Hailed as a lynchpin of digital transition, AI is beginning to establish itself in the Polish economy.
At the beginning of May 2022, the National Centre for Research and Development (NCBR) opened a competition for innovative applications of artificial intelligence (AI). The Centre is looking for projects involving a smart speech processing system for doctors, a mail dispatcher, an assistant for persons filing official applications, and a smart apple picking machine. The total budget is PLN 190 million.
The biggest budget of PLN 100 million has been allocated for the latter project, even if a fruit orchard is not the environment that first comes to mind in the context of AI applications. However, NCBR argues that “automating fruit picking is an answer of sorts to the looming problem” of a shortage of workers in agriculture in the near future. The machine will lend a robotic arm while AI-based on tailor-made software will separate good apples from bad ones.
This is expected to reduce harvesting costs to the advantage of fruit growers. According to NCBR, half of the 4.8 million tonnes of apples produced in Poland in 2018 came from modern orchards, most of them with an area of 15-20 ha each, focused on good quality dessert fruit (not for preserves). Although work on building apple harvesting robots has already begun in several locations around the world, it has not yet reached completion anywhere. The development of a Polish machine is also supported by the fact that orchards are structured in different ways in different countries. If AI proves itself in apple harvesting, the technology can be adapted to produce robots that harvest other fruits.
The speech processing system for doctors will streamline the process of producing medical records, writing prescriptions, and entering patient data. The application assistant, in turn, will operate like Siri or Google Assistant, except that it will additionally be familiar with the specificities of a municipal office, tax office, or other public institution in order to facilitate the administrative process. The smart mail dispatcher will use software installed on the computer or server of an institution or company. It will sort incoming emails by category of business and suggest replies based on the content of the letter and historical data.
“We are looking for solutions which can be directly applied in practice,” said Dr Remigiusz Kopoczek, acting Director of NCBR. “Supporting the digital transition of the economy is one of our priorities. Practical AI applications will enhance the competitiveness of our economy and improve the quality of various aspects of life,” said Dr Kopoczek.
AI in the warehouse
The current competition is the fourth under the strategic programme “Infostrateg” aiming to boost Poland’s AI potential. In 2020-2029, NCBR will spend up to PLN 840 million on projects covering scientific research and the implementation of AI. In previous competitions in the second half of 2021 and in 2022, several projects were selected using AI for medical image recognition, debunking fake news, and assuring product quality and safety, among others.
A grant of nearly PLN 3 million from the National Centre for Research and Development was awarded to the Warehouse Intelligence project developed by engineers from PSI Polska which brings an AI application to logistics. A pilot of one of the project stages has been implemented at LPP, owner of the fashion brands Reserved, Cropp, House, Mohito, and Sinsay.
There are almost 24 million m2 of warehouse and industrial space in Poland. The rising cost of fuel, materials, and labour is forcing manufacturing and retail companies to look for new options to optimise their operations. Warehouse automation can help.
“Technologies available on the market support employees very selectively, mainly by streamlining order picking routes. There is no AI-based technology that would optimise all warehouse processes in a holistic way,” said Jerzy Danisz of PSI Polska, referring to the origins of the project.
Warehouse AI has reached perfection by learning from its own mistakes. “An ML (machine learning) model was given a task and asked to find the best possible solution. If it managed to successfully optimise a process, it scored. Otherwise, it had to keep trying until it succeeded,” said Danisz. The learning takes place in a digital replica of a real warehouse (a digital twin) so as to quickly try out hundreds of thousands of warehouse operational scenarios at almost no cost.
It takes about three months to train the algorithm to operate at the target location.
The pilot implementation of Warehouse Intelligence in the LPP warehouse shortened the transport routes of goods by 30%. This generates savings as over 1/3 of warehouse logistics costs are attributable to order picking. Sebastian Sołtys, Logistics Director at LPP, said that the cost of picking and packing goods at LPP represents a “significant number”. “Therefore, optimisation in this area is crucial for us. It has a direct impact on warehouse performance and order handling efficiency.”
Specialists in full swing
If AI succeeds in Polish warehouses, this will be an important step in the country’s digital transition. For now, Poland’s economy is still in an early stage of the process. According to the “State of Polish AI 2021” report published by the Digital Poland Foundation, only 6% of large companies in Poland use machine learning, one of the lowest proportions in the European Union where the average is 11%, and typical of Central and Eastern Europe as other Member States in the region reach a similar score.
In contrast, Poland ranks seventh in the European Union in terms of the number of experts who develop or implement AI. Companies in the sector usually employ between five and 10 people. More than half of companies developing AI products and services generate seven-digit profits, and 8% generate more than PLN 50 million in revenue. Nearly 40% sell abroad, the main markets for Polish AI solutions being Europe and the United States.
Domestically, AI is finding its way into the most digitised sectors, such as telecommunications, banking and finance, retail, media, and insurance. This is partly due to the huge swaths of data generated in such businesses. AI is proving its worth when processing big data.
COVID-19 and its consequences have had a two-pronged impact on the implementation of AI in the Polish economy. According to Autodesk’s “Digitalisation in the Age of Pandemic” report published last year, about 15% of the companies which stopped implementing Industry 4.0 technology decided to halt AI development in particular. On the other hand, more than a quarter mentioned AI as a solution introduced in connection with a pandemic even in the absence of such plans before its outbreak. As a result, the percentage of companies that have implemented AI has risen from less than 20% in 2020 to almost 30% in 2021. Another 32% say they are planning to do so.
“The development of AI-based technology can be construed in terms of the seasons,” said Aleksandra Przegalińska-Skierkowska, Professor at the Department of Network Society Management at Kozminski University, speaking at this year’s EEC Trends debate. She also said that AI has suffered a harsh winter of insufficient funding. “Now it is spring, which I expect to be followed by a long sunny summer,” said Professor Przegalińska-Skierkowska, with the reservation that AI in Poland is still at the stage of research and early implementation. “A change in the approach to implementation is happening but there is not yet a degree of implementation that would saturate the modern economy with artificial intelligence.”