The approach of LLMs without some sort of symbolic reasoning layer aren’t actually able to hold a model of what their context is and their relationships. They predict the next token, but fall apart when you change the numbers in a problem or add some negation to the prompt.

Awesome for protein research, summarization, speech recognition, speech generation, deep fakes, spam creation, RAG document summary, brainstorming, content classification, etc. I don’t even think we’ve found all the patterns they’d be great at predicting.

There are tons of great uses, but just throwing more data, memory, compute, and power at transformers is likely to hit a wall without new models. All the AGI hype is a bit overblown. That’s not from me that’s Noam Chomsky https://youtu.be/axuGfh4UR9Q?t=9271.

“Freedom of choice Is what you got Freedom from choice Is what you want” -Devo

They’re also adding a lot more incompatible text formatting and shit to keep Android incomplete with their real chat protocol. Gotta keep those teens bullying Android users. Also E2E encryption would be nice, but the EU didn’t force them to do that.

Still great because MMS is garbage and ruins photo and video quality.

They could offer a way to download a copy and steganographically tag it to hell with your id so that they know if you distribute it. You can “loan it out” by letting friends stream off your Plex or whatever. If you start selling that streaming service or it shows up in torrents, it has your ID on it.

Boom, you own it forever and you’re incentivized not to over share.

Or you know sell DRM free versions and let people do whatever, but that probably has a snowballs chance in hell.