I love listening to podcasts about GPT-3. This one is really special, it's about a founder's experience building with the API early on and even building a business around it now.
Here's the episode synopsis:
Imagine being able to learn about the theory of relativity from Einstein himself? Mckay Wrigley is the Founder and CEO of LearnFromAnyone, a service being incubated that aims to leverage OpenAI’s GPT-3 API to build an AI teacher that is an expert in any topic and can assume any personality and voice too.
Mckay joins us on the show today to speak about learning code, building a brand in today’s social media landscape, and working on a platform that leverages one of the most exciting and contentious areas of tech – AI. Mckay tells us about how he quit law school and enrolled in a coding bootcamp while sharing his learning process on Twitter, which landed him a job at Hivewire as a full stack developer after only six months. At Hivewire he was able to upskill at a rapid pace and when he heard the rumors of GPT-3’s imminent release, he signed up for an API key and was promptly selected. He talks about his early experiments with the API, getting the idea to build a chatbot that could teach using real-world personalities, and how the first version of this project went viral just after its release. We discuss some early stumbling blocks which kick off an interesting exchange about the pitfalls of AI, how far we are from general artificial intelligence, and the challenges Mckay is facing as a full stack developer building an AI product using another company’s API. Along with all this, we also get into Mckay’s early experiences as a Founder, his optimistic attitude, and the case for positivity at a time like this where many of the narratives we hear are all doom and gloom. For a super exciting episode about the horizon of AI and how it can be harnessed for good, be sure to tune in.
Key Points From This Episode:
OpenAI’s GPT-3 API and how Mckay founded LearnFromAnyone
Uses of OpenAI’s GPT-3 API; a language generator that learns from all the internet.
How our modern world allows people to build personal brands in 2-3 years.
Mckay’s journey quitting law school in order to learn how to code
Tech and soft skills Mckay learned after getting his first developer job at Hivewire.
The role leveraging Twitter to document his coding journey played in Mckay finding work and getting access to the GPT-3 API key.
Mckay’s idea to use the API to build a chatbot that allows you to learn from anyone.
How well the first version of LearnFromAnyone did, going viral in a few hours of being made.
Subjects LearnFromAnyone is better at teaching.
Old challenges of having to dig up relevant data to make a chatbot like this which the GPT-3 API solves.
Limits to GPT-3’s knowledge and the goal of LearnFromAnyone to reduce these.
The human ‘constant learn’ feature versus GPT-3’s snapshot-based neural nets.
Mckay’s thoughts on OpenAI as a company.
Snags in LearnFromAnyone’s initial release; toxic responses thanks to being trained on the internet.
Dangers inherent in AI products and the need to implement safety precautions.
Mckay’s thoughts on going all-in on a product built off another company’s API.
Mckay’s experiences from starting a company during COVID.
The tech stack Mckay has used to build LearnFromAnyone; Python, Flask, Node, React, AWS.
Companies that inspire Mckay, from OpenAI to Tesla, Stripe, and more.
A case for optimism and why it is exciting to be alive today.
This weekend I'll mainly be resting and learning more about the GPT-3 pricing model. Will report back if I discover anything interesting. Hope you all take some time to rest too!