21

ChatGPT plugins to your IDE - How to prepare?

Profile picture
Data Engineer at Financial Companya year ago

With AI being all the rage and that only becoming more so, I'm trying to predict the future of development along with the rest of the developer community. According to Tech with Tim, the next big thing is ChatGPT plugins to your IDE so it has the context of your entire project when you ask questions. This will allow it to generate a lot more code.

Assuming this is the near future, is there any way to prep for that now?

Are there other developments on the horizon that we can/should try and prep for?

1.7K
3

Discussion

(3 comments)
  • 18
    Profile picture
    Tech Lead @ Robinhood, Meta, Course Hero
    a year ago

    Also, for folks who are interested, here's the official waitlist for ChatGPT plugins: https://openai.com/waitlist/plugins

    It seems like there's some ChatGPT plugins already for IntelliJ, but they aren't from OpenAI and are therefore unofficial. I would be very wary installing these as they may be malware trying to steal your data and code.

    If you're currently working, especially if you're at Big Tech, I highly recommend talking to your manager before giving AI tools access to your codebase. We all know how poor tech companies can be at following security and privacy rules, both intentionally and unintentionally.

  • 16
    Profile picture
    Tech Lead @ Robinhood, Meta, Course Hero
    a year ago

    My advice to prepare for ChatGPT plugins and the rise of consumer AI in general is to get extremely proficient at asking sharp questions, both the initial ones and follow-ups.

    This is because at the end of the day, AI (at least in its current form) is still just a function:

    • It takes in input
    • After processing the input, it returns an output

    Like any human, AI cannot read your mind (at least not yet), so the quality of its output will vary based on the quality of your input. This means:

    • If you ask great questions, you will get great answers
    • If you ask bad questions, you will get generic, non-actionable, and maybe even incorrect answers

    AI poses another risk by excessively hand-holding software engineers, similar to how some engineers are overly carried by their senior engineer mentors:

    • There's not much point to using AI if you don't fundamentally understand the solutions it's feeding you: If you don't truly get what's going on underneath, you will eventually be exposed as an engineer and this severely holds back your career growth.
    • ChatGPT does a pretty good job proactively explaining its solutions, but different people learn in different ways and its initial explanation may not resonate with you.
    • If you don't 100% get what ChatGPT or whatever other AI interface is feeding you, keep pushing it with follow-up questions until it frames the concepts in a way you're able to fully grok.
    • In general, I recommend trying to come up with your own solution to every problem you have ChatGPT solve for you. What's the difference between your approach vs. the AI's? Why is that way?

    Here's the Taro resources I recommend around getting good at asking questions:

  • 10
    Profile picture
    Senior Software Engineer at IBM
    a year ago

    I'd just keep informed of a lot of the different tools out there the best you can. You can always come across the gem and just bring it into your workflow as long as legal concerns allow for you to, but for you to predict the future is exhausting and will likely end in burnout.