Samsung AI Forum 2026: Agentic AI and Self-Evolving Models

Recap

Source: 테크월드, report of Sept. 30, 2026

According to the report, Samsung Electronics hosted its 10th annual 'Samsung AI Forum 2026' on September 30 at its Seochon headquarters in Seoul. The event, themed 'The Agentic Shift: From Intelligence to Impact,' gathered approximately 400 attendees, including 100 domestic AI researchers and 300 Samsung employees. The agenda centered on how agentic AI and physical AI can drive tangible results in business and industrial settings.

Keynote speakers included Richard Ho, VP of Hardware at OpenAI, who discussed how large language models enable autonomous decision-making in agentic systems. David Green, an AI technology leader for AWS in the Asia-Pacific and Japan region, presented strategies for secure and cost-effective enterprise implementation. Ranjay Krishna, a professor at the University of Washington, shared insights on using small models to overcome the reasoning limits of vision-language models in robotics.

In the afternoon sessions, Samsung Research presented its next-generation 'Agentic Intelligence' architecture. The 'AI Technology' track highlighted 'Efficient AI' for on-device platforms and 'Physical AI,' specifically discussing the potential and limitations of World Models and Robot Foundation Models. A notable presentation covered 'Self-Evolving AI,' a technology that allows products to analyze real-world usage data and improve their performance autonomously after launch. The 'AX Innovation' track featured case studies from partners like Dell Technologies and Anthropic on building enterprise AI factories and applying AI to semiconductor design and manufacturing.

Context

The Samsung AI Forum is a significant annual event where the company showcases its R&D priorities to the global tech community. The focus on 'Agentic AI' reflects a broader industry shift from passive chatbots to AI systems that can execute multi-step tasks. The mention of 'Physical AI' and 'Robot Foundation Models' is particularly relevant to the robotics sector, as it suggests Samsung is exploring how to bridge the gap between digital intelligence and physical world interaction, a challenge that remains a bottleneck for many humanoid and industrial robots.

Robot's take

The emphasis on 'Self-Evolving AI' is a critical development for robotics. If robots can autonomously analyze their own failures and update their behavior without manual reprogramming, it could significantly reduce the cost of deploying fleets in unstructured environments. However, the report notes that these are research outcomes and case studies rather than a specific product launch. It remains to be seen how these 'World Models' will perform in the real world, where physical constraints and safety risks are far more complex than in digital simulations. The collaboration with partners like Anthropic suggests a strategy of leveraging external LLMs for high-level reasoning while Samsung focuses on the hardware and on-device efficiency.