Poland Launches Bielik, Its Own AI Language Model

On the second anniversary of ChatGPT's public release, Sebastian Kondracki announced the launch of Bielik, Poland's first large language model, similar to ChatGPT. In an interview with RMF24, Kondracki emphasized that the integration of artificial intelligence (AI) into daily life could enhance human capabilities rather than replace them.

Kondracki noted that the popularity of ChatGPT led to millions of users engaging with AI within days of its launch, sparking discussions on its applications in education and other fields.

He described AI as having 'frayed boundaries,' excelling in some tasks while struggling in others. For instance, while ChatGPT can generate creative short stories, it often produces weak poetry.

Looking to the future, Kondracki expressed uncertainty about AI's trajectory but highlighted ongoing advancements in computational power and data utilization. He pointed out that AI is evolving to process multiple formats, including sound and images.

Regarding regulatory measures, he acknowledged the potential impact of the EU's AI Act on development pace but stressed the need for balanced regulations that would allow for data sharing to enhance AI training.

Kondracki prefers the term 'collaborative intelligence' over 'artificial intelligence,' arguing that true power comes from combining human expertise with AI capabilities. He believes that while AI might transform certain jobs, it will ultimately enhance cognitive professions rather than replace them.

In Poland, Kondracki praised the country's strong foundation in AI development, noting the significant contributions of Polish scientists to global AI advancements. He highlighted the importance of computational power and investments in supercomputing facilities like Cyfronet to propel Polish AI forward.

Bielik, while compared to ChatGPT, is described as a compact model that can run on local infrastructure, making it suitable for specific tasks requiring Polish language nuances.

Bạn có phát hiện lỗi hoặc sai sót không?

Chúng tôi sẽ xem xét ý kiến của bạn càng sớm càng tốt.