Will Artificial Intelligence fail to catch on because it costs too much?
Artificial Intelligence (AI) has already changed our world and will no doubt continue to do so. But if you take a closer look at individual projects, especially in what is called Industry 4.0, you will see that the costs of development are quite high. The automotive industry, for example, has abandoned the idea of a completely self-driven car after initial euphoria, for the same reason.
Could AI be abandoned due to the high cost?
In the 1950s, people believed the future would bring rocket- or nuclear-powered flying cars, which would have been technically feasible. But the development effort was too small for the advantage they would have over good old internal combustion cars – at that time. Today, people are thinking again about air taxis: nuclear-powered cars would at least not damage the climate any further.
Though the development costs for AI are indeed enormous, there are other hurdles. There are already many highly developed AI systems, but implementing them would also involve high costs. Like any new system, it has to be adapted and managed. This of course requires resources. Thus, implementation may not fail due to cost alone, but because it cannot take place due to lack of capacity.
For the development and implementation of AI, specialist IT developers are needed. Unfortunately, they do not yet exist in sufficient numbers to provide the necessary capacities, even with retraining and immigration in the medium term.
Structures on the entrepreneurial side are also not well-suited to promote the introduction of AI on a broad front. Digitization is often grafted on top of it, as if this would then somehow take root in the company from the top downwards. It would be better if digitization and thus also AI were an integral part of the company and above all its products and services.
A critical look at the much-vaunted start-ups is also sobering. About 80 percent fail within three years. Of course it’s good to have the courage to make mistakes and fail, but this still involves a great deal of sunk costs. Mathematician Gunter Dueck calls it the age of over-innovation:
That means we make the world a better place very quickly, but we burn a lot of money unnecessarily…
AI is a broad field, but whether and where it can be usefully applied is not something that a company can easily assess. If you want to be part of it, you just have to do it. So far so good, but meanwhile, there is a great danger of getting bogged down. Many projects are working on solutions for identical or similar problems and, in the process, cannibalizing each other.
Even with the hands-on mentality, a well thought-out strategy and concentration on the essential goals are required. What do we actually want to achieve with this specific AI application? Is this really the right direction, or are we just assuming that it will somehow work out?
Billions in state funding go down the drain for unsuccessful research projects, which at best only serve as medium-term structural aid. It’s similar to what Henry Ford said about advertising:
Half the money I spend on advertising is waste, and the problem is I do not know which half.
In the case of start-ups and research projects, not only don’t you know which half it is, you lose a lot more than half.
Even with all the good will and combined forces of the AI proponents, there is a broad front of rejection against them, as with all major economic projects. Europe’s desperate attempts to keep up with AI are counteracted by a general, and particularly German, rejection of innovation. Here, data protection is often the lever that throws the threat of digitization off course. Like fire protection, data protection also costs money, and if you overdo it, it costs a lot of money.
To counteract any misconceptions here straight away: I am neither in favor of nuclear-powered cars nor against data protection, and I have nothing against start-ups. The costs of innovation, by which I do not just mean monetary costs, are inherent and therefore often unavoidable; however, they could be put to better use if we had the right focus and a well thought-out strategy. The potential of AI ranges from intelligent office solutions to complex production processes of the Smart Factory throughout the entire value chain. If properly harnessed, these potentials far outweigh the cost of AI and would make our world a much better place. Let’s not let digitization and artificial intelligence run out of steam halfway through just because the initial costs are too high – that would be too bad.