Chat GPT-4 can't remember it's writing a FireFox add-on not a Chrome Extension.

It's like the most amazing coder ever, but always half-drunk, and completely confident, and always. Here's how almost every single response started after the first....

• Apologies for the incomplete response.
• Apologies for the confusion. The Express server I provided earlier ...
• I apologize for the inconvenience. After reviewing the code, I've noticed some inconsistencies in the code
• I apologize for the confusion. It appears that the context menu was removed due to a typo in the content.js file.
• I apologize for the confusion. To make the changes you requested, follow the steps below:
• Apologies for the confusion, and thank you for providing the additional information. Here's an updated implementation that should resolve the issues:
• I apologize for the confusion. Here's an updated solution that should display the response in the popup window and clear the input field on submit. Additionally, I added an indicator that shows the addon is thinking.
• Apologies for the confusion, and thank you for the clarification. Based on your requirement, you can make the following changes:
• Apologies for the confusion. You are correct that you cannot trigger the reviseMyComment() function in the content script without sending a message from the background script.
• My apologies for the confusion. The error you are encountering is because the sendToOpenAI() function is not available in the content script content.js
• Apologies for the confusion. I made an error in my previous response.

Data In -> Data Out

I don't think they're having any religion re-enforced on them, but think of it this way:

You know how mad some super religious extremists get when you even use words that imply gay people are normal, or trans people exist (and aren't just mentally ill),

Imagine if people got as mad every time someone said "oh my god" or "JFC" etc. This imaginary group would be claiming "micro-reglious-agression" all. day. long.

I think that Abrahamic religious are soooo ubiquitous in the training set that the AI is likely to just go with the flow on it.

Great point! I
actually really enjoy AIExplained's videos on this. There are a bunch of different ways ways to measure 'consciousness' and many of them are passed by GPT4, which really just means we need new tests / definitions for AI models.

That's helpful. I built a music rec prompt / prompt generator, and I had to use Spotify's API and some fuzzy matching to double check if it hallucinated.