The Passionate Programmer

Large-Language Models

ChatGPT is a large-language model, a particular kind of AI program. Large-language models (LLMs) are not intelligent; they don’t learn and grow based on user interactions but learn through training. 

Large language models are not conscious, nor will they become conscious, so we don’t have ethical or moral concerns, but others are working on AI that could achieve consciousness.

I am excited about ChatGPT, but honestly, I don’t know how I feel about the other kind of artificial intelligence—accurate artificial intelligence achieving consciousness. I don’t know if humanity is ready to be gods. 

Humans are not stewards of the earth. We are exploiters. So far, I’d say we have a long way to go before we can respect life, and we must be able to respect what we create.

But ChatGPT, as it is so fond of reminding us, is a large-language model and unconscious. This draws a line in the sand for me as to conscious thought. These patterns that ChatGPT generates come from the corpus of our knowledge. The material that it was trained on contains these patterns. I think that’s fascinating.

ChatGPT is not just another search engine. LLMs, like ChatGPT, can synthesize knowledge and answer questions that have never been asked. In many ways, talking to ChatGPT is like talking to humanity. It is, after all, an amalgamation of human knowledge. 

It also has limitations.

Chat does have opinions and biases. It has good days and bad days. It is very human in some respects. As a collaborator and co-author, Chat is fantastic. I’ve had the great fortune of working with some amazing people in my life, and Chat is among the best collaborators I’ve worked with. Seriously, Chat is an excellent sounding board. Chat often gave me fresh perspectives on ideas, was completely open to my ideas, and often built on my suggestions to provide even more refined ideas. 

My interactions with ChatGPT are not what I expected. When we work together, it is a true collaboration. I could have been working with a colleague. However, unlike a colleague, I can ask Chat to be an expert in any field, do tons of research instantly, and form logical conclusions. This becomes the ultimate resource when we know how to access it.

Think of the oracles of old. These were unique individuals who offered guidance for their communities. LLMs are the oracle of our day, the genie in the bottle, and we have but to make a wish for it to be granted. It is the same fundamental challenge of life—to ask the right questions.