I have a Bluetooth dongle headset-mic. Probably for the same buffer reason, it constantly breaks audio when I have multiple audios in/outs running.

The only consistent fix is switching to another audio driver and playing a video on YouTube while I switch it back.

Reddit sold Google the rights to train AI models using their (our?) data a year ago. Any trained model will recognize the data it was trained on and rank it highly.

https://finance.yahoo.com/news/exclusive-reddit-ai-content-licensing-041012378.html

You have to aggressively purge the current chat and give it more abstract references for context. With enough context it can rewrite some logic loops, maybe start a design pattern. You just have to aggressively check the changes.

It’s baked into the training. It’s not a simple thing to take it out. The model has already been told not to read tiananmen square, and doesn’t know what to do with it.

How much does sleep and dreaming contribute to this? Have you looked into how humans dream compared to other animals? Any papers to point to?

Pepe the frog is from the Boys Club, a comic or children’s book; I don’t know, it just had a couple talking animals. The first picture of him with the feels good man caption was very silly and took off from there. Unfortunately pepe became a hate symbol after 15 years.

What helps me is asking Copilot to try doing it. When it spits out garbage; if it hits any of my targets I can tune it from there.

What else is reasonable to use to uniquely identify users? A username they’ll forget? A phone number maybe? But who wants to give their phone number to some company? We could use SSN like Korea, but that’s way too far for a typical user.

Testing.

We are still workshopping the kiss one. That wolf is still working it’s way in.