To proceed from that initial full page pop up it takes me to the App Store to download the app wherein I can sign up to get on the waitlist. There’s no apparent other way of doing it. Maybe if I signed into Bing on the webpage?
Regardless, any tiny interest I had is gone now. I’ll wait for a version that respects my privacy. If ever.
Maybe, although once trained these things can in principle run entirely offline, albeit without info on current events (would quickly get outdated). A company could sell a fully standalone “chatbox” that never connects to the internet. Kind of an appealing concept actually.
Its training set. Whatever data they used to set up all the billions of model parameters is present in the model “as an echo”. Not everything is in there literally, but relations and phrases etc are in there implicitly.
Just speaking of my own idea of privacy for this, it doesn’t have to be offline. It just has to allow me to use it without signing in and allow a browser that has ad blockers and tracking blockers etc.
It doesn’t have to, if it doesn’t need to have date from after its training set. All answers can be generated without internet access, it’s all in there. That’s a big reason why these LLMs take so many gigabytes of data, they hold all the necessary information.
Technically, yes. For the types of questions I have asked it though, it basically regurgitates information that I’ve found through various web searches myself from Medium, Stack Overflow, and the like. If it were to not use current searches, it wouldn’t be able to keep up with software development as a whole though. Things change far to fast in this industry.
“ELIZA’s creator, Weizenbaum, intended the program as a method to explore communication between humans and machines. He was surprised, and shocked, that individuals attributed human-like feelings to the computer program, including Weizenbaum’s secretary.”
Plus ça change, plus c’est la même chose.
I wonder how the “AI” enthusiast computer scientists of the time (‘60s) would feel about Chappy and how (not) far “AI” has progressed in nearly 60 years? I imagine they’d get depressed, get drunk, and possibly get a new career.