Lately, I've been using Google Bard and ChatGPT way more often, often for Python code samples, dealing with CSVs and Other data. I realized That Google Bard could push me way faster than I was before, and I appreciate the tool now, especially for the speed it responds in.
Why Google Bard is often way better than Stackoverflow
Speed
Stackoverflow is a great platform, and I've used It a lot and mostly used it for one library, Phaser.js. This library is very active on Stackoverflow, my very first question was looked into by Photonstorm, the creator of the library. One amazing user on StackOverflow, winner_joiner, almost always responded to these posts within 12 to 24 hours. Thanks to him, I accomplished way more than I could have on my own.
Yet the moment I tried asking questions about mainstream frameworks, like Flask or About an error I got on Firebase, Both questions were ignored. I later tested both of these questions on Google Bard and after a couple of back-and-forth error reports and code changes, I was able to solve both issues.
Yet also on the flip side, when it comes to less mainstream frameworks like Phaser.js, models struggle at providing the same level of detailed knowledge humans have given me, with most models, at best, rendering a couple of squares on the scene*.
So, for now, StackOverflow wins if the community is lively, beginner friendly, and a place for fast, quality answers. Yet Bard wins everywhere else.
Efficient
As AI models get better at Programming, with GPT-4's sandbox a thing, Code can test itself, and AI code is transforming into less and less buggy code, and more and more optimized, secure, and efficient code.
With Bard, I give it my generic requirements, it whips up some code, I add more details, it incorporates them, and in the end, I get the code I want, and use it. Overall, this code has broken less than 10% of the time and is often fixed by reporting the error and taking the newly generated code.
Documentation
Nobody feels like writing documentation, except AI! You can give AI your code and simply ask to comment doc it or to generate documentation text for a Docsaurus page, Making documentation easier than ever!
Popularity
The final reason is, of course, the rising popularity of such tools. Github Copilot is amazing. Replit has Ghostwriter. The day IDX.dev happens, we'll get Codey. Opensource code completion like codeparrot exists. Many people are already using it, many more will probably use it heading into the future. Getting good at these tools is probably good for your future.
* this is based on March performance rates