Algorithms We Live By

This episode is brought to you by:

OKX. You may have seen OKX on McLaren’s Formula 1 race car or Manchester City’s football kit. But what is OKX? OKX has over 730 spot trading pairs, 280 derivatives markets, and 1000 options markets. It processes 400,000 requests per second with 99.95% uptime. That’s why over 20 million traders and institutions choose OKX when they want to trade. Visit okx.com to learn more.

[00:02:23] - [First question] - His impressions of recent innovations by ChatGPT

[00:03:51] - A rundown of the technical breakthroughs that led to today’s most cutting-edge AI

[00:06:08] - The types of questions he asks to test and evaluate new AI like ChatGPT

[00:08:13] - Questions or requests that would stump ChatGPT

[00:09:02] - Intuitive human tasks that would deeply confuse chat-based AI

[00:11:42] - Impressive R&D spend at OpenAI to optimize machine learning

[00:13:27] - The prospect of ChatGPT as competition for major search engines in the future

[00:15:27] - How his time at Google and Twitter informed his understanding of algorithms 

[00:17:20] - How social media algorithms work and how users reverse engineer them

[00:22:44] - His take on the amount of people it takes to run tech companies like Twitter

[00:26:39] - How protocols like Twitter Spaces factor into the modern tech ecosystem

[00:28:23] - His transition from big tech to Wall Street and how the two career paths intersect

[00:30:35] - How proximity to the investing side of the tech industry boosted his finance skills

[00:32:16] - The degree to which AI tech shows promise for universal applications

[00:34:01] - How his affinity for poker led him to crypto and NFTs

[00:35:55] - What inspired his jump to creating an NFT-based startup

[00:38:23] - The comparative value of powerful machine learning models versus the data itself

[00:40:03] - How he thinks about different NFT value models and their tradeoffs

[00:42:14] - The criteria that tend to make a given NFT more valuable than other similar ones

[00:43:48] - How his model for NFTs valuations actually works

[00:47:44] - Feedback he gets from users about his model

[00:48:55] - High valuations associated with both minimalistic and cluttered NFT visuals

[00:49:38] - The NFT liquidation auctions of FTX and Three Arrows

[00:51:09] - The proportion of NFT profits from instant buying versus bidding

[00:52:08] - How NFT traders make use of the model through the DeepNFTValue open API

[00:53:59] - The future of their profitability given that their goal is widespread distribution

[00:55:38] - The potential for lending and levering in future NFT markets

[01:00:24] - What he’s most excited to build over the next six months and six years