OpenAI released ChatGPT to the public on Nov 30, 2022. Since then, we have been learning about LLM’s and how they work. The discovery process has been unique since the technology itself is unique.
LLM’s don’t operate in the same way typical software programs operate. The standard perspective that usually works well to analyse your typical software architecture hasn’t worked in analysing these new LLM’s structures.
Although the process has been slower than usual, we are continuously learning more and more about how these new types of systems function. Both on a technical research level by academics and on a societal level by everyday users.
It seems as though the advances to LLM’s have continuously gained momentum since it’s introduction to the public. With all of the investment flooding into AI projects, t’s been progressing at an alarming rate. It appears we now live in a world with technological advancement that outpaces the rate of human learning. By the time you learn the latest technology, new technologies will have replaced it.
Whether this is just a feature of the early adoption phase of its life cycle or a sustained attribute over its lifetime is yet to be fully seen. Moore’s law, which states the number of transistors on a microchip doubles about every two years, has remained generally true since 1965.
Out biggest limitation currently is the way we’ve been taught to conceive of and interact with computers. LLM’s share the same genetic makeup of computers but are a different beast altogether.
Checkers and chess share the same fundamental structure. The same board, two sets of coloured pieces, controlled by two opposing players waging battle. Yet due to the nature of chess, it is much more intricate in its execution. As a result, the games a played very differently, and just because you know how to play one very well doesn’t mean you’re even at a beginner level of the other.
Once we learn to interact with LLM’s in a way that maximizes their leverage, we can utilize them to their full potential. Like a checkers player learning to play chess for the first time, once we see and understand the pieces and moves we can really start to compete.
Contrary to what the mainstream media will have you believe, as AI accelerates, the demand for software and the job of the software developer is not going anywhere. The role of a developer is changing however.
Like anything else, once we are exposed to something long enough, we eventually adapt. Fools rush in, but at some point, you have to read the writing on the wall and see the signs pointing to where things are heading when they do shift away from the status quo. It’s better to get ahead of the AI curve now because like the dissemination of the internet in the 90’s, AI is only going to become more and more prevalent in our society and daily lives.
Peter Linton
Software Developer
