Welcome, ladies and gentlemen. This session will be in French. We shouldn't leave technology only in the English language. I thank Aiman Ezzat for being with us this morning, CEO of Capgemini. I had the honor of being on the board of Capgemini for several years. It's a great company. Dear Aiman, let's try to cover a wide range. To start the discussion, I'll ask you to say a few words about the idea of digital as a true industrial revolution, perhaps a super revolution, and also about artificial intelligence as a revolution within the revolution.
Good morning, thank you for having me. On whether it's a revolution equivalent to the industrial revolution: yes, we have a digital revolution that changes all rules of competition. What consumers value is intelligence, the digital part. Look at company valuations: a few US companies represent 17-20% of S&P variation. Value is created in digital. A car's software becomes crucial for competitiveness. So all rules are changing, hence a digital revolution. Software developers and data analysts are the workers of this digital revolution. We are moving from a physical world to a convergence of physical and digital. It's more than an industrial revolution in terms of societal change. AI within the digital revolution is also a revolution in itself. AI is not new; it became mainstream with ChatGPT, but it's based on statistics and has been around for decades. Its importance grows because computing and data management capabilities become exponential. Generative AI allows handling text, voice, video. It's about exploiting data to improve decisions, accelerate innovation, increase efficiency. This is a true revolution because today's computing power allows things unimaginable even 10 years ago.
On AI, especially generative AI, the general public was surprised by ChatGPT and its miracles, which made everyone fantasize. The socio-political question is: haven't we come to expect too much from AI, especially in business, as if it would suddenly change everything? That is the question of reality versus expectations.
Absolutely. We had the AI hype cycle. Early last year, everyone talked about AI and said it would completely change everything in the short term. The reality is the difference between technological adoption and time. The potential of AI and generative AI exists, but adoption time is much longer because it requires changing processes, organizational modes, reflexes, and how people work. Changing reflexes is another matter. That's the reality of any technology.
Can you give some examples? Capgemini has plenty of experience.
Examples: What accelerated the hype was productivity in customer service, like call centers with smarter agents through generative AI. That created the hype, but applying it everywhere is not simple. In other areas, there are many assistants: in sales, in factories as operators' assistants suggesting the best next decision. The concept of human-in-the-loop is important; assistants don't decide but suggest options. There are client service assistants in many places. One area with great potential is R&D, especially in pharma: reducing development cycles, anticipating molecule failures, and shortening the 10-year drug-to-market cycle. Generative AI can have a significant impact there, and concrete advances are seen.
And can you give an example of failures, false leads?
The false leads are all attempts at productivity. I think the reports, including from the IMF, that said 400 million jobs would disappear due to generative AI were completely alarmist. That had a very negative effect: people were scared, so in companies, when people talked about generative AI, they said 'I don't want it because my job will disappear tomorrow.' That was too alarmist. In two years, we haven't seen the anticipated productivity levels. Historically, the relationship between employment and technological progress has always had early alarms (steam machine, railways), and they have always been wrong so far. It takes more time. Value arrives but takes much longer because it works through humans, not machines. Education, etc.
Still on AI, there is increasing demand from governments and regulators to regulate AI. I think this is a subject you have thought about a lot. We are impacted by what is happening. The idea is to find the right balance between protecting citizens and innovation.
My principle is to regulate the use of technologies, not the technology itself. But there are divergences: the US recommends guidelines, while Europeans regulate through the AI Act, regulating the development of technology itself. That is very dangerous because it kills innovation. Secondly, trying to regulate something new that we don't understand is dangerous. If we approached nuclear energy the way we approach AI, we would never have nuclear energy. So we are overregulating something we don't understand at the source, killing technological innovation.
It's the Europeans, I imagine?
Yes, the Europeans. The Chinese don't regulate, so they can do everything. The Americans give guidelines but leave it open. California risks swinging into the European camp. So it becomes very disparate across the world. For a company to function, if its models are in Europe or the US, they have completely different rules. It becomes nightmarish. The risk is that all innovation goes abroad because startups find it too complicated to operate in Europe.
Can you elaborate on that? California swinging into the European camp?
In terms of regulation, there is talk of legislation similar to European regulations on data protection and technology regulation. That is at the California state level.
Let's continue to cover the topic. There is more talk about sovereignty, the relationship between digital, AI, and sovereignty. It doesn't take a genius to see a potential contradiction: digital crosses all borders, so sovereignty implies erecting barriers. Would you like to elaborate?
