007 - Danielle Boccelli on the limitations of language

Apr 12, 2023

Ask episode

Chapters

Transcript

Episode notes

KMO speaks with data scientist Danielle Boccelli about the possibilities and limitations of learning strictly from language. The topic is large language models, but this conversation was recorded before the announced release of GPT-4. So much of what we know and what we are is not encoded in language, so there's only so far AI can progress without learning from something other than just text scraped from the internet.

Episode Title:

AI and Large Language Models: Bridging the Gap between Public Discourse and Research Progress

Episode Keywords:

Artificial Intelligence, Large Language Models, AI Research, Transformer Technology, Text Generation, AI in Education, AI Dilemma, GPT-4, AI Applications, AI Limitations, Public Discourse, Embodiment, Content Creation, Contract Writing, Textural Artifacts, Commercial Pressures, Chatbots, Fact-Checking, Polanyi's Paradox, Artificial General Intelligence

Episode Tweet:

In this episode of the KMO show, join us as we explore the fascinating world of Artificial Intelligence and Language Models with Danielle Boachelli. Get ready to gain valuable insights into AI and its implications! #AI #languageModels #KMOshow

Episode Summary:

In this episode of the KMO show, I had an insightful conversation with Danielle Boccelli, a data scientist and PhD student, about artificial intelligence and large language models. We discussed the gap between public discourse and actual research progress, the limits of learning from text, and the potential applications and limitations of large language models and transformer technology. We also explored the complexities of text generation, the impact of AI on education, and the commercial pressures driving the development of AI language models. Finally, we touched on the AI dilemma presented by Tristan Harris and Aza Raskin, and the evolving capabilities of large language models like GPT-4. Join us as we delve into these fascinating topics and gain valuable insights into the world of AI and language models.

Chapters:

(0:00:21) - The Reality of AI

(0:07:14) - Limits of Learning From Text Alone

(0:18:21) - Limitations of Transformer-Based AI Technology

(0:25:07) - Exploring the Complexities of Text Generation

(0:31:01) - The Impact of AI on Education

(0:42:52) - Language Models and Consequences

(0:57:03) - Exploring AI Dilemma

Chapter Summaries:

(0:00:21) - The Reality of AI (7 Minutes)

Full show notes at: https://www.patreon.com/posts/81435006