Why artificial intelligence is a philosophical topic

What is actually the USP of humans? Which activities can and should we delegate to artificial intelligences? Jonas Moßler, whose company SUSI&James offers the industry "digital employees", talks about the limits of AI, its sensible use – and his own fears.

ALUMINIUM: Mr. Moßler, 'artificial intelligence' is a term that is used quite differently. How then do you define AI?

Jonas Moßler: You can of course define AI as the automation of human behavior, as the simulation of cognitive human abilities or of neural networks. But I wouldn't link the topic exclusively to humans, because while humans are a relatively complex entity, you can also negotiate intelligence at the level of a gnat. So basically, in order to assess where artificial intelligence begins, you first have to think about: What actually is intelligence? And what does 'artificial' mean? After all, we ourselves are part of nature. So when we create something – is it artificial? You see: It quickly becomes philosophical, it quickly goes into depth.

That sounds almost a little apologetic. Isn't one of the problems of the discussion that it too seldom gets philosophical?

Moßler: I agree with you there. In fact, society should first think about this in a much more fundamental way. It starts with the terms, with our framework of thinking. We are dealing here with profound issues that have to do with our humanity.

Don't we also have to ask about the limits of meaningful application, both technologically and philosophically? After all, it's also about the future of work itself.

Moßler: If I answer this question soberly from our business model, it's about repetitive activities and processes that are manageable in their complexity. These can be taught to a machine, at the various sensory, motor and cognitive levels. And that's no small thing, because after all, it's already part of our unconscious world. As a human being, I myself am a small AI, I myself am a system that has been trained and trains itself. The human being is incredibly exciting for me, but at the same time, with his physicality and his shape, he is also a vehicle and a highly developed biochemical robot. All our activities – brushing our teeth, riding a bicycle, assembling a piece of furniture – are vulnerable in this sense.

But where is the limit?

Moßler: The boundary is where we see ourselves as human beings. And we don't deal with that at all anymore: What is our USP? That's a fundamental discussion, but it's not being held, because it would also bring about a bit of a leap in culture and consciousness if we were to recognize that we can automate a lot, but that somehow people are still valuable. But perhaps not for things that machines can also do.

Maybe we still want to do these things ourselves?

Moßler: Absolutely right, it can also be sensual to make a chair yourself, to drive a car yourself. But as I said, from a purely technical and mathematical point of view, many processes today can actually be rationalized, described and trained. And to a high degree. This goes as far as the work of doctors or lawyers. In a conversation like the one we are having right now, it becomes more difficult.

But what remains of the USP of the human being?

Moßler: That's exactly what I'd like to have a social discussion about: What is actually exciting about humans? And why does it perhaps also make sense to let go of things to which we don't have to bind ourselves exclusively, because machines can also take over? How useful is that for us as humans? And not in the sense of business models.

Artificial intelligence also evokes fears. How do you counter them?

Moßler: By sharing them. I think it's presumptuous and dangerous not to share them. The applications may still be a bit terse today, but we all sense that something bigger is happening here right now. Of course, when you see robots doing flips, the image quickly arises of what would probably happen if they wore weapons. We should be afraid.

And what do we do with this fear?

Moßler: In my view, the culture of talking about feelings is being lost a bit right now. As a society, we are learning to blindly accept certain things: Silicon Valley is doing it, it's cool! We learn to follow suit, but we're not sheep! We have to actively bring in our fear, because it is justified. The military example is, of course, an extreme one – it is much more obvious, for example, that it is becoming increasingly possible to digitally eliminate people. In China, people are already being digitally scored today – that's no fantasy.

But the solution will probably not be to put the genie back in the bottle. Is it about more people with an ethical mindset taking matters into their own hands?

Moßler: Yes, but it's about more than that. I firmly believe that this discussion of values must also take place in the technical implementation of AI. Indirectly, the goal is to create sovereign artificial intelligences. That's more of a hidden track at the moment; we don't really know how we're going to get there yet. But at some point we will get there, and we need to think about what will happen then. AIs need a core that we can really identify with. We need to set the framework, or it's going to get ugly.

Your company is designing 'digital employees'. Somehow, that's a peculiar term.

Moßler: When we started using it a few years ago, there was actually resistance to the term. But in the meantime, we feel very comfortable with it, because it's honest: We talk about it when AIs can communicate, with natural language, via different channels. And when they learn to operate programs and implement processes independently in the background. For us, this combination is a digital employee.

Who is used in which fields?

Moßler: We are always moving into applications where humans can no longer perform, but the customer has precisely this expectation. If you call a car dealership today, you'll find out: Unfortunately, they're not very responsive. That may be down to the individual in each case, but it's probably usually just too much. They have other things to do, it's an overload. To fill those gaps, to eliminate those overloads: That's our approach. We can relax issues.

For example?

Moßler: Think, for example, of complex environments, of corporations! There are 500 SAP modules running there, and one is uglier than the other, slow, vulnerable. IT is arbitrarily ugly. It's exciting to say: I don't want to deal with this ugly IT landscape, this clicking around, I have better things to do as a human being! None of this hurts SUSI.

You are aware that this also sounds like rationalization?

Moßler: No one was laid off anywhere because of SUSI, we just took out stress. And released human energy. I'm not saying that rationalization won't be the result at some point. But that's a development I'm actually less afraid of. It just has to happen at the right speed, people have to be picked up. And of course we need a social discussion about it.

Does artificial intelligence also have to be perceived as such? Or is your goal the perfect simulation?

Moßler: That's exactly what we've been thinking about in depth. I am convinced: If AI remains unrecognized, that's dishonest. I want it to be authentic, so we also always make it transparent when it's a digital employee: 'You're talking to an AI – it's not as powerful as the human colleagues, but unfortunately they're busy right now.' That's a humble attitude. The AI becomes an apprentice, and you also know you're talking to one now. Just with a digital apprentice.

The interview was conducted by Bernhard Fragner.

Experience Jonas Moßler at the

ALUMINIUM Business Summit 2021
September 28th­–29th, 2021
Altes Stahlwerk Düsseldorf

Further Informationen and Program