Samsung Electronics will host the Samsung AI Forum 2025 for two days starting on the 15th.
Now in its ninth year, the Samsung AI Forum is a venue for technical exchange where academic and industry experts gather each year to share the latest research achievements in artificial intelligence (AI) and explore future research directions.
Global AI experts will deliver keynotes at this forum, including Yoshua Bengio, a world-renowned scholar in deep learning and a professor at the University of Montreal in Canada, and Joseph Gonzalez, an authority on language model and AI agent research and a professor at UC Berkeley.
Vice Chairman and CEO Jun Young-hyun of Samsung Electronics said in his opening address, "Samsung Electronics is applying AI technology across diverse business areas and developing foundational technologies that enable easy and fast use of AI anytime, anywhere."
Day 1 of the forum, organized by the Device Solutions (DS) institutional sector, was held at The Universe at Samsung Electronics in Yongin, Gyeonggi Province, under the theme "Vertical AI strategies and vision for the semiconductor industry."
In his keynote, Bengio explained potential risks that can arise in existing AI models, such as "evasion of human control" and "malicious use," and introduced a new model, "scientist AI," designed to serve as a safeguard.
Amit Gupta, vice president at Siemens EDA, a semiconductor design automation company, said, "Integrating AI into electronic design automation tools for semiconductors is more important than anything else," adding, "In particular, to fully harness AI's potential, an end-to-end system that operates across the entire workflow is needed."
In a technical session led by Executive Vice President Song Yong-ho, head of the AI Center in the DS institutional sector at Samsung Electronics, Pohang University of Science and Technology Professor Kang Seok-hyung, and Korea Advanced Institute of Science and Technology (KAIST) Professor Moon Il-chul, participants discussed the latest AI application research achievements and future outlooks in semiconductor design and manufacturing.
Three scholars were selected for the Samsung AI Researcher Award: University of Toronto Professor Nicolas Papernot, UC San Diego Professor Rose Yu, and New York University Professor Lerrel Pinto.
The second day of the forum, held on the 16th, will be organized by the Samsung Electronics Device eXperience (DX) institutional sector and run online under the theme "Beyond generative AI, toward agentic AI."
President Jeon Kyung-hoon, chief technology officer (CTO) of the Samsung Electronics DX institutional sector and head of Samsung Research, said, "Generative AI has already become an essential tool in everyday life and across industries," adding, "Samsung Electronics will prepare AI technologies that provide practical help to users in line with the emerging era of agentic AI."
In the Day 2 keynote, Gonzalez will present case studies on advancing agent capabilities based on large language models (LLMs).
Subbarao Kambhampati, a professor at Arizona State University, will share research results on a "large reasoning model" to address limitations of existing large language models and will present key tasks that must be solved in language models, including guaranteeing accuracy, context-adaptive computation, and providing intermediate reasoning interpretations.
Stefano Ermon, co-founder of the startup Inception and a Stanford University professor, will unveil a "diffusion language model" (DLM) that applies diffusion models—previously used for image, video, and audio generation—to language. This technology is expected to overcome the limitations of sequential text generation and present a more efficient paradigm for language models.
In the technical session, Samsung Research researchers will share the latest research and development成果 on: ▲ AI technology for automatic camera color temperature adjustment ▲ efficient large language model training techniques and applications using knowledge distillation ▲ on-device technologies for loading large language models onto consumer electronics such as smartphones and TVs ▲ AI technology that automatically generates dubbing voices in real voices.