This week I worked primarily on finishing our implementation of data storage within a session so that the user is able to have a personalized experience when they log in. I also implemented functionality for users to be able to log out of a session and be redirected back to the website’s home page. Finally, I fixed some tables on our user home page and coordinated with the backend team on how they can use the data storage information in additional queries.
This week, I learned a couple of very important lessons and also had some lessons further reinforced.
First, you really have to be as specific as possible in your prompts to the AI. I asked ChatGPT the same question in 3 different ways and got 3 different answers each time. It was up to me to determine which answer was the best, even if all 3 answered my question to some degree. It is important to be as specific as possible so that the AI tool has as much information to properly answer the question.
Next, I learned that the AI NEVER thinks it’s wrong. And this plays a little into the fact that it is terrible at debugging, but the way it confidently answers questions sparks confidence. So when something doesn’t work, it is frustrating but also can be annoying trying to figure out why it doesn’t work. You definitely need to do your own testing/legwork to verify it’s answers.
And finally, it was reinforced to me this week just how terrible AI is for debugging. I’ve said it almost every week, but I cannot understate how bad it is. It regurgitates the same information every time you ask about an error and usually puts the fault on you, rather than thinking about ways to fix the code. Most of the time, it seems to be due to simply using outdated syntax or practices depending on the framework.
Even despite this, AI really is a great tool to get you started, but it just needs to be used to supplement your own skills.
Leave a Reply