In this episode, Dr. Anastassia Lauterbach interviews Anthony Scriffignano, Ph.D., a renowned expert in AI and data science. They dive deep into the intricacies of large language models, the phenomenon of AI hallucinations, and the role of prompt engineering. Anthony shares fascinating insights on spotting AI-generated inaccuracies and the human approach to counteract them, making this a must-listen for anyone curious about the intersection of technology and human cognition.
About Anthony Scriffignano
Anthony Scriffignano, Ph.D. is an internationally recognized data scientist with experience spanning over 40 years in multiple industries and enterprise domains. Scriffignano has an extensive background in advanced anomaly detection, computational linguistics, and advanced inferential methods and has multiple patents worldwide in these areas. Scriffignano was recognized as the U.S. Chief Data Officer of the Year 2018 by the CDO Club, the world’s largest community of C-Suite digital and data leaders. He is a member of the OECD Network of Experts on AI working group on implementing Trustworthy AI, focused on benefiting people and the planet.
Scriffignano has served as a commissioner for the Atlantic Council, most recently contributing to a Report on the Geopolitical Impacts of New Technologies and Data. He has briefed the US National Security Telecommunications Advisory Committee and contributed to three separate reports to the President, on Big Data Analytics, Emerging Technologies Strategic Vision, and Internet and Communications Resilience. Additionally, Scriffignano provided expert advice on private sector data officers to a group of state Chief Data Officers and the White House Office of Science and Technology Policy. Scriffignano serves on various advisory committees in government, the private sector, and academia. Most recently, he has been called upon to provide insight on data science implications in the context of a highly disrupted datasphere and the implications of the global pandemic. He has published, delivered keynote presentations, and participated in panel presentations extensively, in various settings, internationally, concerning emerging trends in AI and advanced analytics, the “Big Data” explosion, artificial intelligence applications and implications for business and society, multilingual challenges in business identity, and malfeasance in commercial and public-sector contexts.
Timestamps
03:33 Defining large language models (LLM’s).
05:23 Understanding parameters in LLM’s.
07:48 How adding data to an LLM affects performance.
10:05 Benefits and challenges of open vs. closed LLM’s.
15:13 Defining hallucinations in LLM’s.
19:39 The role of context in reducing hallucinations.
26:50 How to spot non-obvious hallucinations in LLMs.
31:11 Independent organizations tracking hallucinations.
39:10 The impact of jargon on communication with LLMs.
41:49 Romy & Roby: Roby Hallucinates for the first time.