How to Tweak a Neural Net to Correctly Reproduce Words

2min Snip

00:00

Play full episode

Summary

Transcript

Episode notes

GBT is a mathematical function that happens to be couple hundred billion terms long. Out of it comes a number, and that number or collection of numbers tell you the probabilities for different possible words to come next. We might have a trillion words that, that we have from things that are out there on the web. Let's try and tweak this neural net so that it will correctly reproduce sentences.

In this thought-provoking podcast episode, we discuss the mysteries surrounding Large Language Models (LLMs) and their implications for the future of artificial intelligence (AI) with Stephen Wolfram, creator of Mathematica, Wolfram | Alpha & Wolfram Language.

We dive into the fascinating world of Large Language Models (LLMs) and their surprising capabilities. He also discusses the underlying scientific principles behind LLMs, including the concept of computational irreducibility, and how they function as probabilistic sentence finishers. Wolfram reflects on the potential applications of LLMs in computational contracts and the challenges of aligning AI systems with human aspirations.

As the conversation delves into the future of AI governance, Wolfram explores the complexities of regulating AI and the importance of finding a balance between human intervention and autonomous decision-making. This episode offers a captivating exploration of the evolving landscape of AI and its impact on various industries and job roles.

If you enjoy the podcast, please follow us on ⁠Spotify⁠ and rate us 5 stars on ⁠Apple Podcasts⁠.

Socials

Follow Delphi Digital

Subscribe to Delphi Digital: ⁠https://members.delphidigital.io/home⁠
Twitter: ⁠https://twitter.com/Delphi_Digital⁠
YouTube: ⁠https://youtube.com/@Delphi_Digital⁠

Disclosures: This conversation is for informational purposes only and does not constitute legal or investment advice. Actual results may vary materially from any forward looking statements made and are subject to risks and uncertainties. This podcast is not investment advice. Do not buy or sell tokens based off this episode.