Let’s talk about AI. Or, as I prefer to call it ML - Machine Learning. Despite the fact that people have been working on artificial intelligence for many decades (a friend of mine was working on a masters thesis around AI in the early 90’s), we still aren’t that close to something the average person would call intelligent.
We are making leaps and bounds in ML, but at what price? Right now, we are in the middle of the “Wild West” of ML and LLM - Large Language Models. People are applying ML to EVERYTHING. It remains to be seen how a lot of these applications are going to work in the future, as we grapple with the implications for privacy, copyright, intellectual property, and personal rights.
At this moment in time, all the online services are rushing to add ML features to their product offerings. But some of those offerings come with a steep price in privacy, where you must agree to make all your information stored or accessible to the service available to the ML algorithms, in order to opt-in to the service. I anticipate mega-corporations locking down these features to protect corporate secrets and intellectual property. But individuals should be thinking about these implications as well. Artist rights are under siege, affecting not only high-profile figures in Hollywood but also cartoonists, caricature artists, fiction writers, and copywriters, among others.
And we have code being created by ML. Github Copilot was just the beginning, and the controversies around Copilot abusing some of the license terms in open source projects is unlikely to go away either. Should the developers of software be worried? At this point in time, ML code is looking at lot like the ML image generation (do you have 6 fingers?). ML can generate code that appears good at first glance, but upon closer inspection, it’s riddled with bugs. I wouldn’t rely on it for anything I care about.
Are there good uses for ML and LLMs? Sure. Some of the new tools I’m using (the Arc browser, the Warp terminal) are starting to REALLY make an impact on my day to day work. Speeding things up, making annoying things easier. And how about… no code solutions? I’ve been digging into those, for a series of training videos I’m producing for my 6 Pack Apps for Entrepreneurs program. There are some really interesting “so called” AI features that make pretty intelligent assumptions about what you are trying to do, and help… fill in the blanks. Sometimes they do it wrong, but mostly they are getting it right.
This week (in fact, tomorrow) I’ll be hosting a live session on Zoom called “No Code - Can you build an app in an hour?”. Come and watch me turn the app prototype I created a few months ago into a “working” app… in an hour? I guess you’ll have to sign up to find out if I can!