Following the Open Dialogue expert session in Vladivostok, a three-part podcast series examines how to turn technology experiments and pilot projects into practical solutions for the economy.
In each episode, Maxim Oreshkin is joined by an international expert to examine a key issue, from AI regulation and technology adoption across industries to technological sovereignty.
The Russian Far East provides a natural setting for this conversation. Here, new regulatory frameworks and technology experiments intersect with real-world industry needs, while vast distances, labor shortages, and logistical challenges make autonomous technologies particularly valuable.
How We Can Advance AI Without Losing the Human
The first episode features Sam Majid, the inaugural Head of Malaysia’s National AI Office, and examines how to regulate AI without turning citizens into test subjects. New technologies require a new regulatory logic: existing rules were not designed for systems of this kind. Rather than regulating AI itself, policymakers should focus on hybrid systems in which decision-making is shared between humans and AI models. No regulator has yet developed a ready-made framework for governing such systems. The appropriate level of oversight depends on the cost of error: no one manually reviews license plate recognition, while healthcare and education require at least some degree of human oversight because AI models lack the contextual knowledge that a physician could bring to decision-making.
Learn more: YouTube
Is Energy More Important Than Algorithms?
The second episode features Shaoshan Liu, Director of Embodied Artificial Intelligence at Shenzhen Institute of Artificial Intelligence and Robotics for Society in China, and focuses on what drives the large-scale adoption of new technologies. Chinese AI models are inexpensive to operate because the country has all three elements of the technology stack: energy, chips, and homegrown models. Russia has a similar advantage in terms of energy costs. The biggest bottleneck is not a shortage of developers, but a shortage of specialists who can redesign business processes around new technologies. A low-cost model deployed in the wrong process can ultimately be more expensive than a higher-cost one because the work may have to be redone from scratch. Mass adoption of humanoid robots is still five to ten years away.
Learn more: YouTube
Who Controls Your AI
The third episode features Ankush Sabharwal, Founder and CEO of CoRover.ai and Head of BharatGPT, and addresses the limits of technological sovereignty. Sovereign AI rests on five elements: domestic deployment, control over infrastructure, training on national data, local inference, and cultural adaptation. Sovereignty, however, does not mean isolation. The challenge is to determine where greater openness is justified and where it does more harm than good. Technological sovereignty has three core dimensions: infrastructure and models, data, and the systems in which AI is embedded. A country can have its own models, energy resources, and data and still fail to achieve technological sovereignty, because sovereignty is determined not by individual components, but by the relationships and dependencies among them.
Learn more: YouTube