Closed thehunmonkgroup closed 9 months ago
We should start off with our own version of the Turing test though. I have some of my own questions that ChatGPT fails at which could prove to the public that ours is more human in nature.
I have some ideas as well for further testing that we should discuss at some point past MVP stages. Im thinking a more cognitive approach for testing outputs. Having an AI analyze the output of ours to detect anything unusual. This would allow quick testing iterations and output suggestions. A more automated advanced turing test that also utilizes its own internal cognitive architecture, but built mainly for unnatrual/rude/etc speech detection. GPTs can pick up oddities to some degree if they are tasked with that purpose and if we combine that with other memory modules this could be interesting.
We're not using the PITCH structure anymore, closing.
PITCH: Testing framework
Problem
We need a testing framework.
Appetite
3 seconds
Solution
pytest
Rabbit Holes
Message delivery by carrier pigeon.
No-gos