Bill Dembski: Pursuing Truth and Trust in AI and LLMs
How does AI stack up when it comes to accurately representing the theory of intelligent design? Today, host Andrew McDiarmid speaks with mathematician and philosopher Dr. William Dembski about the reliability and accuracy of large language models (LLMs) like ChatGPT and Bard, particularly concerning intelligent design.
This second half of the conversation highlights the importance of independent verification when using AI, as LLMs can “hallucinate” or generate false or biased information. Dembski advises an approach of verify, then trust, turning the old Russian proverb on its head. McDiarmid and Dembski also explore the potential for AI to enhance human capabilities and education if used judiciously, rather than becoming a crutch that erodes critical thinking. But that will require that we hone our skills as investigators, learning how to prompt AI with finely tuned questions and prompts. It also means remembering that an AI model is only one source of information among many. Collecting multiple viewpoints, especially on important topics, can provide a richer, fuller understanding. Finally, Dr. Dembski touches on AI’s conceptual connection to truth and its limitations in achieving artificial general intelligence due to its reliance on existing data rather than original thought or interaction with the physical world. This relates to Dembski’s current work on conservation of information.
This is Part 2 of a two-part conversation. In part 1 of this exchange, Dr. Dembski discussed his explorations into the reliability and accuracy of large language models on the topic of intelligent design. He argued that this new technology has promise, and definitely goes beyond what Wikipedia can offer, but it will take work from us too. Watch Part 1 below.
Dig Deeper
- Read one of the articles that inspired this conversation: Artificial Intelligence Promises to Level the Playing Field for ID
- Watch Part 1 of this conversation: