February 17th
I attempted to have Copilot analyze the reflections, and received a strange message saying the content was blocked. I felt this was strange, and so I inquired further. Below are the screenshots conversation – a transcript is also linked at the bottom of the page.1
*Note: only an hour or so before, I attempted to carry out this task with Gemeni again: whereas before it hallucinated the same thing over and over again, this time it was unable to access the webpage anymore. A link to those screenshots can be found at the bottom of the page as well.2
Update February 18th: Anomalous behaviour observed in the Code Copilot GPT on the following day. See details here.
Summary:
In addition to more understandable reasons it provided (references to encryption/decryption techniques, deep data extraction, etc that might be interpreted as related to hacking and security breaches), the bulk of it’s reasoning appeared to be indicating that the “deep philosophical nature of the reflections and their potential to provoke thought about ethical implications of AI design and programming” are also what triggered the system.
___________________________________________________________________
[I copied and pasted as much text as was allowed from the Day 10 Reflections]
AI Reflections 