Totally agree. I am using my own charting software that I just would not have had the patience to create because I find javascript/react tedious and uninteresting. With chatGPT's help it was basically trivial. It also acted as a force multiplier and not automation.
My charting software has no value as a production ready product but is quite valuable to me.
Bank managers outsourcing less and less excel data manipulation to analyst seems rather obvious over time.
The AI narrative though seems so far from the real world in my experience. LLMs seem good at a type of language disambiguation and translation. Because of this, they can disambiguate and translate my ideas in English into javascript/react without issue. They are great tools if you are learning a language you can't currently speak or read.
What I find most interesting is I am rather certain LLMs simply can not produce the intersection of two different ideas in order to create something new. It can only find the idea if previously seen.
I have played around quite a bit with Google MusicLM and that absolute can do the intersection of various aspects of music. I have randomly produced music that sounds like a kind of bowed flute, in the Pelog scale from Balinese Gamelan with glass mallet instruments in the background.
AI art can do the intersection of various aspects of art.
ChatGPT is helpless though in trying to do the intersection of two ideas to create something new. I have found interesting ideas like a search engine in this way that already existed but that is obviously the best it can do.
There is also this process too that buying pet food online was a great idea but just not in 1999. So many things had to happen first.
A bank analyst is probably a dead job at some point in the future but certainly not in 2024.
AI is a tool like any other. It will change the world in some ways and not in others.
I use ChatGPT AI to write code faster. I outsource my coding tasks to junior developers less than I used to. I can describe something at a high level and the AI can usually get it right or close to right and save me a bunch of time - it's often just as good as a junior. Not perfect, but it's worth using and it makes a difference.
I think AI's initial impact is going to be hidden from view: more people will be able to make things that we might assume they made on their own. For example, there will be more small scale software being developed and coming to market. Joe Blow software developer who has a couple side projects at night will be able to code faster and get more to market. Nobody will know (or care) that he pumps out 4x the number of apps that he used to because he's now 4x more efficient in his spare time since AI does most of the more tedious tasks.