FireFox is blocking Twitter content

To view content on tw-rl, follow these steps...

  1. Click on the shield in the address bar.
  2. Toggle the switch at the top of the panel.
Sign In →
Sign In →
Read Thread
ChatGPT has changed software development! However, 68.1% of Software Engineers still don't use it. As a CTO, I'm telling my teams to use ChatGPT in their daily work to increase both velocity and quality. This is how πŸ‘‡
In my poll last week, I was surprised to learn that most Software Engineers aren't using either GPT-3, nor Github Copilot, nor any other AI tools. It seems they either see no value, or they don't know where to start.
Sergio Pereira πŸš€ @SergioRocksQuick poll: - If you're a Software Engineer, are you using AI tools to help you write code? (This vote is anonymous, in case that's relevant)
I'm describing below 10 use cases where ChatGPT will make your work easier and/or better as a Software Engineer. You can steal the prompts and use them in you day-to-day work. Happy codingπŸ‘‡
1/ Generate boiler plate code To build a new project from scratch, I ask ChatGPT to create a skeleton of the app I need, using the technologies, frameworks and versions I choose. I can even make it part of my team's documentation. It saves at least an hour or so each time
2/ Research and compare Many times there are different ways to implement something. So my usual approach is to build a rough PoC of two competing approaches and compare. ChatGPT can do it for me: - React vs Vue - Algorithm A or B - etc This saves half day of work, easily.
3/ Explain code How many times our work is simply to understand a code base that wasn't built by us? Spaghetti code, with no comments. We die a little bit line by line. Now we can ask ChatGPT to explain what the code does, and we save a bunch of time.
4/ Comment code We can extend the point 3 above by adding line by line comments to that code base we didn't build. But we can also make our own code properly commented before shipping, by asking ChatGPT to add comments. No more uncommented code in PRs.
5/ Write test cases Ask ChatGPT to generate test cases for a list of scenarios. Even tell which framework, version etc you want to be used. Boom, it will do it for you. No more untested PRs.
6/ Write documentation You can ask ChatGPT to write documentation on things like: - How to spin up a certain code base, - The packages required for it to run - What the code does, - What are the know limitations. - Etc. It might not be perfect, but the 80/20 rule applies.
7/ Generate regexes Regexes are one of those specific syntaxes we don't use from time to time. We need to Google, look up syntax and spend significant time every time we need one. Not anymore, we can just ask ChatGPT to generate it.
8/ Rewrite code using correct style I just wrote down some code conventions from one of my teams and added it to the prompt, along with some spaghetti code. Very useful when merging code from different repos/teams, that would need to be refactored before merging.
9/ Find bugs in your code When you know what you want, but the code doesn't do what you want. Instead of adding console.logs all over the place, you can ask ChatGPT to spot the bug for you. This works best on function level bugs, not repo level. Still, a massive improvement.
10/ Leetcode type algorithms If you need a specific isolated function, to run on optimal complexity. ChatGPT can get you a very fast start. An obvious use case for this is coding interviews. But sometimes it's useful in the day to day work too. See these 3 examples:
Sergio Pereira πŸš€ @SergioRocksChatGPT is the last nail on the coffin, for coding interviews. Three examples:
Get all prompts from the 10 examples above in the link below. It's totally free, just download and use them as you please.
For all Software Engineers who are shitting on AI tools. Don't do it. You should NOT be afraid of AI taking your job. Instead, you should be excited that you can leverage it to be more productive and/or create higher quality output.
Inb4: ChatGPT has obvious limitations. It's answers are incomplete or wrong sometimes. In my 13 years of experience writing and reviewing code, critical thinking is a must-have skill, both with humans and with AI. Don't push ChatGPT's responses to prod without reviewing first.
I personally apply the 80/20 rule. I leverage ChatGPT on the 80% of tasks that create 20% of the value. That's usually tedious time consuming tasks that must be done regardless. This way, I can focus more on the high leverage work that creates most of the value.

My Notes:

Select to add to your #gallery:
Sergio Pereira πŸš€

Pro Curator

$99 /yearPay what you can