Presentations by the main world researchers of AI and the last
Research activities
Samsung Electronics announced today the opening of the Samsung AI Forum 2025, from September 15 to 16. Now in its ninth year, the forum serves as a global place where the main academics and industry experts meet to share the latest breakthroughs of AI and explore future research guidelines.
“Samsung applies AI in our operations to develop fundamental technologies that make AI more intuitive and seamless,” said Young Hyun Jun, Vice-President and Chief Executive Officer of Samsung Electronics, in his opening remarks. “The Samsung AI forum of this year brings together leading experts from industry and university world to discuss how AI transforms society and industry, and sharing information in what we expect will be a significant exchange of ideas.”
This year’s forum presents main conferences of IA world renowned researchers; Including Yoshua Bengio, professor at the University of Montreal and a pioneer in in -depth learning; As well as Stefano Ermon, professor at the University of Stanford and co-founder of the Startup Inception, who led the development of the dissemination-based linguistic model (DLM).
First day: world researchers explore the future of AI semiconductors
The division of Samsung devices solutions (DS) welcomed the first day of the forum under the theme “AI vertical strategies and vision for the semiconductor industry”, near the Samsung semiconductor site in Yongin, Korea.
Professor Bengio pronounced a main address describing the large -scale risks of today’s AI models, including their ability to bypass human control and abusive potential. As a safeguard, he presented Scientist IA, a new model designed to help mitigate these concerns.
“Unlike the models constructed to imitate or please humans, Scientist IA focuses on the supply of truthful responses based on facts and verified data,” said Professor Bengio, highlighting the potential of the model to improve the security of the AI and accelerate scientific discovery. He also shared his latest research results on AI for the development of materials in discourse.
Amit Gupta, Siemens EDA’s main vice-president, has led a session entitled “the future of electronic design led by AI”. During his presentation, he highlighted the importance of integrating AI into the automation tools of electronic design (EDA) and noted that end -to -end systems covering the entire workflow will be essential to unlock the full potential of AI.
The technical sessions featuring Yong Ho Song, executive vice-president and chief of the AI Center of the DS division, Professor Seokhyung Kang of the University of Sciences and Technologies of Pohang (Postch) and Professor Il-Chul Moon of Korea Advanced Institute of Science and Technology (Kaist). Each speaker has shared recent research developments in AI applications for the design and manufacture of semiconductors, while offering perspectives on the evolution of the field.
“AI is already an essential tool in flea design and software development,” said EVP Song. “While the manufacture of semiconductors is complex, we expect AI to help meet the technical challenges that arise.”
The forum also recognized three winners of the Samsung AI researcher of the year: Professor Nicolas Papernot at the University of Toronto, Professor Rose Yu at the University of California San Diego and Professor Lerrel Pinto at New York University. All the winners have been invited to present their work and share information from their latest research.
Second day: focus on the agentic AI era and stimulate productivity
The second day of the forum will be held online by division of the device’s experience (DX) of Samsung and broadcast live Samsung’s developer’s YouTube Channel Under the theme “generative at agentic AI”.1
“Generative AI has already become an essential tool in daily life and industries,” said Paul (Kyungwhoon) Cheun, CTO from the DX division to Samsung Electronics and chief of Samsung Research. “While we are entering the era of agentic AI, Samsung will continue to focus on the development of AI technologies which offer users tangible advantages.”
The second day includes Joseph E. Gonzalez, professor at UC Berkeley and a leading researcher in Language models and AI agents; Subbarao Kambhampati, professor at Arizona State University and World Authority in the planning and decision -making of AI; And Stefano Ermon, professor at the University of Stanford and co-founder of Inception.
Professor Gonzalez will present his research on improving the agent capacities of systems based on the large language model (LLM). In particular, it will present the paradigm of calculating the sleep time, which allows agents to use inactive interaction intervals to reason, learn and plan.
Professor Kambhampati will share his research on major reasoning models (LRM) aimed at approaching the limits of LLM. He will emphasize that if the current models excels in the generation of texts, they remain limited in factual precision, planning and complex reasoning – highlighting key challenges such as guaranteeing response guarantees, allowing a contextual adaptive calculation and providing interpretations of the intermediate reasoning stages.
Professor Ermon will introduce the diffusion language model, which applies diffusion models – widely used for the generation of images, video and audio – in the linguistic field. This approach aims to overcome the limits of traditional methods of sequential text generation and to offer a more effective paradigm for language models.
During technical sessions, Samsung research representatives will present their latest developments, in particular:
- AI Camera AI technology for automatic color temperature adjustment
- Methods based on the distillation of knowledge for more effective training of LLMs and their applications
- AI technologies available to provide LLM to consumer electronics such as smartphones and televisions
- Automatic lining technology that generates off votes in the voice of the original speaker
- Deep diving technology that uses multi-agent systems to automatically analyze and generate various reports
- Document AI technologies that automatically convert the formats of various documents into structural data for SMMs and agent systems
- A Disk AI studio for developers who shorten the development cycle of generative AI models
1 Intelligent systems capable of making independent decisions and performing tasks by themselves.
