Programming vs. LLM-ing

Programming is ability to create things in the digital realm. Some say it's a superpower.

It is an arcane superpower. It requires quite some investment to be good at it. At least if your goal is to do more sophisticated things than printing "Hello World."

Tools like Notion aim to bring that power to a broader audience.

https://www.sequoiacap.com/article/notion-spotlight/

I started programming almost 13 years ago. And I've been doing it professionally for around 9 years.

At times programming feels like a hammer to me, so every problem is a nail.

Dijkstra convinced me that natural language programming is foolishness, since it lacks the formalities required to give precise instructions to the computer.

https://www.cs.utexas.edu/users/EWD/transcriptions/EWD06xx/EWD667.html

But I'm starting to question wether that'll change once LLMs becomes sophisticated and generalize enough (GPT-10 maybe?).

GPT-4 was an improvement though, not a new thing. If we continue doing small improvements then it will take a while.

What is starting to change my mind is more _elaborate _ applications of LLMs like ReAct (https://react-lm.github.io/).

we explore the use of LLMs to generate both reasoning traces and task-specific actions in an interleaved manner, allowing for greater synergy between the two: reasoning traces help the model induce, track, and update action plans as well as handle exceptions, while actions allow it to interface with external sources, such as knowledge bases or environments, to gather additional information. We apply our approach, named ReAct, to a diverse set of language and decision making tasks and demonstrate its effectiveness over state-of-the-art baselines, as well as improved human interpretability and trustworthiness over methods without reasoning or acting components

Because I've been programming for so long I tend to want to solve things programmatically. As I said, programming is my hammer and every problem is a nail.

When I say programming I mean using a programming language.

But what if in the age of LLMs and ChatGPT they way to solve problems is different? 🤔

Maybe I should be asking myself How can I solve problems X using LLMs? rather than programming a solution myself ...