In this conversation, Dr. Erik J. Larson discusses the limitations and challenges of artificial intelligence (AI) models, particularly focusing on the misconceptions surrounding AI progress. He highlights the failures and biases of chat GPT models and emphasizes the need for a principled way to evaluate AI errors. Dr. Larson also addresses the future of AI, the dangers of misinformation, and the difficulty of training new models. He points out the shift from garage startups to corporate AI and the lack of investment in alternative approaches. Overall, he argues that AI is a disruptive technology but not yet feasible for many organizations. Dr. Erik J. Larson discusses the plateau of AI progress, the limitations of Moore's Law, the misconceptions of AI progress, the anti-human component of AI, the centralization of power in big tech, the fallacy of the singularity, the dangers of granting rights to AI, the loss of confidence in higher education, preparing for the future job market, and the fear of AI misuse.
AI scientist Erik J. Larson explains why today's large language models, including ChatGPT, may impress but still fall far short of true artificial intelligence—and how that misunderstanding threatens culture, knowledge, and innovation.
Guest Bio: Dr. Erik J. Larson is an AI scientist, tech entrepreneur, and author of The Myth of Artificial Intelligence, known for his critical insights into generative models and their cultural impact, shared through his Substack Colligo.
Topics Discussed:
Top 3 Quotes: