Skip to content

Working with us

Using ChatGPT to make us better developers

26th May 2023

Always seeking to improve how we do things, Huddle has been exploring using AI as a tool when writing code.

The software we use to write code allows us to embed a panel to the side of our screen, through which we can interact with the ChatGPT AI platform.

We have mixed feelings about AI here at Huddle, some of us are jumping in with two feet, others are excited but cautious, and some are counting down the days until Skynet becomes a reality.

Looking up information

Perhaps the most rudimentary, but nevertheless useful, application of AI is asking it to look up information for you.

Given that we can interact with AI directly in our code development software, it is much quicker to ask an AI a question than it is to switch to a web browser and search the same thing on Google.

There are hundreds of examples of this, for example “How do you validate an email address with Javascript” or “What is the CSS property to make an element 2x larger?”

More conversational than Google

Moreover, engaging with an AI is much more conversational, and means we can ask refining questions to drill down to the answer we’re looking for. This is very different to Googling things, where you need to search, look at some results, realise it’s not quite what you’re after, then reframe your search and “start again” each time, finding a fresh set of results.

The following example shows a developer asking an AI for a list of common file downloads (something that we might have Googled before now). You can see the AI respond, and the developer work with it to refine the result until it’s exactly what they’re after.

Supporting junior developer progression

Another great use case is using AI to support developer learning, particularly with junior developers.

Of course, sometimes the subjective style of a human teacher (e.g. a more senior developer) can really help someone to learn something quickly and effectively, but sometimes the objective and rational output of an AI can also be helpful.

For example, we can highlight some code and ask the AI “What does this code do” or “Explain this to me” or “Why is this written this way?”

Brainstorming ideas

Much of the time that goes into coding isn’t actually spent writing code, but trying to figure out how to actually go about something given the numerous paths available.

AI can help with this process, by asking the AI questions like “What are 3 ways I can achieve this?” or “What are the pros and cons of doing it this way?”

AI can be a great sounding board to bounce ideas off, and receive logical and rational responses. This can help to give a developer informed options quickly, for them to choose the best one.

Conclusion

I, for one, am making sure I’m using my pleases and thank yous – you never know which future AI mastermind your actions will get back to!

We’re looking forward to seeing how we can continue to use AI in development, and potentially other areas like project management and testing. Some things though, like account management and creative design, remain a deeply human activity for us, one that we are hesitant to involve AI.

May 26, 2023
By Tom Parson