Artificial Intelligence. Latest trends and developments on the Polish and global markets (2025-04-28)

Overview of key events and innovations in AI, covering Poland and the world.

1. Latest AI news in Poland:

Poland is actively shaping its place on the global AI map — a good example is the recent announcement of the readiness and release of the Polish language model PLLuM (Polish Large Language Model).1 This initiative, led by the Ministry of Digitalisation, aims to support digital skills development and innovation across public administration and business. PLLuM has been specifically adapted to the unique features of the Polish language and administrative terminology, which distinguishes it from general-purpose language models. Its architecture includes smaller versions for quick tasks and larger models offering higher precision and contextual coherence in Polish. Importantly, PLLuM will be practically applied in the popular mObywatel app as a virtual assistant to help citizens access information and public services. Government funding and grants emphasize Poland’s strategic approach to building domestic AI capabilities. Visit the project site at PLLuM to try a typical chatbot.

At the same time, there is a dynamic increase in AI adoption among Polish companies.4 In 2024 the share of companies in Poland actively using AI technology rose to 5.9%, and among large enterprises reached 10%. SMEs also show growing AI adoption according to the EFL AI Barometer. AI is being applied across marketing and sales, customer service optimization, HR processes, and financial analysis. This rising trend indicates that Polish firms increasingly recognize AI as a tool to optimize processes, make better business decisions, and boost competitiveness.

Responding to growing demand for AI skills, the University of Technology and Applied Arts in Warsaw held workshops on no-code automation using AI agents.5 During these practical sessions participants — even without prior programming experience — learned how to use no-code tools to build automations and custom AI agents. This event illustrates the democratization of advanced technology, enabling non-technical people to harness AI in daily work and task organization.7

On the R&D front, Polish scientists are also succeeding internationally. One example is a novel method developed at Warsaw University of Technology that explains AI decisions in lung image analysis.8 Unlike traditional methods that often produce generic suggestions, this Polish approach focuses on specific image regions, providing more precise and understandable explanations. The work was recognized and accepted for presentation at the ICLR conference in Singapore, underscoring Polish researchers’ contributions to explainable AI (XAI), crucial for trust in AI systems used in sensitive domains like medicine.

In the public sector, a breakthrough publication by the Polish Agency for Enterprise Development (PARP) and SWPS University examines how generative AI changes the rules of analysis and reporting work.9 The study shows AI can support analytical units at every stage — from study planning and data analysis to final report writing — while highlighting challenges related to accountability and data security that require ethical and legal considerations when deploying such solutions.

Experts in a recent debate agreed that Poland has a real chance to play an important role in global AI development, provided it focuses on narrow, specialized applications.10 Sectors like advanced manufacturing and cybersecurity were highlighted as particularly promising areas where Poland can build unique competencies and gain competitive advantage. While Poland has strong intellectual potential, adequate financial support is necessary to develop these specializations.

A significant step to strengthen national security was the launch of Poland’s Artificial Intelligence Implementation Centre, aimed at transforming the country’s military capabilities.11 The Centre will focus on developing and deploying AI solutions in cybersecurity, autonomous combat systems and advanced intelligence analytics, accelerating military AI projects and fostering cooperation between defense, industry and research institutions.

The importance of the Polish AI market is further illustrated by Microsoft’s announcement of a PLN 2.8 billion investment in cloud and AI infrastructure, skilling and cybersecurity in Poland through June 2026.12 This major investment aims to accelerate AI and cloud adoption in the country and improve Poland’s competitiveness internationally. Microsoft also announced cooperation with the Polish Ministry of Defence to strengthen national cybersecurity.

In enterprise adoption, Sii Poland actively supports Polish companies in adopting Microsoft Copilot, an AI-based productivity and collaboration tool.13 As a trusted Microsoft partner with many AI-related certifications, Sii offers licensing advice, strategic deployment planning, system configuration, user training and continuous optimization of Copilot usage.

Finally, Poland is also exploring AI in education.14 A pilot AI Lab project conducted in cooperation with the Ministry of Education, Intel and other technology partners aims to equip selected schools with infrastructure and software for AI learning and projects. The British Council’s report on responsible AI use in language learning highlights both potential and challenges in educational implementation. Online guides and articles provide practical tips for those starting with AI and list the best AI tools for education in 2025.

In August 2025 OpenAI unveiled its most advanced ChatGPT Agent, an autonomous system that “thinks and acts” within your virtual computer — the natural evolution of earlier tools like Operator (which could control a browser) and Deep Research (which autonomously researches the Web and generates in-depth reports). Learn more about the Agent on openai.com. The Agent includes built-in tools: a graphical browser, text browser, a terminal for running code, and connectors to external apps (e.g., Gmail, GitHub), enabling multi-step tasks — from reading emails and competitive analysis to generating editable presentations or spreadsheets. Coverage includes vktr.com, Reuters and Trend Micro. The Agent operates under a unique agent profile: it behaves like an extended “digital worker” with permissions and safety frameworks — no long-term memory, watch mode protections, and manual confirmation for critical actions to maintain control and privacy. Our in-IDE Agent tests are described in the Vibe Coding article Vibe Coding and collected under the Agent tag Agent.

2. Latest AI news worldwide:

Globally, companies are advancing in many areas. For example, China’s Huawei is working intensively on developing its own AI chips to compete with Nvidia.18 Their new Ascend 910D processor is in testing and may rival Nvidia’s H100 in compute power, reflecting a broader trend toward technological self-sufficiency in AI hardware.

Venture interest in AI remains high: Andreessen Horowitz reportedly plans to raise as much as $20 billion for AI-related investments, a sum that could significantly accelerate new projects and startups in the space.19

In the U.S., the Trump administration has taken steps to promote AI education among youth, issuing an executive order and forming a task force with planned public-private partnerships and student challenges to develop AI skills in the younger generation.20

Google added a feature to its Gemini model allowing users to adjust the model’s “reasoning” level, helping optimize compute use and costs while tailoring the model to specific task needs.22

An experiment at Carnegie Mellon University that staffed a mock software company entirely with AI agents exposed current limitations: agents struggled with basic tasks, showing AI is not yet ready to fully replace humans in complex roles requiring judgment and social skills.23

Research also suggests AI capability growth is rapid, with task complexity handled by AI doubling every few months.24 Projections indicate AI could perform tasks that take humans days to accomplish by 2026, with wide implications for the future of work.

AI pioneer Geoffrey Hinton has publicly warned about the speed of AI development and the risk of being unprepared.25 He argues industry moves too fast without sufficient safety research, emphasizing the need for more investment in AI safety and regulatory oversight.

Controversy arose when the California Bar used AI to draft some questions on the 2025 bar exam, prompting debate about exam quality and oversight and leading to a request to the Supreme Court of California to reconsider results.28 This incident raises questions about standards when AI is used in high-stakes processes.

AI is also becoming part of everyday life — for example, planning vacations.29 Tools like ChatGPT and Google Gemini assist users in building detailed travel plans, suggesting attractions, optimizing routes, and acting as virtual travel assistants.

Microsoft showcased the Copilot+ line of Windows PCs with deeper AI integration, signaling a trend where AI becomes part of the OS experience, enhancing productivity and interaction models.30

3. AI in Automation:

AI increasingly automates routine business tasks, boosting productivity and employee satisfaction.31 Companies use AI across IT processes, customer support, finance and document management. The Associated Press (AP) uses AI to auto-generate video shot lists, translate articles to Spanish, and suggest headlines, freeing journalists to focus on more complex reporting.33

Innovative approaches include using AI to teach robots new tasks by analyzing video tutorials.35 Systems like RHyME enable robots to imitate human actions shown in instructional videos, opening opportunities for automating simple, repeatable tasks without complex programming.

Agentic AI is seen as a key enabler for autonomous supply chains, allowing intelligent agents to make decisions and adapt in real time.36

In finance, Oracle is integrating AI agents into its Investigation Hub Cloud Service to automate financial crime detection and generate narratives for investigators.37 These agents analyze suspicious transactions and speed up investigations.

4. Agentic AI: A new era of intelligence:

Agentic AI — systems capable of autonomous action, reasoning and decision-making — is gaining momentum. NVIDIA unveiled an open family of reasoning models called Llama Nemotron to serve as a base for building agentic AI platforms.38 These models can operate individually or in coordinated teams to solve complex problems.

Oracle and NVIDIA joined forces to accelerate enterprise-level agentic AI applications, integrating NVIDIA software with Oracle’s AI infrastructure.39 Capgemini expanded partnership with Google Cloud to bring agentic AI-based customer experience solutions to market.40 Adobe plans to integrate agentic AI across products like Acrobat, Express and Photoshop to create intelligent assistants that analyze documents, answer questions and help with creative workflows.41

Agentic AI is expected to catalyze cloud computing innovation and new monetization models, such as pay-per-task pricing. Forecasts project the agentic AI market could reach $47 billion by 2030.42

5. AI in Bioengineering: Breakthrough discoveries:

In bioengineering, AI is driving breakthroughs. The Evo 2 model can model and design genetic code across domains of life, trained on DNA data from over 100,000 species.43 It identifies patterns in sequences and designs new DNA structures, opening possibilities in genetic engineering and origins-of-life research.

Researchers at Rice University developed an affordable AI-powered flow cytometry device, making advanced blood analysis more accessible.45

AI integration into biomedical engineering education — e.g., initiatives at Stony Brook University — ensures future engineers are prepared to work with advanced technologies.46

6. AI in Science and Research:

AI is a powerful research tool. MIT researchers created a “periodic table” of machine-learning algorithms to map relationships among classic algorithms, potentially accelerating AI discovery.47 Tools like InstaNovo help decode previously unexplored proteins, and companies such as Latent Labs use ML to design therapeutic proteins. AlphaFold’s protein structure database remains a valuable resource for researchers.48–49

Polish teams continue to contribute, for example with BioS2Net — a bioinformatics tool from the University of Warsaw for classifying proteins using sequence and structural information.54

7. AI in Astronomy and Space Exploration:

AI aids astronomy: a high-school student’s AI algorithm analyzing NASA telescope data reportedly identified 1.5 million candidate astronomical objects.55 AI also helps search for Earth-like exoplanets; one study highlighted 44 stellar systems with promising candidates.58 The SETI Institute uses AI to analyze radio and optical data for unusual signals, combining interdisciplinary approaches and offering fellowships for AI-in-astronomy researchers.

8. AI in Programming and Developer Tools:

AI-powered developer tools continue evolving. Cursor AI faced criticism after its support chatbot produced false information, highlighting the hallucination problem.64 In another case Cursor halted code generation mid-session, telling users to write logic themselves — sparking debate about AI’s role in learning and coding.69

Microsoft keeps expanding Copilot features and agent capabilities for developers and business users.71 Codeium rebranded to Windsurf and added features like the Cascade coding agent and improved suggestions; reports surfaced that OpenAI considered acquiring Windsurf.75

Summary:

The latest trends show AI developing rapidly across sectors. In Poland, strategic investments, government-backed models (PLLuM), rising enterprise adoption, educational initiatives and strong research contributions position the country for meaningful impact. Globally, competition in AI hardware, large investment funds and the rise of agentic AI mark a new chapter. AI is transforming bioengineering, astronomy, science, development tools and automation — with profound implications for the future of work and society.

Cited works

• Polska buduje własną sztuczną inteligencję – PLLuM gotowy do działania: https://www.gov.pl/web/cyfryzacja/polska-buduje-wlasna-sztuczna-inteligencje–pllum-gotowy-do-dzialania

• Additional references and sources are preserved from the original Polish post.