Embedded Files
Adi
I am a PhD student at CS faculty at the Technion and advised by Yonatan Belinkov. I am interested in improving the truthfulness and safety of LLMs. My research focuses on hallucinations, safety, and interpretability in LLMs. I received the Council for Higher Education (VATAT) Scholarship for PhD students in data science and artificial intelligence.
If you find it interesting, feel free to reach out!
News
News
New paper - Seeing the Forest for the Trees: Emergent Agreement from Individual Conversations!!! - showing how agreement between LLM agents is hard to see in a single conversation’s internals, but when aggregated across conversations, behavior follows a Markov process and the internal representations follow it through a power law.
Selected Publications
Selected Publications
Seeing the Forest for the Trees: Emergent Agreement from Individual Conversations
Adi Simhi, Yaniv Nikankin, Jonathan Herzig, Idan Szpektor, Yonatan Belinkov, Shay B. Cohen
Under review
Old Habits Die Hard: How Conversational History Geometrically Traps LLMs
Adi Simhi, Fazl Barez, Martin Tutek, Yonatan Belinkov, Shay B. Cohen
Got accepted to ICML 2026!
ManagerBench: Evaluating the Safety-Pragmatism Trade-off in Autonomous LLMs
Adi Simhi, Jonathan Herzig, Martin Tutek, Itay Itzhak, Idan Szpektor, Yonatan Belinkov
Got accepted to ICLR 2026!
Trust Me, I'm Wrong: High-Certainty Hallucinations in LLMs
Adi Simhi, Itay Itzhak, Fazl Barez, Gabriel Stanovsky, Yonatan Belinkov
EMNLP 2025 Findings
Distinguishing ignorance from error in llm hallucinations
Adi Simhi, Jonathan Herzig, Idan Szpektor, Yonatan Belinkov
Arxiv 2024
Constructing benchmarks and interventions for combating hallucinations in llms
Adi Simhi, Jonathan Herzig, Idan Szpektor, Yonatan Belinkov
Arxiv 2024
Talks
Talks
Old Habits and Hard Choices: How LLMs Navigate Management Decision-Making and Earlier Interactions
CSAIL, MIT. Cambridge, MA. June 2026.
The Bau Lab, Northeastern University. Boston, MA. June 2026.
PRISM Reading Group, Harvard University. Cambridge, MA. June 2026.
Characterizing Hallucinations: A Two-Dimensional Analysis of Knowledge and Certainty
University of Edinburgh. Edinburgh, UK. October 2025.
University of Oxford. Oxford, UK. September 2025.
Research visits & Industry Collaborations
Research visits & Industry Collaborations
Visiting Researcher, University of Oxford – host: Fazl Barez. September 2025.
Visiting Researcher, University of Edinburgh – host: Shay Cohen. October 2025.
Research Collaboration, Google Research – with J. Herzig and I. Szpektor. 2023–present.
Research Collaboration, Pfizer – LLM hallucination in biological domain. 2024–present.
Google Sites
Report abuse