Samsung Electronics Vice Chairman and CEO Han Jong-hee said at the Samsung AI Forum that it is becoming increasingly important to use AI more responsibly, and that Samsung Electronics will fulfill its responsibility to build a more efficient and sustainable AI ecosystem.

Samsung Electronics held the 'Samsung AI Forum 2024' for two days from the 4th to the 5th at the Suwon Convention Center, hosted by Samsung Electronics SAIT (Samsung Advanced Institute of Technology), and discussed the future of AI with global AI experts.

The Samsung AI Forum, now in its 8th year, is a technology exchange venue that invites world-renowned scholars and experts in the fields of artificial intelligence (AI) and computer engineering to share the latest research results and explore future research directions.

This forum will feature the following speakers: △ Yoshua Bengio, a world authority in the field of deep learning, a professor at the University of Montreal in Canada; △ Yann LeCun, Meta’s chief AI scientist and professor at New York University in the US; and △ Ian Horrocks, a world authority in the field of knowledge graphs. Horrocks), a professor at Oxford University in the UK, and other global AI experts gave keynote speeches.

“AI is changing our lives at an incredible pace, and as it becomes more powerful, it becomes increasingly important to consider how we can use AI more responsibly,” said Han Jong-hee, Vice Chairman and CEO of Samsung Electronics, in his opening remarks. “Samsung Electronics will fulfill its responsibility to build a more efficient and sustainable AI ecosystem.”

The first day of the forum was held under the theme of “Sustainable Innovation with AI and Semiconductors.”

At the forum, attendees shared research results on AI and CE (Computer Engineering) technologies that can strengthen next-generation semiconductor development and production capabilities and discussed future development directions.

Professor Yoshua Bengio of the University of Montreal in Canada emphasized building safer and more reliable AI systems in his keynote speech titled Bayesian Oracles for AI Safety.

Bayesian Oracles use advanced statistical models, such as the Bayesian methodology, to collect and analyze information and improve predictive power by updating existing knowledge based on new information. Accordingly, it can support effective decision-making in complex and highly volatile environments.

Professor Bengio introduced research results showing that AI's performance has surpassed human levels in various fields, along with the future risks of AI, such as large-scale labor market impact, hacking using AI, and the emergence of superintelligence, and stated that in order to ensure AI safety, not only policymakers but also the public must have an accurate understanding of the current status and future of AI.

Professor Bengio emphasized that in order to ensure the safety of AI, △ safe AI design must be done in advance to prevent AI from performing dangerous actions, △ AI's behavior and goals must be aligned with those of humans, and △ more coordination and cooperation is needed in AI competition between countries and companies.

Professor Bengio, who is considered a world-renowned AI expert and winner of the 2018 Turing Award, has been a regular participant in the Samsung AI Forum since the first one in 2017 and has been working as a Samsung AI Professor since 2020, conducting industry-academia cooperation with Samsung Electronics.

Professor Yann LeCun, who won the 2018 Turing Award together with Professor Bengio, explained the level and limitations of the current Large Language Model (LLM) in his keynote speech and emphasized that additional technological innovation is needed for machines to reach the level of human intelligence.

In the technology session, AMD CTO and Vice President Joseph Macri introduced AMD's AI solutions on the topic of AI everywhere, and explained the importance of collaboration with AI platforms and AMD's strengths.

Along with this, Samsung Electronics SAIT Master Young-Sang Choi had time to discuss AI technology trends and semiconductor AI directions with the speakers.

The Samsung AI Researcher of the Year Award, announced on this day, was given to five people, including Professor Shuran Song of Stanford University in the United States. An on-site lecture was also held.

Professor Suran Song focuses on research utilizing AI across robotics, and was highly praised for her contributions to the development of recognition and inference algorithms for 3D modeling.

The second day of the forum was held at Samsung Electronics’ Seoul R&D Campus under the theme of “Device AI for Our Daily Lives.”

In his welcoming speech, Vice President and Head of Global AI Center at Samsung Research, Kim Dae-hyun, said that changes in everyday device AI are accelerating further due to the development of generative AI technology, and that he hopes this forum will be a place to discuss and share new possibilities in the coming AI era.

Professor Ian Horrocks of Oxford University, UK, co-founder of Oxford Semantic Technologies (OST), gave a keynote speech on the topic of personalized AI service technology using knowledge graphs.

Last July, Samsung Electronics acquired OST, a company that possesses world-class knowledge graph source technology.

Professor Horrocks explained the importance of knowledge graphs and how they store and process data in a way similar to how humans remember and recall knowledge.

In addition, it attracted the attention of the attendees by introducing major use cases such as search and recommendation that well reflect the characteristics of the knowledge graph system and suggesting a plan to implement customized services based on a flexible data model and logical inference.

In his presentation on AI for science, Samsung Electronics SAIT AI Research Center Director (Vice President) Choi Chang-gyu said that AI and semiconductor technology are very important in the field of science as they not only improve our lives but also reduce development time and costs. In particular, AI is of great help in cases where experimental data is insufficient or material synthesis is difficult. He said it was possible.

Samsung Research's Master Lee Hae-jun explained that high-quality data, efficient architecture, and stable training techniques are necessary for efficient learning of large-scale language models, and that this can improve both cost and performance.

He also emphasized that the efficiency and performance of learning a new model can be improved by reusing previous models, and that these methods are important factors to consider when developing a language model.