I’m a GPT Plus user and novice OpenAI API user.
I’ve run into a disconnect with the current state of what’s available and I wanted to share and see if it might get others to share what they’re doing in this vein.
I’m an expert coder, but I’ve taken the approach that GenAI ought to be able to do a lot of the heavy lifting with coding, especially against the API to use it. But this is really not the case given the following things I run into when I try to use it on a project building a custom chat based website:
You can turn on Advanced Data Analysis and upload files, images, etc. This can sometimes result in great help. For instance, you can download the zip available on any GitHub repo and upload that and the analysis can often be quite helpful in advancing the state of code in that repo.
But sadly, currently, you can’t combine (1) with searching the web. Briefly you could choose Browse with Bing but not (1) at the same time. And now it seems you can’t choose Browsing at all (at least not in an OpenAI, non-plugin, option). Choosing to Browse allowed you to incorporate search terms that effectively broadened ChatGPT to know about the current API, without this, any advice you’re getting on API coding is starkly and hopelessly out of date to the point of being unhelpful due to the data cutoff.
Ask Bard. This can be helpful since Bard at least can incorporate current web data about the OpenAI APIs (ironic, I know). But Bard is really weak sauce for helping with coding in general so far.
So has anyone else taken a similar tack and found a groove that works very well?