There is a lack of AI and ML experts

A second important field of application are production environments. This is where AI and machine learning solutions come into play in quality assurance in production as well as in research and development (54 percent each) and logistics (53 percent).

Users want to understand algorithms 

Nevertheless, there is no unconditional trust in AI and machine learning algorithms. For more than a third of those fathomed, an important point when choosing machine learning software is that its results are understandable for the user. This item ranks second in the ranking of the most important selection criteria, after costs. 

But this traceability is likely to reach its limits in the foreseeable future, especially with AI applications. Because experts admit that, for example, the decisions made by neural networks are no longer comprehensible to people, at least not in their entirety. Thus, in the medium term, a new type of trust is required between the users of AI or ML applications and the underlying algorithms.

There is a lack of AI and ML experts

In practice, however, users of AI and ML currently have less to contend with such "philosophical" challenges than with more trivial problems. As in 2020, one of the biggest is the lack of specialists : around 37 percent of companies said that on thelabour marketthere are not enough AI and machine learning specialists to be found. As a result, projects are delayed or cannot be tackled at all.

Above all, large companies with more than 10,000 employees (43 percent), which are otherwise an interesting alternative to smaller companies, especially for young professionals and "high potentials", complain about a shortage of skilled workers. This is probably unpaid to the fact that larger companies carry out a greater number of AI and ML projects than smaller companies. Therefore, the need for specialists is higher. Techcrunchblog

One option to compensate for the lack of specialist knowledge is to impart knowledge of AI and machine learning to your own employees . The study showed that there is a considerable need for further training , especially in the IT departments (50 percent) . Almost a quarter of those surveyed see know-how deficits in all areas of the company. An internal training andfurther educationcan fill such gaps. However, this presupposes that the companies set up appropriate programs and give their employees the time to take part in such measures.

 Too few specialists and a lack of know-how are currently the biggest problem areas in the implementation of ML projects.

Take the human factor into account

According to the study, it is also important that further training is not limited to imparting specialist knowledge. It should become clear to employees that AI and machine learning are not "job killers", but can strengthen their employer's competitiveness. Obviously, many employees are not aware of this. A third of the executives surveyed stated that the lack of acceptance of AI and ML by employees posed a major challenge.

That means, putting such solutions "in front of the nose" of the specialist departments is counterproductive. Such an approach will in all probability lead to projects not delivering the desired benefits. Rather, a holistic approach is required. It should be tailored to both the technology and the people who use it. In short: the corporate culture must also be put to the test. And that should be at least as big a challenge for a number of companies as the technical aspects of AI and machine learning solutions.