Artificial Intelligence as Inventor and Developer
Examples of cases in which human inventors and developers are encountering increasing competition from the world of ones and zeros are beginning to pile up.
- AI-Supported Materials Research: In the race to develop novel batteries and fuel cells, Toyota is betting on artificial intelligence. With a budget of around 33 million euro, AI is being deployed to identify and evaluate promising new materials. Studies have shown that this approach can work: from a total of over 200,000 types of magnets, it proved possible to identify the 14 most attractive candidates with the aid of computer simulations. In a word, the question as to “what to produce with” will not be decided by material scientists alone going forward.
- Automated Procurement: In the factory of the future, the decision as to how much should be produced will be taken autonomously and not by the production manager. Otto’s top buyer – an artificial intelligence algorithm – can predict which products will be sold in the coming month with a hit rate of 90%.
- AI-Driven Design: Even answering the question of why something must be done just so will not be left solely in human hands in future. Artificial intelligence algorithms are being deployed as intelligent and creative designers, capable of producing varied sets of design proposals. This is already resulting in improved chassis structures today. Mirroring natural selection, an intelligent algorithm is able to produce billions of design options, then play them through and winnow out the failures using sensor data and simulations. This is done in line with specific targets and guidelines such as mathematical problems, game-theoretical decision-making criteria and trade-off decisions. Based on the underlying learning algorithm, the artificial intelligence system outputs the optimum design or else several design options each with different benefits.
A harmonic merger from all developments will be important for the factory 4.0, because machines need to be capable of more than simply working off blueprints. Gearing up for high-volume customer-specific production will require more than simply customising the colours of simple products. For customisation to offer real added value it must be possible to modify multiple product features and to manufacture more complex products. To preserve the structural integrity of the products in question, modular production cells will need to react to customer requirements in a flexible manner and to understand what they can achieve “with what”, and “why it must be done just so”. Every customer has their own quirks and wishes. Real added value is achieved when the production plant is capable of working towards an unrestricted target corridor whilst taking full account of but not being limited by the customer’s specific decision criteria and issues. In this way, innovative and bespoke product variants will indeed be created directly in the factories of the future – perhaps even before the customer has placed an order for them.
AI and sustainability: a new strategic area of activity
In the global debate about the risks to society posed by and the economic potential of artificial intelligence, two complementary lines of argument stand out in striving for specific approaches to a “sustainable” use of AI. Both topics are of relevance to all companies wishing to develop a long-term AI strategy. However, they are usually not differentiated sufficiently in practice. An attempt at clarification by Andreas Neef, Managing Partner of Z_punkt.