When I say I wasn’t sure, I really wasn’t sure. It’s not that I didn’t think using ChatGPT would help me be more productive, or that I wouldn’t be able learn to steer it effectively. Rather, what I wasn’t sure about was whether regularly employing ChatGPT would erode some of the fundamental parts of my working day.
My concerns were two-fold:
- I take pride in (and get joy from) writing code - simply put, it’s fun, and there’s always something new to learn. Do I really want a magic bot to tell me all the answers?!
- I often use writing as a tool to help me develop my thinking, and like to take my time doing it. Will I lose this if I use an LLM to do the bulk of my writing for me?
Surprise, surprise, I was wrong. Here’s what I’ve found after a few weeks of regular use of ChatGPT.
Accelerating technical work and learning 🔗
Using ChatGPT, the path I follow to take an idea and realise it has shortened by an order of magnitude. As a CTO I can’t spend hours and hours in the weeds of different programming languages and frameworks. This meant that I previously had to scour documentation or Stack Overflow whenever I wanted to do something simple, like bootstrap a new project. Now, all I need to do is ask ChatGPT, and it will give me clear steps to follow to get up and running. This means I get to spend more time learning and doing, and less time searching.
ChatGPT also saves me tons of time when I’m debugging and fixing issues. I can give it an error message (often without needing to bother to give it any context - a simple copy and paste job), and more often than not it will either provide an answer, or some additional troubleshooting steps.
Prioritising writing tasks 🔗
Like I said above, I like to take my time when I’m working on a piece of writing to really think through how I’m communicating. I find it helps me solidify my thinking and see things in new light. But not every writing task offers those rewards. To be honest, sometimes you just need to get words on a page, and I’m not great at switching off the part of my brain that gets worked up by poorly written sentences. That’s where ChatGPT comes in: I can outsource those lower value writing tasks and focus my time on the jobs that will have the most impact. Obviously the language model’s output will always need reviewing and refining, but it tends to get me 90% of the way there.
There’s also the ‘staring at a blank page’ hurdle that most people are familiar with. It’s the stage where you know you need to write something, but have no idea how or where to start. ChatGPT can give you that starter for ten - even if you end up discarding everything it produced, at least you have overcome that first hurdle.
Collaborative thinking 🔗
The conversational nature of ChatGPT allows me to engage in a dialogue that rapidly refines and validates (or invalidates, as is often the case) my thinking about a topic. By providing some context and then outlining an idea I have, I can quickly and iteratively enhance, modify, or scrap a train of thought.
During the last few weeks I’ve changed from being skeptical about the effects of embracing ChatGPT in my daily work, to being fully on board. It has undoubtedly changed the way I work for the better. If you haven’t tried it yet then I would encourage you to approach ChatGPT (or any other large language model tool) with an open mind, but also to remember two important points:
- Any information you give to ChatGPT will be used to train the language model, making it publicly available to other OpenAI users
- ChatGPT - and generative AI tools in general - are not a silver bullet. Don’t expect it to magically solve all your problems for you!
(P.S. before you ask - no, this post wasn’t generated by ChatGPT 🤖)