The latest post in the #SentientSyllabus project is a bit of breaking news, an accessible introduction to a paradigm shift that is rapidly gaining stem, although it is only about a week old.

While we are still struggling to fully grasp the implications of "speech from a box", there is a veritable explosion of interest and activities that take this a step further, treating individual prompts and responses as a thought, and chaining them together to "thinking".

It turns out this is rather easy to do. While one instance of #ChatGPT only has responsive agency, two together can sustain this and have full agency over an extended domain. Add in memory and remembering, the ability to call on external executor AIs, to access the internet and other services, and you get full-fledged computations, a novel paradigm that we may call "generative programming". Generative, because the program flow itself is not pre-specified, but it is generated on-the-fly with the evolving context of the task at hand. And it is all glued together by natural language and the everyday understanding of our world that language brings with it.

We discuss the #Smallville experiment, and introduce #AutoGPT, #BabyAGI, #Jarvis, and #LangChain - with a shout out to #pinecone; names that it will be worthwhile to remember - at least for a week.

https://sentientsyllabus.substack.com/p/trains-of-thought

Drop in, enjoy the thinking, and share with whoever would benefit.

Language is learning to talk. "Die Sprache spricht."

#GPT4 #HigherEd #AI #generativeAI #Education #University #Academia #GenerativeProgramming