Polish enterprises will have problems using the potential of AI

Artificial intelligence is one of the most important trends in business today. Both global giants and smaller companies declare investments in this area. Today, the Polish economy is at a transitional stage – between experimentation and its real, strategic use.
— Rapid automation is necessary to offset the negative effects of the declining number of working-age workers and rising business costs. The pace of AI implementation determines whether we will maintain our current standard of living and we will remain an important economy – comments Łukasz Radosz, managing director of Euro-Funding Polska.
Without people there is no innovation
Data from the Polish Agency for Enterprise Development and the Jagiellonian University show that Today, only 23% use AI. companies in Poland. As much as 77 percent enterprises have not implemented this technology, and 64 percent lacks implementation capacity or remains undecided.
It's not about lack of tools. As experts emphasize, AI technologies are widely available today and increasingly affordable. The problem lies in the ability to implement them.
The competence gap is larger than expected
The most pressing problem is the lack of people who can combine technology with business. Companies are looking not only for AI specialists, but also for data analysts and people able to translate technological capabilities into specific operational processes.
Staff shortages are especially stark among companies already using AI — almost half of them report a competence deficit. Experts emphasize that organizations started building their own competences and developing teams capable of implementing new technologies too late.
Read also: Working with AI is a new standard in Poland. Those who don't understand this will be left behind
Most companies still test solutions point-by-point – without a coherent strategy and without integration with key business processes. As a result, artificial intelligence remains only a tool supporting individual actions. Meanwhile, this is it organizations that use AI systematically achieve the highest return on investment.
Architects of autonomy – a competence unknown to the market
— The PARP report only touches the tip of the iceberg. In my opinion, the problem is not that there are not enough people to support ChatGPT – because there are more and more of them. The real gap that we at Comfy AI Studio feel with almost every large project is lack of “architects of autonomy”. People who understand that AI is not just a clever assistant for writing emails, but a full-fledged employee who can take over the entire process from A to Z – notes Łukasz Zmywaczyk, president of Comfy AI Studio.
Read also: It was supposed to be a sure job, but it was a disappointment. What about El Dorado in IT?
What specific competencies are missing? — First of all, what I call “technological instinct”. When we implemented a video analytics system at the last major winter sports event in Italy, we were working on AWS Agent Core at a time when Amazon had not yet finished writing documentation for it. There is a lack of experienced engineers on the market who can find their way in such technological guerrilla warfare. People who can build the solution themselves, relying on intuition, experience, raw code and wise AI support – emphasizes the expert.
As he adds, another competency that is in short supply is: process thinking. — It is difficult to find specialists who can build the so-called superagents. This requires business imagination, not just the ability to type in a prompt, he explains.
Without data there is no AI. Polish companies also have a problem here
A separate, but equally important problem hindering the development of technology is data quality. As follows from Cloudera researchonly 26 percent organization has data fully organized and properly supervised. The remaining companies struggle with significant technical barriers: their data is scattered, not integrated between systemsand theirs visibility is limited.
This is a serious strategic limitation. As Sergio Gago, chief technology officer at Cloudera, emphasizes, Artificial intelligence is only as effective as the data that powers it. Low data quality also affects project finances: 18 percent companies indicated it as the main reason for low return on investment.
Michał Żelichowski from PSI Polska draws attention to the same problem in the context of the Polish market in a commentary on the report of PARP and the Jagiellonian University. In his opinion, the lack of data management policies means that algorithms do not have a solid foundation to work on. This not only reduces their effectiveness, but above all generates high risk of making wrong decisions business.
Most companies are just starting out
PARP and UJ experts divide Polish companies into three groups in terms of advancement in AI implementation. The best performers are the so-called Pioneers – only 6%. marketbut almost all these companies have access to qualified specialists.
In a group Digital Practitioners (12 percent of companies) 66 percent does not report any staffing problems. The situation is completely different among… traditionalists, who constitute as much as 46 percent Polish business. Only 2 percent of them have competences in the field of AI.
However, the most alarming data are the reactions of companies to the competence deficit. As much as 48 percent companies admit that they do nothing about it. Only 28% organize training. companies. Only a few reorganize their structures or count on independent training for their employees.
— This is a serious mistake, especially since tools are available to significantly reduce the costs of investing in AIsuch as R&D tax relief or training programs with co-financing of up to 80%. – reminds Łukasz Radosz from Euro-Funding Polska.
There are also less tangible factors standing in the way of wider use of AI: organizational culture and approach to risk. Many companies are still dominated by caution, lack of willingness to experiment and employees' fear of change. Meanwhile, implementing AI requires a completely different approach – iterative action, learning from mistakes and the ability to manage change.




