One thing about this current crop of “AI” tools going around that sticks in my head, I think I got it from a Shawn Wang article, is that “AI” is good at:
- Small Input -> Big Output
- Big Input -> Small Output
So type “bernie sanders wearing a blue web standards beanie” into an image generation “AI” tool and those 50 characters of text become millions of pixels of image representing it (yeah). Small Input -> Big Output.
On the flip, train a model on, say, 15 million websites, then ask it to write stories where your kids are friends with Mario and Luigi. Or train an image model on every X-Ray ever taken so you can give it one more for analysis and see if that really is a fractured rib or not. Big Input -> Small Output.
On the Small Output side, I saw Paul Stamatiou recommend a Chat GPT-based summarization Chrome Extension. No time to read the full article? Summarize it and get the gist now. Mixed feelings there. I don’t hate the time savings but feel “AI” is going to strip away the feel of the piece. I need that proper feel to know if I really care or not. I have it installed, so let’s see if I end up using it much or not.
Related, I’ve still been cracking open Artifact app on my phone sometimes. They still haven’t solved that “it’s nigh intolerable to read content sites on mobile without high powered ad blockers” problem quite yet. They have “reader mode”, which is nice but you have to flip it on every single time so I wouldn’t consider it solved. I prefer reading the designed website anyway, just sans-horrible-ads. A new take on the solution is summarization:
That’s neat. I like it.
“AI” is best when it’s just a feature and I don’t have to type anything.
I think it’s possible that if every article is summarized it will become very evident that most people just read the same thing every day, just rehashed by a different person with a different style.
Doubly so when more and more articles are AI generated themselves.