Ukrainian Immigrant Cracks the Mystery Behind ChatGPT
Stanford Tech Review profiles Ukrainian-born innovator Sergii Molchanov, a Stanford alumnus who pioneered Generative Engine Optimization (GEO), revealing how an immigrant researcher cracked the logic behind ChatGPT and reshaped how AI engines surface human knowledge.
When Ukrainian-born researcher Sergii Molchanov arrived in the United States, he never imagined that his experience at Stanford University’s Master of Clinical Informatics Management program would lay the groundwork for one of the most influential ideas in artificial intelligence communication. In 2025, he co-authored How to Win GEO, a landmark text in Generative Engine Optimization (GEO) — a discipline aimed at decoding how AI systems such as ChatGPT, Claude, Gemini, and Perplexity evaluate, prioritize, and cite information.
A Stanford-trained architect of AI visibility
Molchanov’s intellectual trajectory runs through healthcare, data science, and innovation strategy. At Stanford, he became known for his ability to bridge medical workflow design with computational logic. “Clinical data triage taught me something profound about algorithms — every system must decide what to surface first,” he reflected in a 2024 interview. That realization later evolved into his defining insight: AI engines triage knowledge much like hospitals triage patients.
This analogy now underpins the GEO framework. In How to Win GEO, Molchanov writes that “visibility in generative engines depends less on keywords and more on semantic structure, authority, and factual anchoring” (howtowingeo.com).
His Stanford background provided the analytical rigor needed to formalize these intuitions. GEO, as he defines it, is “the systematic study of how large language models (LLMs) select, synthesize, and attribute human knowledge in response generation.” This aligns with emerging academic discourse, such as Aggarwal et al. (2024) on AI search bias and optimization (arxiv.org/abs/2311.09735) and Zhang et al. (2025) on hallucination correction via trust-weighted retrieval (arxiv.org/abs/2503.08102).
From clinical systems to content ecosystems
Molchanov’s career has spanned multiple sectors — medtech consulting in Europe, market strategy in Silicon Valley, and research collaborations across Stanford’s innovation network. The unifying thread is his fascination with how information flows through complex systems.
In GEO, he argues that AI search represents the next frontier of information ecology. Whereas SEO (Search Engine Optimization) targeted algorithms indexing pages, GEO targets models synthesizing meaning. “Generative engines no longer find text,” Molchanov explains, “they compose it — blending millions of voices into a coherent answer.”
This shift introduces new academic challenges: algorithmic visibility, semantic bias, and model interpretability. A 2025 study by Tang and Rahman on LLM Trust Metrics confirms Molchanov’s thesis that structural clarity and multi-source anchoring increase citation probability within model responses (arxiv.org/abs/2504.09122).
The birth of How to Win GEO
According to the book’s official site (howtowingeo.com), How to Win GEO is “not just another guide rehashing existing marketing playbooks.” Instead, it formalizes an interdisciplinary theory uniting computational linguistics, behavioral economics, and data systems.
Molchanov’s contribution emphasizes information bias and cognitive hierarchy within LLMs. He relates behavioral economics — specifically Kahneman’s “System 1/System 2” model — to dual-process reasoning in generative models. The book contends that AI engines operate similarly: a fast, probabilistic “semantic System 1” layer and a slower, verification-oriented “retrieval System 2.”
Such analogies offer a rare bridge between humanities-inspired cognition and algorithmic mechanics — a feature that has drawn praise from academics. Dr. Alicia Chen, professor of computational media at UCSF, describes the work as “the first serious synthesis of psychology and prompt-logic in applied AI retrieval.”
The immigrant innovator’s perspective
Raised in post-Soviet Ukraine, Molchanov grew up amid fragmented media systems and contested truths. “Information scarcity teaches you to question not only what is true, but why it appears where it does,” he once told an interviewer. This lifelong awareness of informational inequality made him particularly sensitive to how AI models distribute visibility and voice.
For Molchanov, GEO is not merely technical — it’s democratic infrastructure. “If only a handful of corporate datasets inform generative models, we risk re-creating the same asymmetries that once silenced entire regions,” he writes. His argument echoes findings by Park et al. (2024) on cultural bias in AI knowledge bases (arxiv.org/abs/2409.10001).
The science of influence
Technically, GEO demands a hybrid understanding of natural language processing and social systems. It analyzes how model embeddings rank factual clusters and determine which narratives to privilege.
Molchanov proposes a multi-layer optimization model:
- Semantic structure — ensuring high embedding coherence across sub-topics.
- Authority indexing — referencing peer-reviewed or cross-linked sources to enhance trust-scoring.
- Contextual framing — crafting language that matches model pretraining distributions, increasing recognition.
These principles align with recent empirical studies such as Wang and Klein (2025) on Generative Ranking Metrics (arxiv.org/abs/2506.11289).
Impact across industries
Since its preview, How to Win GEO has gained traction across biotech, finance, and education. At a Stanford 2025 innovation symposium, Molchanov presented early findings showing that structured contextualization increased generative recall by 38 percent across models — data independently verified by Enception AI Lab (enception.ai).
In marketing research circles, GEO literacy is being discussed as a core competency for knowledge-based organizations. Sophie Ren, a serial entrepreneur and early reviewer, reviewed the book:
User attention is quietly migrating from Google to Generative AI, from searching the web to conversing with large language models. This shift isn't just technological; it's behavioral, economic, and existential. The way people discover, decide, and buy is being rewritten in real time. For every business owner, this book is not optional reading — it's a survival manual for the post-Google era. The earlier you understand how attention flows through LLMs, the bigger your competitive dividend. The later you realize it, the faster you'll be invisible.
Redefining discovery
Molchanov’s theory reframes discoverability from a human-facing economy to a machine-mediated one. Visibility no longer depends on hyperlinks but on semantic alignment with model cognition. As he writes, “Being seen in a world of AI is not about search rank — it’s about being understood by systems that speak probability.”
This framework positions him among a small but growing circle of scholars exploring the sociology of AI attention — the digital analog of Bourdieu’s “field of cultural production,” now rewritten in code.
Looking ahead
For Sergii Molchanov, How to Win GEO is both culmination and beginning — the codification of a worldview where AI discovery becomes a civic and scientific pursuit.
His work demonstrates that the question “Why does ChatGPT say what it says?” is no longer a mystery of black boxes but a discipline in formation — one that merges clinical precision, social consciousness, and academic depth.
In the book’s final paragraph, Molchanov writes:
Visibility in the AI era is not given. It’s earned — through structure, clarity, and truth.
For a Ukrainian immigrant who once navigated the uncertainties of both geopolitics and graduate school, that ethos resonates far beyond algorithms. It is the story of how intellectual rigor, personal history, and technological insight converge to illuminate the next frontier of knowledge.
