In a conversation with a friend about generative AI and autonomous agents, I wondered if one could create a pipeline to evaluate if any given article on the web would be interesting to a me:
I have years’ worth of articles that I’ve saved and then read on the read-later service Instapaper. I know which ones of them I found interesting. I could use this text as training data for an AI.
Then I could ask the AI to compare the text of any potentially interesting article I found on the web with the data it has already analysed and output a suitability score – a probability of me actually finding the article worth reading.
If I said yes, it could add it to Instapaper and to its training data set.
This is deeply personalised. One person’s Instapaper/ Pocket/ Evernote/ Safari Reading List is likely to be very different from another’s.
It’s also the kind of work that AI could improve human quality of life with. With the explosion of published material on the web, there’s also an abundance of articles, posts, Twitter threads that I’d find interesting. But even as this grows exponentially, my capacity to discover, evaluate and read is linear – in fact, it has a hard limit of both time and cognitive capacity. This is where I’d like technology to do a priori prep work for me.
Over time this could get interesting.
A generative AI could take articles of topics I like but which are poorly written (subjectively) and rewrite them to make them more appealing to me.
An AI could also expand to other media – it could extract audio from YouTube videos or podcasts, convert to text, feed it into the same suitability analysis as above and recommend whether I’d like to listen or watch them.
It could give me the option to read a podcast or listen to an article or switch between them at will.
It could sequence articles in such a way as to introduce a new topic to me and educate me step by step about it.
AI is right at the point of making knowledge infinitely personalised by making it infinitely malleable.
The next several years are going to be fun.
One reply on “AI’s about to make knowledge infinitely personalised by making it infinitely malleable”
Extremely interesting! Boggles the mind as to the sheer scope of something so dynamic. The metaphor that comes to mind is of the little paper kaleidoscopes we enjoyed as children, where the constant rotation delivered new and exciting permutations as you turned it around, thus creating from a paradigm that not only didn’t exist before, but that was also discovering itself as it took shape