While the hype surrounding artificial intelligence remains at an all-time high, the reality on the ground is more nuanced. Here is the translated and polished article detailing the current state of AI adoption in business.
AI in Business: Great Expectations, Modest Reality
The use of Artificial Intelligence (AI) in business is rising, yet the pace of real-world implementation is lagging significantly behind market expectations. Recent data reveals that few companies have fully integrated AI, with managers increasingly citing regulatory hurdles, high costs, and a lack of reproducible business results as major barriers. In practice, the technology is largely confined to isolated testing environments, while key decision-making remains firmly in human hands. Even in firms that have adopted AI, disappointment is a common byproduct.
“To be honest, AI has very few actual achievements today,” says Professor Wojciech Czakon, Head of the Department of Strategic Management at Jagiellonian University, in an interview with Newseria. “The fascination persists regardless, but a realization is slowly dawning on users and investors: the hopes pinned on this technology were perhaps not excessive, but certainly premature. This is an expensive, largely unproven technology clashing with boundaries set not necessarily by ethics, but by regulation.”
The Adoption Gap: By the Numbers
According to 2025 Eurostat data, 19.95% of EU enterprises with more than 10 employees used at least one AI technology, up from 13.48% the previous year. However, the scale of adoption is heavily skewed toward larger organizations:
- Small firms: 17.13%
- Medium firms: 30.36%
- Large enterprises: 55.03%
Sector-wise, AI usage is highest in knowledge-based industries. The Information and Communication sector leads with 62% adoption, followed by professional, scientific, and technical services at roughly 40%. In other industries, AI remains a rarity as companies wait for more predictable outcomes.
The Polish Context: A Wait-and-See Approach
In Poland, adoption lags behind the EU average. Eurostat reports that only 8.36% of Polish enterprises used AI in 2025. Statistics Poland (GUS) confirms this, noting that 8.7% of firms declared AI usage—primarily for text/voice generation (5.3%) and text analysis (2.1%).
A report by the Polish Economic Institute (PIE) titled “AI in Polish Enterprises” highlights a striking trend: 77% of non-users state they will not implement AI until it becomes absolutely necessary. The primary obstacles are high implementation costs and a lack of internal expertise. Furthermore, many AI initiatives are “bottom-up” rather than strategic, making them difficult to scale.
Productivity vs. Financial Impact
The MIT NANDA report, “The GenAI Divide: State of AI in Business 2025,” paints a stark picture of global investment. Despite billions poured into Generative AI, only 5% of pilots generate million-dollar value. Tools like ChatGPT and Copilot boost individual productivity but often fail to move the needle on the corporate bottom line.
“Managers look to AI for productivity—which, in plain terms, means cutting costs,” explains Prof. Czakon. “The promise of replacing 15 or 20 positions with an AI agent costing 120 PLN a month is a huge temptation. The question is whether we are dealing with an equivalent of value. Currently, the ratio of ‘unmet hopes’ to ‘positive effects’ stands at 9:1.”
The Trust and Responsibility Barrier
A significant hurdle is the lack of reliability in current models. While machine learning is advanced in production processes, language models are often deemed “unreliable” by academic standards.
“I’m not sure anyone would want to be in the middle of a storm at a major airport where an AI, rather than a human, manages the backup traffic and decides if you fly to Warsaw via Oslo or Istanbul,” Prof. Czakon remarks.
This lack of reliability—and the question of accountability—prevents AI from being used in sensitive areas like:
- Insurance claim settlements
- Medical procedures
- Employee promotions
- Approval of financial statements
The EU’s AI Act is a response to this, categorizing systems by risk levels and mandating human oversight for high-risk applications. While these regulations provide a legal framework, the potential for financial sanctions is leading many firms to take an extremely cautious, phased approach to adoption.
