Posts by Collection

Position: LLMs can’t plan, but can help planning in LLM-modulo frameworks

Published in International Conference on Machine Learning (ICML) as Spotlight, 2024

This work proposes a novel neuro-symbolic approach for reliable planning with LLMs by having external model-based verifiers and critics in a bi-directional interaction regime.

Recommended citation: Kambhampati, S., Valmeekam, K., Guan, L., Verma, M., Stechly, K., Bhambri, S., ... & Murthy, A. B. (2024, June). Position: LLMs can’t plan, but can help planning in LLM-modulo frameworks. In Forty-first International Conference on Machine Learning.
Download Paper

Incorporating Human Flexibility Through Reward Preferences in Human-AI Teaming

Published in Arxiv, 2024

This work establishes a human~Al preference-based reinforcement learning cooperation game where the RL agent queries the human-in»the-Ioop to elicit task objectives and human preferences on joint team behavior.

Download Paper

LLMs Need to Go Beyond Computational Confidence Metrics to Establish Trust

Published in AAAI 2025 Symposium on AI Trustworthiness and Risk Assessment for Challenged Contexts (ATRACC-25), 2025

Perspective Paper on highlighting the Insufficiency of Computational Assessments of Trust and the need for Human-Centered Evaluations of Trustworthiness in Generative AI Systems.

Recommended citation: B Murthy, A., & Sanneman, L. (2025). LLMs Need to Go Beyond Computational Confidence Metrics to Establish Trust. Proceedings of the AAAI Symposium Series, 7(1), 131-136. https://doi.org/10.1609/aaaiss.v7i1.36878
Download Paper

Metrics for Holistic Evaluation of LLM Reasoning about Action, Change, and Planning

Published in NeurIPS 2025 Workshop on Evaluating the Evolving LLM Lifecycle: Benchmarks, Emergent Abilities, and Scaling, 2025

New informative Metrics for Evaluation of LLM Responses in Planning and Reasoning tasks.

Recommended citation: Murthy, A. B., Mink, J., & Sanneman, L. (2025). Metrics for holistic evaluation of LLM reasoning about action, change, and planning. In NeurIPS 2025 Workshop on Evaluating the Evolving LLM Lifecycle: Benchmarks, Emergent Abilities, and Scaling.
Download Paper

CSE 571 Artificial Intelligence

Graduate-level Course in Artificial Intelligence, SCAI, Arizona State University, 2023

Instructor: Dr. Subbarao Kambhampati
Graduate Teaching Assistant for around 120 Students. Responsibilities involved Design of programming Assignments and grading schemes, holding Office Hours, Assessment, Review and Grading of exams, course coordination and logistics.

CSE 565 Software Verification, Validation & Testing

Graduate-level Course, SCAI, Arizona State University, 2024

Instructors: Dr. James Collofello, Dr. Robert Lynn Carter, and Dr. Ayca Tuzmen
Graduate Teaching Associate for around 500 Students. Responsibilities involved Course instruction, holding Office Hours, Design of assignments and grading schemes, Review of exams and course coordination with grading staff.

AAAI Special Track on AI Alignment

Program Committee, NeurIPS, 1900

ICML (International Conference on Machine Learning)

Program Committee, ICML, 1900

NeurIPS (Neural Information Processing Systems)

Program Committee, NeurIPS, 1900

Intro to Informatics

Undergraduate Course, SCAI, Arizona State University, 2023

Instructor: Dr. Dianne Hansford
Teaching Assistant for around 60 Students. Responsibilities involved holding office hours, project guidance, course assessment and grading. Topics covered Java, SQL, Web Design including HTML, CSS, D3.jS, and Game Development.