I always struggled with the question of how one man can see so easily a connection,

• between Quantum Gravity and Psychology or

• between Planck’s law of radiation and the Dunning-Kruger- or mount-stupid-effect

• between the quantum well problem and perfect advertisements

• between the last theorem of Fermat and the structure of the Quantum Einstein Field Equations

• between the co-evolution of man and machine and the entanglement of dimensions in Riemann spaces

• between Hawking radiation and the impossibility of imprisoning thoughts and ideas

• between the ignored degrees of freedom in the Einstein-Hilbert action and the realm of the dead and the unborn

• between Darwin’s laws and the evolution of spirituality

• between love and quantum fields

• …

When asking Norbert how he manages to spot certain connections others don’t, he gave me an interesting illustration of his way to think. He explained to me that he uses some kind of inner done which he sends high up into the sky in order to get an overview. This concept of the inner drone also helps him to avoid getting caught in deep holes he might have dug for himself “when focusing too hard on one particular problem thereby potentially letting the bigger picture become out of sight”.

The following abstract of a presentation Norbert once gave only for a small group of friends shall give us an example of how Norbert connects apparently strange dots in order to come to often surprisingly clear explanations and partially shockingly simple solutions:

Abstract: Metric Coating Optimization and Neural Network Entanglement – A Funny Similarity

Norbert Schwarzer, Tankow 2, 18569 Ummanz, Germany, SIO, www.worldformulaapps.com, www.siomec.de, n.schwarzer@siomec.de

Material optimization problems are – in principle – not so different from typical applications of neural networks like e.g. picture recognition. In both cases we usually have huge parameter fields, non-linear dependencies, often uncontrollable and fairly big uncertainties and – apparently – chaotic behavior regarding the effect of small parameter changes. Meaning, what in picture recognition becomes a “0” instead of a “6”, because of just one neuron misinterpretation, results in disastrous stability problems for a wrongly adjusted interlayer, which was supposed to create better adhesion, but acted as mode II or mode III defect instead.

The culprit for such disasters is quickly to be found: Linear optimization and tools of linear algebra. In the presentation it will be demonstrated how more general concepts applied in material science, especially surface or coating optimization can not only lead to better, which is to say more reliable and stable results, but also provide holistically enough uncertainty information for suitable life-time predictions.

Interestingly, with the similarity to neural network treatments, mentioned above, the question automatically arises whether such generalized concepts may also help to make Artificial Intelligence a bit more natural and a bit less artificial. Or formulated differently: Taking our results from the treatment of surface problems, could it be that linearity is a dead end with respect to the development of truly intelligent machines?

Thereby it appears as a funny coincidence, that it is thin film technology which originally brought the computer technique this far, that the question of artificial intelligence could be asked at all. Now one might ask what has AI to do with material science, especially thin film and coating technology. Well, as there are already complex problems where neuronal networks are applied for material and structural optimization, it may not be a bad thing to have the latter giving the first a small leg-up. This way, hopefully, in some future, material science could greatly profit from better AI concepts.

It still holds, after all, that a bit of artificial intelligence is better than natural stupidity.