Part of the TPC Seminar Series

---

Abstract:

The rapid advancements in artificial intelligence (AI) and the emergence of AI agents are revolutionizing scientific research and technological innovation. The introduction of AI agents has ushered in a paradigm shift, enabling machines to autonomously learn, reason, and make decisions in complex environments. In this talk, I’ll talk about the future of AI agents and their transformative potential in various domains, with emphasis on the work initiated through CACTUS (Cheminformatics agent) integrating science tools and open source LLMs. CACTUS represents a proof of concept of LLM-powered agents for molecular discovery, highlighting their ability to accelerate drug discovery and streamline the development of autonomous experimentation. However, the path to AI agents is not without its challenges. Developing AI agents that can effectively integrate domain knowledge, reason abstractly, and make ethical decisions requires significant advancements in machine learning algorithms, computational infrastructure, and interdisciplinary collaboration. The success of recent scientific agent development serves as a testament to the importance of bridging the gap between AI research and domain expertise, highlighting the need for close collaboration between computer scientists, biologists/chemists, and other subject matter experts. As we forge ahead in our quest for superintelligence, it is crucial to address the ethical considerations and potential risks associated with the development of agents. Ensuring the alignment of AI goals with human values, maintaining transparency and accountability, where superintelligent AI agents serve as partners in scientific discovery and technological progress.

Biography:

Dr. Neeraj Kumar is a Chief Data Scientist at Pacific Northwest National Laboratory (PNNL), where he works with a team of scientists and professionals to address important challenges in AI, energy, health, and biothreat sectors. As one of the co-founders of TPC (and PNNL lead), he contributes to the consortium’s efforts and collaborates with the ANL leadership team to drive progress. With experience in quantitative research and data science, he focuses on developing innovative solutions and leading programs at the intersection of fundamental discovery and AI-driven tools development. As one of the co-founders of TPC (and PNNL lead), he contributes to the consortium’s efforts and collaborates with the ANL leadership team to drive progress. In his talk, he will discuss the challenges and opportunities associated with developing AI agents that can integrate domain knowledge, reason abstractly, and make ethical decisions, while emphasizing the importance of collaboration across disciplines and considering the ethical implications of developing advanced AI agents.