ChatGPT Part 1

I love computer technology. All aspects of it. That's why it has been my career choice my entire adult life, even though my jobs have changed over time (software developer, operations, systems administration, security, etc.). To an outsider, computer technology is akin to magic. But for those of us in the industry, it is our job to understand how that magic works. It is not a stretch to say that demystifying magic is a shibboleth within the tech industry. It's also why there's not a lot of new technology that feels magical, like it did when I was just starting in this industry.

Enter ChatGPT. This is the first time in decades that I have felt that magic again. Typically with new technologies I can see through the hype, but ChatGPT is something on another level. ChatGPT is a large language model (LLM), but instead of having me screw up explaining what it is, let's here ChatGPT explain itself.

ChatGPT is a language model developed by OpenAI.

Think of ChatGPT as a super smart AI that has learned how to have conversations like a human. It has been trained on a lot of text, so it knows a lot of information and can answer questions. When you ask ChatGPT a question, it uses its knowledge to generate an answer.

A Large Language Model (LLM) is a type of artificial intelligence (AI) model that is trained on a massive amount of text data to perform various natural language processing (NLP) tasks. The goal of an LLM is to generate human-like text based on a given prompt or context. For example, it can be used to generate answers to questions, write coherent articles or stories, translate text from one language to another, summarize long documents, and many other NLP tasks. Some of the well-known LLMs include GPT-3 by OpenAI and BERT by Google.

What I did was give a "prompt" to ChatGPT and it responded with the most probable phrases that meet the goal of my sentence. But when I first heard about ChatGPT I didn't play with it. I waited to try it out for many reasons:

  1. I saw it as a pseudo job killer.
  2. I saw it as an overhyped tool used by non-technical people.
  3. I didn't like having to login.
  4. It's cheating.

In short, I was prejudiced. None of my reasons had anything to do with the tool itself. I listened to others without doing my own research. I didn't even give ChatGPT a chance. Last week I dove in feet first to see what ChatGPT was all about.

I was blown away. I ended up staying up past my bedtime because I was so entranced by it.

But I also see its shortcomings. I see ChatGPT performing most of the boilerplate work of modern society which, unfortunately, most people don't realize now appears to make up most of society at large. But this is a good thing because it means we can start focusing on being creative and innovating new ideas, while tools like ChatGPT perform the menial work of producing boilerplate that we seem to think it actual work, but is in fact just red tape.

It reminded me of using the Web in the late 90s and early 00s. I needed to improve my keyword skill set (akin to the premise in Rainbow's End) otherwise the search results wouldn't be useful. With ChatGPT, it is all about prompts. While I can now use natural language to ask questions, I still need to set the environment and scope of my request to ChatGPT, otherwise its answers won't be useful. And doing this prompt work requires some semblence of skill on my part.

Once ChatGPT comes up with an answer, it requires knowledge about the topic to discern whether it is right or wrong. I found that it was correct about 80% of the time, which is already amazing, but when it failed it did so in ways that someone without training on the topic wouldn't catch.

So ChatGPT is more like an extension of my brain rather than a replacement for it. This was good to know as I tried to get it to help with a problem I was struggling with. I'll pick this up in Part 2.


Just for fun, I also prompted ChatGPT to "talk like a silicon valley executive and re-explain what ChatGPT is." Here was its response.

ChatGPT is a cutting-edge AI language model that is disrupting the NLP industry. At its core, ChatGPT represents a major step forward in our ability to harness the power of machine learning to generate human-like text at scale. With its unparalleled ability to understand and respond to natural language inputs, ChatGPT is positioning itself as the go-to solution for businesses looking to integrate advanced language capabilities into their products and services. Whether it's answering customer queries, generating marketing copy, or providing valuable insights, ChatGPT is poised to revolutionize the way we interact with language-based technologies. In short, ChatGPT is the future of NLP and we're thrilled to be at the forefront of this exciting development!

« Previous: 2022 Tech Bust
Next: ChatGPT Part 2 »