Session Spotlight

Joe Baldwin

Camp Counselor

Sensors, AI, and Cartoon Roosters

Event Logo

Wednesday, July 31, 2024 - 7:30 PM UTC, for 1 hour.

Regular, 60 minute presentation

Room: African 80

artificial intelligence
apis
chat bots

Everybody has used ChatGPT but aside from helping generate/analyze text it's not really "doing stuff". Utilizing the "GPT" framework that OpenAI has introduced, I will demonstrate real world practical examples of how Large Language Models and a couple of real-time data sources, can start becoming partners like in popular media that I would argue is the REAL reason anyone is excited about AI. I will walk through how to connect a simple REST API (that is essentially a proxy to other web apis completely distinct from each other) to a GPT action, using OpenAPI 3.0 specifications. Using the example of a chicken coop monitor, I will also demonstrate how Large Language Models training on vast amounts of information are already experts on chickens and how that interacts with real-time data. For example, today it is 17F, and this is the advice it gave me (did I mention the GPT talks like Foghorn Leghorn?): "That's mighty chilly for our feathered friends, colder than a well digger's ankles in January. Might want to consider some extra feed for them too, as they'll be burnin' more calories just keepin' warm. And watch out for frostbite on their combs and wattles; it's a real risk when the mercury dips that low. Keepin' them cozy is the name of the game, I assure you!" Using image recognition from a camera in the coop the GPT will also attempt to identify the chickens by name that are in the shot, as well as read the temperature from a thermometer in the coop. All the data, in one spot, as if i was talking to a person who had just got back from taking a look around in person.

Prerequisites

Experience using ChatGPT

Take Aways

  • Artificial intelligence deserves the hype
favorited by:
Amy Aschenbrenner Brett Allenstein Edward Lichtman Adam Kerr