How do I log for myself and others queries that I’d like to have better (or maybe just different1) answers to, but perhaps are not urgent enough to continue persisting in-the-moment in pursuing other avenues?
Example:
- Is there a quick way to verify these sorts of claims: “Enter your OpenAI API key here (the key is not stored by this app)” “This is a client-side web app, your API key will not be shared with us.”
- Why don’t these websites or API providers directly support verification?
I received very helpful replies in the I School’s Slack channel for #no-stupid-questions, but it seems would be too much work for myself or too much of an ask of others to push it along at this time.
Aside: In this case, you can definitely look at the network traffic (and check local storage), which might be a reliable guide for some websites (those that are client-side apps). But even then the code could change. I’m imagining instead an an icon surrounding an area of the webpage that indicates it itself is sent directly to some claimed server. Or a link to a certification page. I’m not sure how often this sort of use case comes up. Though there are a lot of examples where people present exploratory tools for the OpenAI APIs (ex. ChainForge which I trusted without any of this verification). It has also been tough for me to effectively word my question (ex. here on Perplexity)
Update: As I was writing the above I received another helpful response, a reply to my tweet, which re-affirmed my frustration that led my asking the question, and gave me some new language.
For some such questions I suppose it is that I want the world to be different. We can keep asking some questions, as little gadflies perhaps.↩︎