Updating Foresight Approaches and Methods

Participants of International Symposium "Foresight in a Rapidly Changing World" at the XXVI April International Academic Conference Named After Evgeny Yasin attempted to find answers to questions such as how to process large amounts of unstructured data, how to integrate artificial intelligence into analytical processes, and how to understand when innovation is ready to enter the market. "The world is changing, and we need to change our Foresight methodology along with it," emphasized Olesia Maibakh (HSE ISSEK Foresight Center), moderator of session "Foresight: New Approaches and Methods." The four presentations complemented each other, forming a unified picture: from a specific instrumental solution to infrastructure, from infrastructure to governance, and finally, to how to determine the moment for specific actions.

From Data to Technological Intelligence: Policy-Oriented Foresight

Cesar Costa, Jackson Maia(Centre for Strategic Studies and Management in Science, Technology and Innovation, Brazil) presented a methodological data processing chain they elaborated for working with the results of national public consultations on the development of the Brazilian Science, Technology and Innovation Strategy (ENCTI 2024–2034). The researchers faced large-scale task: processing and structuring 2,192 free-form proposals submitted by 858 representatives of academic community for national STI strategy. The authors developed a five-step process: data extraction; intelligent sorting using GPT-4.1-mini; semantic mapping through embeddings; clustering using the Louvain algorithm; and consensus editing, where AI suggests wording and people make the final decision. Ultimately, the original queries were transformed into 695 unique semantic clusters—without losing a single vote. Maintaining the method complete transparency was of fundamental importance to the authors: they deliberately chose TF-IDF and cosine distance instead of more powerful but less explanatory semantic models. "The methodology shouldn't be a black box; it should be mathematically defensible. The final decision always rests with human: AI structures data, and human controls the outcome," S. Costa emphasized.

Multi-agent AI systems in foresight and strategic analytics

Anna Aksenova, Valeria Lvova, Danila Kopeikin (HSE Centre for Strategic Analysis and Big Data) presented the results of iFORA team's work on integrating multi-agent AI systems into foresight, strategic analytics, and decision support. The speakers traced approach evolution: from the first text analysis algorithms in 2015—through creation of a multilingual database, which today contains more than 850 million documents—to a multi-agent platform where specialized AI agents work in parallel, and researcher receives structured PDF report with sources and analytical maps. The key principle remains unchanged: human sets goal at input, and human verifies the result at output. "AI cannot yet assume responsibility in foresight," the authors noted. "Success is determined by balance between human expertise and AI capabilities."

Anticipatory management as a new system of action

According to Ozcan Saritas (Rochester Institute of Technology Dubai, UAE), modern technologies are developing exponentially, while governance institutions are adapting linearly—and this gap continues to grow, creating a zone of unmanageable systemic risk. O. Saritas sees solution in transition from reactive to proactive management – ​​through implementation of anticipatory governance principles. This entails flexible regulatory systems creation where policy is considering as iterative process; transition to polycentric governance, where decisions are made in distributed manner rather than solely at central government level; and digital infrastructure development that allows state to respond to changes in real time. The key condition remains human-centricity: AI is a tool, not a subject of governance.In response to the moderator's question about key driver of transformation, O. Saritas emphasized the need to develop indigenous AI systems that reflect values ​​and cultural context of specific societies, rather than universal solutions developed elsewhere.

How to find the perfect moment to implement innovation?

Nikolay Khlopovand Olga Shaeva (Algorithm Trend Intelligence) concluded the session with presentation on combined the Gartner curve and Everett Rogers' innovation diffusion model for determining of innovation readiness. The Gartner Hype Cycle describes how technology expectations change; the Rogers Curve describes how it spreads among people. Separately, they capture current state. Combined, they explain dynamics and allow to identify three "windows of opportunity": Vision Gate (peak of expectations, moment for pioneers), Validation Gate (bridging the gap, forming an early majority) and Expansion Gate (entering the mass market). The speakers demonstrated history of General Magic, which developed smartphone concept in 1989 — 18 years before the first iPhone. The technology was right, but the timing was premature. Practical application of the model was demonstrated on Alfa Activity product, launched jointly with Alfa-Bank: understanding where market is on curve allowed to choose the perfect moment and right communication. “The perfect moment is as important as perfect product,” the authors concluded.

The session demonstrated that methodological renewal of Foresight is not an abstract task. It occurs regularly, in specific projects and tools: methodological chain for processing public consultation data, use of multi-agent platforms for strategic analytics and decision-making, concepts of anticipatory governance and innovation readiness models.

PRESENTATIONS

Cesar Costa, Jackson Maia (Centre for Strategic Studies and Management in Science, Technology and Innovation, Brazil)

From Data Science to Technological Intelligence: Reframing Evidence-Based and Policy-Oriented Foresight in Brazil (PDF, 2.54 Мб) 

Anna Aksenova, Valeria Lvova, Danila Kopeikin (HSE)

Multi-agent AI Systems: Applications for Foresight, Strategic Analytics, and Decision Making (PDF, 2.75 Мб) 

Ozcan Saritas (Rochester Institute of Technology Dubai, UAE)

Towards the AI State: Anticipatory Governance as the New Operating System for Global Growth (PDF, 1.01 Мб) 

Nikolay Khlopov, Olga Shaeva (Algorithm Trend Intelligence)

The Correlation between Hype Cycles and Innovation Diffusion Theory: Toward a Combined Model of Innovation Readiness (PDF, 23.03 Мб)