Europe to rival U.S. and China with 'Responsible AI"?

Hamburg News: Mr. Bogula, what is actually meant by Responsible AI?

Werner Bogula: Responsible AI is a holistic approach designed to evaluate and guide the use of AI beyond technical feasibility. We at ARIC are mulling its use in technology, economics, law and society. Human beings are at the centre of all these fields. A number of stakeholders are impacted by the technology when it is used in business and government and their voices need to be heard.

Hamburg News: Which stakeholders are they?

Bogula: I am referring to employees into whose work this technology is being introduced. If they are not involved from the start, distrust and resistance can arise that ruin the advantages of AI. No one likes to be involved in new processes that are determined by machines. That applies particularly to AI, which is fraught with fears and reservations.

Hamburg News: Aren't such fears justified perhaps? AI could replace 300 million jobs, according to a study by the U.S. bank, Goldman Sachs, in April.

Bogula: There are different forecasts and opposing views such as a study by the International Labour Organization (ILO) published in August, according to which generative AI will create more jobs than it wipes out. But at the moment, the bigger problem is the shortage of skilled labour - not job losses - in Germany and Europe. Nearly 68,000 IT vacancies nationwide including 4,000 in Hamburg's local government were registered by the Competence Center for Securing Skilled Workers in 2022. Given the backdrop of demographic change, businesses and local government would do well to digitize existing areas of responsibility so that AI relieves employees of routine tasks and frees up space for higher-value activities. This would lead to more qualified activities instead of more pressure.

Hamburg News: How can other stakeholders i.e., customers and consumers be won over to AI?

Bogula: Through education. First of all, it is important to comply with the framework conditions of data protection and data security. But beyond that, we have to address distrust and reservations about technology, e.g., through information events. We at ARIC explain how AI works and discuss the implications in over 150 annual events. The focus is on informed judgement to overcome a fundamentally defensive, technophobic attitude. Trust in the reliability of technology can only be created through education.

Hamburg News: So is trust a key factor?

Bogula: That's right. Even the best arguments are useless without trust. Therefore, when using AI responsibly, care must be taken to communicate openly and transparently what exactly AI does. Legal certainty, traceability and fair use of AI are crucial - basic trust that AI results are legally compliant, that decisions can be justified or challenged and lastly certainty that no one is favoured or disadvantaged by AI because of certain traits.

Hamburg News: These factors will be included in the EU's upcoming law to regulate the use of AI throughout Europe. Doesn't that create additional hurdles and make the use of AI even more complicated?

Bogula: It is better to establish a regulation in advance for all of Europe than to see the use of AI torpedoed in lawsuits and individual proceedings afterwards. Naturally, it is important to balance all stakeholders' demands including those of the business community. Ultimately, a purely fear-driven culture of prevention harms employees and consumers. What happens then can be seen in medicine where data sets are bought from the U.S. and China because of German data protection laws. Another example is AI models, where U.S. companies create products that spread unasked to European companies via standard software such as Google Search or Microsoft Office. The fact that Europe does not have a strong AI industry to represent its interests means that the U.S. - not us - is setting the standards. 

Hamburg News:  The final regulation is not yet clear. How should German companies deal with the uncertainty at present?

Bogula: The worst attitude is wait and see. Interestingly, employees, not management, in many companies initiate the use of AI to simplify their work. Companies must take a proactive approach and use AI with staff to improve productivity and to become more engaging for everyone. That includes guidelines for channelling the proliferation of ChatGPT and then developing their own AI procedures. Given the rapid development of AI, five-year plans with a polished AI strategy are less helpful than flexible trial and error in individual areas. Just keep at it.

Hamburg News: Is flexible trial and error enough to keep up with the market leaders in the U.S. and China?

Bogula: Of course not. It's about overcoming the typical "wait and see" approach and showing everyone how productive and efficient AI is for both employees and customers. That creates a demand for reliable and trustworthy AI, for which a market is developing in Europe. However, to meet this demand, many more enabling technologies such as large language models like ChatGPT should be developed here.  We are also very strong in AI research in Europe, and many start-ups are working on interesting products. We still need to step up when it comes to deep tech companies. Deep tech means basic technology for application-oriented AI. The Heidelberg-based company Aleph Alpha, has developed a European version of ChatGPT for use in industry and administration according to European standards. We can use that to counter the U.S. and China.
ys/pb

Many thanks for the interesting conversation, Mr. Bogula.

Interview by Yvonne Scheller.

Read the other parts in our AI series:
 

1) Hamburg clearly an AI hotspot, says Alois Krtil

2) AI Summit 2023 focuses on doom or boom aspects

3) European AI law explained

4)  AI’s impact on medicine