe needs courage
The debate has gained new impetus through the advent of generative AI, i.e. the applications that – like ChatGPT or Dall-E2 – generate new content from the analysis of very large amounts of data and thus open up completely new possibilities, but at the same time pose completely new questions. How does the user know how such an AI was programmed, can he rely on the results, who is liable for wrong decisions? There are hardly any suggestions for this in the current draft – and we should see that as an opportunity.
Volker Wissing (FDP), Federal Minister for Digital Affairs and Transport : Image: EPA
Forcing a rapidly evolving technology into the rigid framework of a law is not the best way to bolster Europe’s uncompetitiveness in this area. What Europe needs is courage. Courage not to limit innovations in their development, but only to set guidelines for them.
In concrete terms, this means not expanding the scope of the AI regulation any further and instead closely accompanying the development of generative AI – and preferably not only nationally or European, but globally. The G-7 heads of state and government laid the foundation for this back in May and agreed to start intensive discussions immediately. The so-called Hiroshima AI process could complete common guidelines for dealing with generative AI this year.
Three questions arise
It would be good for Europe to translate the results of such a code of conduct into mandatory self-regulation. We have already gained important experience with such instruments in the context of cloud technologies and in the fight against disinformation and punishable hate speech, which are now being applied.
To do this, we have to answer three questions: What do we want to address? When it comes to generative AI, we currently have more questions than answers. A code of conduct could create clarity. This starts with transparency about the data used for programming, goes beyond a minimum of openness to algorithms and does not end with the training of programmers to avoid bias.
Roland Lindner, New York Published/Updated: , Recommendations: 45 Ralf Otte Published/Updated: , Recommendations: 29 Adrian Lobe Published/Updated: , Recommendations: 2
Who should implement it? Here it must be clear that we should not leave it to the big companies alone. Meta, Microsoft and Co usually have no difficulties with the implementation of regulations because they have sufficient resources. However, small and medium-sized companies that work in niches on generative AI and start-ups that not only develop cutting-edge technology in the EU but are also supposed to operate sustainably are left behind. Therefore, we should make selected associations and standardization bodies the central actors of self-government, since SMEs and start-ups are represented there.
How should it be tracked? Of course, mandatory self-regulation must go hand in hand with a tight review process. After two years we have to look at the cards: have we achieved our goal? Do we have to adapt the Code of Conduct? Or is there a need for strict regulation and the integration of generative AI into the AI regulation. With mandatory self-regulation for generative AI as a supplement to the AI regulation, we can strengthen Europe’s competitiveness and at the same time show courage. Courage to allow innovations.
#Guest #article #innovative #regulation