# Dr. Roman V. Yampolskiy — AI Safety Source Note

Accessed: 2026-05-18

## Core public sources

1. University of Louisville faculty page
   - URL: https://faculty.cse.louisville.edu/roman/
   - Verified details visible on page:
     - Roman V. Yampolskiy, Ph.D.
     - Associate Professor, Speed School of Engineering / Department of Computer Engineering & Computer Science, University of Louisville.
     - Founding/current Director, Cyber Security Laboratory.
     - Public university email listed on page: roman.yampolskiy@louisville.edu.
     - Speaking-opportunities email listed on page: rvyamp01@louisville.edu.
     - Official page also links a Facebook profile.

2. Personal/professional site
   - URL: https://www.romanyampolskiy.com/
   - Visible positioning: expert in AI safety, cybersecurity, digital forensics; tenured associate professor at University of Louisville; 100+ publications; speaker on AI ethics/future of technology.
   - Public social links visible on the site include Facebook and Twitter/X.

3. AI for Good speaker profile
   - URL: https://aiforgood.itu.int/speaker/roman-v-yampolskiy/
   - Verified details visible:
     - Faculty member, Department of Computer Science and Engineering, University of Louisville.
     - Founding/current director of the Cyber Security Lab.
     - Author of books including `AI: Unexplainable, Unpredictable, Uncontrollable`.
     - Main area of interest: Artificial Intelligence Safety and Security.

4. Taylor & Francis book page
   - URL: https://www.taylorfrancis.com/books/mono/10.1201/9781003440260/ai-roman-yampolskiy
   - Book: `AI: Unexplainable, Unpredictable, Uncontrollable`.
   - Publisher page describes the book as exploring why AI safety is challenging and addressing questions around intelligence, consciousness, values, knowledge, unpredictability, unexplainability, and controllability.

## Public social / professional links found

- Website: https://www.romanyampolskiy.com/
- University page: https://faculty.cse.louisville.edu/roman/
- X / Twitter: https://x.com/romanyam and older redirect `https://twitter.com/romanyam`
- LinkedIn: https://www.linkedin.com/in/romanyam
- YouTube: https://www.youtube.com/@RomanYampolskiy
- Instagram: https://www.instagram.com/romanyam/
- Facebook personal/profile lead: https://www.facebook.com/roman.yampolskiy and official/fan/page lead: https://www.facebook.com/DrRomanYampolskiy/
- Google Scholar lead from UofL page: http://scholar.google.com/citations?user=0_Rq68cAAAAJ

## Editorial framing for Managing Expectations

Use this line:

> Yampolskiy is one of the darker, more security-focused voices in AI safety. The value is not that every forecast must be accepted, but that his arguments force leaders to ask whether advanced AI can ever be fully explained, predicted, boxed, aligned, or controlled.

## Theories / themes to summarize

- AI safety and security as an engineering/security problem, not just ethics.
- Superintelligence could create catastrophic or existential risks if goals, capabilities, and deployment incentives are not controlled.
- The `unexplainable, unpredictable, uncontrollable` thesis: sufficiently advanced AI may be too complex to fully interpret, too adaptive to reliably forecast, and too capable to guarantee external control.
- The control problem: there may be no universal method to guarantee that a more intelligent system remains aligned with human intent.
- AI boxing/confinement skepticism: containing advanced AI in isolated systems may fail because persuasion, hacking, social engineering, side channels, and human operators become part of the system.
- Safety must include cybersecurity, model access, governance, verification, and deployment restraint.

## Caution

Present his position as an influential AI safety/risk argument, not a settled consensus. Pair with other AI voices later for balance.