FROM PROF. DR. PETER GENTSCH
The potential of artificial intelligence is currently seen primarily in the automation of tasks and processes. The focus is on automation potential, cost savings and efficiency increases. AI is not trusted with innovations – creativity and ideas are still considered the privilege of us humans.
However, more recent approaches of AI show how AI can also produce process and product innovations. These include new production and manufacturing processes as well as the creation of new software products, images, music or entire films. In addition to the economic issues of such innovations created by AI, social and ethical questions also arise.
CREATIVITY, THE LAST PRIVILEGE OF MANKIND?
A lot of fuss is being made about AI. And probably rightly so. Because there is no technology that is as over- and underestimated as AI.
Usually AI refers to processes and how we can automate and optimize these processes. People think about how AI can be used to speed up processes – or they simply think about the fancy bots that we can talk to, of course, and that respond as intelligently as possible using “intelligent” speech processing.
Is creativity perhaps the last privilege of us humans, or is AI also becoming dominant in this domain? For the further development of our society and of technology as a whole, it is elementary to recognize whether AI can be used to create new things, where AI can replace humans, and where ultimately the limits of both – humans and machines – lie.
What drives us humans to develop innovation? Maybe you see some room for improvement, maybe it’s just an opportunity, maybe you feel a pain – and many geniuses had their great ideas in the morning in the shower. It is often emotional and motivational factors rather than purely rational factors that drive us humans to create new things. The question is what drives an AI system to be innovative. AI has no awareness – the feed and driver of AI systems is always data used to train the systems. Training data and analysis set-up are currently still being pre-set and thus determined by humans. However, AI systems are becoming increasingly autonomous and “learning to learn”. Despite the mechanistic and data-driven approach, AI thus increasingly gives us the illusion of creating true innovations.
AI IS CREATIVE AND BEAUTIFUL
This sentence was said by Lee Sedol, long-time world champion in the oldest board game called Go, after he was beaten by the AI System AlphaGo.
Go is a very complex game and requires considerable strategic thinking and – actually – intuition. Lee Sedol has previously spoken to the press and expressed considerable confidence that he could win the games against the machine – precisely because Go is so complex and he didn’t think that a machine could deliver the enormous power of an experienced Go player. But he lost the first game. He lost the second game. And he also lost the third game. After that, he apologized to the press for his failure. But he also said that this result is really interesting and has contributed a lot to our understanding of the topic of “innovation”. He pointed out that although Go has been played for thousands of years, no one had the idea to play like AlphaGo. So the computer suddenly became much more than an opponent: it became a source of inspiration.
Fig.: Impressions of the go game Lee Sedol against AlphaGo
By the fourth game it was clear to everyone who would win. But the winner was Lee Sedol – with the “Train of the Gods”, which had never been played before and had never been predicted by any AI before. Lee Sedol used the inspiration of AlphaGo to take the game of Go to a whole new level. In this understanding, AI challenges us to go the extra mile to achieve new achievements. The formula of super-creativity is therefore Man + AI as a perfect symbiosis.
WHAT INNOVATIONS CAN AI CREATE?
It must be a nightmare for a fashion designer if AI can create new designs and concepts for fashion. Fig. 2 shows first applications. For example, the design generated by AI can be the basis for the designer to revise it until it is ready for the market.
Figure: AI creates fashion
And while Game of Thrones fans were still waiting for the sixth book of the saga, one fan fed an AI with the previous five volumes to predict the sixth book. AI can thus produce new content by predicting the closest words and sentences based on words, taking into account the temporal context.
Another example is the AI-supported production of an issue of the British marketing magazine “The Drum”. A thousand copies of the issue were printed, in which AI selected images, adapted texts and designed the pages. The AI was fed with data from the winners of the Golden Lion at the Cannes Lions International Festival of Creativity. So it wasn’t just a matter of creating the magazine, but also of creating an artificial intelligence that would appeal to the lifestyle audience. The two examples show that AI applications are increasingly evolving from structured, repetitive tasks to creative and innovative applications.
These changes do not stop at advertising either, although this is an area that is initially thought to rely on very human creativity. But as early as 2018, Lexus had new commercials developed by an AI. So the storyline and all the visuals were created by an AI. For this, campaign material from car and luxury brands from the past 15 years was analysed, all of which have won creative awards at the “Cannes Lions”. In this way, the AI learned which contents were particularly well received, which fit a brand like Lexus and how they could ultimately be put together to form a film. And so AI is penetrating more and more into the creative fields.
AI is also moving into the arts. So there is the AI application “The Next Rembrandt”. Based on historical Rembrandt paintings, AI has learnt the artistic style of Rembrandt and can therefore also paint new sceneries in the well-known Rembrandt style. Fig. 3 shows various Rembrandt paintings. Would you have recognized the AI-generated picture? It is the third from the left.
Fig: The Next Rembrandt – how AI paints pictures
However, it can also be much more dynamic. Traditional personalization works by selecting what best suits the consumer from existing content. But it goes far beyond previous personalization when an AI creates completely individual content. Such an approach creates a very customer-centric, data-driven approach: some customers get the movie trailers with the fast cars – others the ones with the beautiful landscapes. If you think this through to the end, you get one-to-one productions of movies and the creation of completely new things. One viewer prefers Nike products: the actor wears Nike. Another viewer prefers Adidas products: the actor wears Adidas. Of course this is an extremely simplified example. The main advantage of AI is that it is not a purely rule-based approach, but that patterns are recognized and used – even those that were completely unknown before.
So this is a way to create new approaches based on data and AI that deliver exactly what the audience wants. So this is a kind of performance marketing: You predict what will generate sales and then produce content and innovations that are highly likely to be seen. Maybe completely automatically.
There are already the first real examples of this: For example, the storyline of the film Sunspring was developed by AI on the basis of a number of science fiction films. The film Zone Out goes one step further, in which not only the storyline but also the faces of the actors are projected onto characters from existing films with AI (face mapping).
Figure: AI generated films
EVALUATION OF THE INNOVATIVE STRENGTH OF AI
The Rembrandt example shows how new works can be developed by imitating or varying existing data. Ultimately, however, this is only innovation through imitation and not real creativity. Examples from the past are used to create a new variation. The AlphGo example shows how AI Prediction and Reinforcement Learning can be used to create innovative solutions.
Figure: Different types of innovation and creativity
Existing data is analysed and a prediction for new solutions is developed. However, there is another, much stronger innovation: radical innovation. For example, electric light was not created by the continuous improvement of candles. It is a radical creation, an invention of something completely new. This is currently impossible for AI. The good news for us humans is that mankind beats AI in terms of radical creativity and innovation. The flipside of this coin, however, is that only 3 percent of innovations are created through radical creativity – and thus 97 percent of innovations are those that AI can achieve.
INNOVATION THROUGH AI – A DESIRABLE SCENARIO?
Many consumers cannot imagine hanging a picture painted by AI on the wall or listening to music composed by AI. Interestingly, many empirical tests show that consumers are no longer able to tell the difference. We may have to get used to this new, artificial way of production.
In many areas of daily life we already do this consciously or unconsciously. For example in press releases, sports reports and newspaper reports about earthquakes. The normal reader can no longer distinguish between a contribution written by AI and that of an editor.
Certainly, a distinction must be made here between different goods. In the case of factual content or commodity products, the acceptance is certainly higher. Whenever brand, emotions and empathy are involved, we want to know and appreciate the company and the people behind the work.
In addition, the question of the target function of AI is crucial. Does innovation and creativity follow an economic efficiency-rational or rather an emotional-emphatic culture-rational. In the case of the self-propelled car, for example, the goal seems clear at first. I want to get from A to B, and the system should be able to find out how this works best – in other words, how it is safest and most economical. But what are the real “fitness functions” of innovation and creation? Is it authenticity, environmental protection, sustainability or shareholder value?
If we simply follow an economic logic, we will certainly end up with a self-propelled car – but also with a “self-propelled” company and a “self-propelled” society. Everything would automatically be optimized for economic sense and cost efficiency. If we put everything in AI’s hands, we will end up in a rational economy based on economic principles – a world in which innovation and creativity will probably also be dominated by AI. Probably nobody wants to live in such a cold, mechanistic and artificial world. So the question arises how we can achieve the best social result by the right balance and combination of man and machine.
AI = AUGMENTED INTELLIGENCE = SUPER-CREATIVITY
After the major technological evolutionary steps Internet, Mobile and Internet of Things Big, Big Data and AI are now taking the biggest evolutionary step to date. If the industrial revolution has enabled us to get rid of the limitations of physical labor, these innovations enable us to overcome intellectual and creative limitations. We are thus in one of the most exciting phases of humanity, in which digital innovations are fundamentally changing the economy and society.
In the early phases of the industrial revolutions, technological innovations replaced or leveraged human muscle power. In the AI era, our intellectual powers are now being simulated, multiplied and, in some cases, substituted by digitalization and artificial intelligence. This creates completely new scaling and multiplication effects for companies and economies.
However, in order to ensure that we are not just spectators in this new world and no longer have control over creativity and innovation, we must understand AI in terms of “augmented intelligence”. Only then can we use AI as a creativity and innovation enhancer. The result: a super-creativity of man and machine