Previously on AI - Episode 2

Exploring Apple Intelligence 🍎

Welcome to another edition of Previously on AI. We have an update on Apple Intelligence as it comes for iOS 18, Claude Artifacts being released on all tiers, and a, I’ll say, clever use of AI for generating fantasy football recaps.

Apple Intelligence

Apple dropped a new iPhone at their iPhone event September 9th. A chunk of the presentation was focused on Apple Intelligence, which will be coming to iOS in bits and pieces throughout the end of the year and beyond.

It’s offering a lot of the standard “personal assistant” type uses for AI like rewriting, summarizing, and proofreading. As well as enhancements to the Photos app like cleaning up images by removing unwanted objects or people. The ability to generate Memories movies within the the Photos app from a description is something that seems very cool.

The thing that jumped out to me was their approach to keeping potentially very personal information private and secure with their Private Cloud Compute offering. Apple aims to provide as much privacy processing data on the cloud as they do on-device. WIRED has a good write-up on Private Cloud Compute and if you want the detailed deep-dive check out Apple’s Security Research post on it.

Claude.AI Artifacts

Anthropic recently made their Artifacts feature of Claude.ai available across all account types. I played around with it a bit on the free version and it’s a very slick interface for working with and iterating on things that you create working with Claude. What I found particularly interesting, being a developer, was the ability to generate a React application and see it generate code as well as a working preview of the application.

I’m still not expecting AI to completely replace developers, but I do think it’s a useful tool for developers to utilize. It helps get past the blank page of starting on a project. In my experience with AI based coding platforms you still need the knowledge of coding practices to prompt, review and modify the code AI generates.

Artifacts can also be used for generating diagrams, visualizations, and more.

Using AI to recap fantasy football recaps

I’ve run a fantasy football league among friends for nearly 2 decades at this point. (I recently went back through Yahoo, ESPN, and CBS league histories to gather records all the way back to high school leagues, more on that in a different post.) We recently made the switch back to ESPN after spending a few years on CBS Sports. One of the things I miss from CBS was the weekly recap they would generate and send out to the league recapping that week’s league action. I guess that’s why CBS wanted to up the league cost to $180 a year.

I was able to recreate a pretty decent facsimile of these emails with ChatGPT and some copying and pasting.

I went to our league matchup scoreboard that features the matchups from the league and highlighted all the content and copied it to the clipboard. This gave me all the matchup data for the league in a standard format.

I then fed ChatGPT what that format was and how it should use the data to generate recaps of each of the 4 matchups between 8 total teams. It worked surprisingly well. Most surprising was the data had player names in the format of F. Lastname and ChatGPT was able to accurately input the player’s full name.

Rad Tabs

LABS.GOOGLE has a collection of all Google’s various aI tools for you to easily try out.

Brilliant is basically Duolingo for learning about machine learning and AI. I ran through their introduction to LLMs course in a hour or two. It covered the concepts in a visual and engaging way.

Fullstack Cloudflare RAG on Github gives an example repo for building a RAG app with Cloudflare. Pretty cool if you’re looking at how to get started on a project.