Why We Aren’t Yet Threatened By The Rise Of The Machines
Some market players perceive Artificial Intelligence (AI) as either being capable of solving all of humanity’s problems, or destined to steal people’s jobs and take over the world. In fact, both scenarios are far from the truth.
That being said, a considerable gap is already beginning to emerge between the companies that implemented AI technologies early and the companies riding the “second wave.” According to Mckinsey projections, the companies that started to implement AI in 2017 can count on 122% profit growth by 2030, and those in the second group can count on just 10%. Many companies realize this. Between 2018 and 2019, the number of organizations that implemented digital technologies tripled, while not yet exceeding 14%, according to Gartner. This still leaves room for improvement.
The fact is a computer can adapt to a sophisticated production process in a relatively short period of time at a level comparable to that of a qualified operator. This is true for a whole range of tasks—a system that has been learning for just a few weeks at most, could give the same recommendations as specialists with many years of experience.
AI, the Internet of Things, robots, and big data technologies also don’t need to be paid a salary. They don’t take sick leave, and there’s no need to create an efficient incentive system for them. These technologies make it possible to solve a high volume of problems at wildly-varying levels of complexity.
With each passing year, changes are happening with increased acceleration, and less time is required for new technologies to penetrate.
However, it’s important to understand that the labor market is extremely interested not in saving jobs, but in retaining those experts who will occupy new job positions created via digitalization. For example, to operate an autonomous dump truck in a mining quarry, the driver will learn to control it remotely and will wind up doing smarter and safer work, which will turn out to be much less physically taxing.
Can AI learn while observing a human’s work by collecting data on how a production engineer makes decisions, depending on external factors? Yes, it can and it must. However, the decisions it makes in the process of work will be limited to its learned patterns. If AI encounters an unknown situation in production that it hasn’t seen before, it will freeze up and have to delegate control to a production engineer.
The capabilities of modern AI must be assessed sensibly, as they are limited. That’s why at the moment AI solutions are implemented in pinpoint fashion at individual production stages, processes, factories.
Here, we’re not talking about intellect, but only about its beginnings. Without a human being, AI can’t do anything yet—neither learn, nor navigate among variable factors. For many years, industry has been created by humans, for humans. So, AI will only be able to replace humans in the very distant future.
In the meantime, finding itself in an area of uncertainty, the AI program will need a human being to be authorized for a particular action. Safety should be at the forefront when using AI in production, and the potential for having a program make erroneous decisions must be shut off by a human being.
Research could well lead to the creation of AI capable of generalizing, incorporating experience and making the right, safe decisions in hazardous situations, but this isn’t likely to happen for some time.
