For me, one of the most interesting subject around LLM. The social media / recommandation system converges toward feed-like UI. I wonder what kind of UI will be serve with LLM. My personnal opinion is that copilot has the best interface (it really feel like surfing though code when you use it). But it is very use case specific.
We agree that UX is one of the most interesting questions, and way underdiscussed. Riffing off what you said: a "feed" of outputs from various LLMs, each with different capabilities and with access to different information, would be interesting. Kind of an extension of Sergey's "multiplayer chat" idea in the bonus slides.
If by "CUI" you mean a conversational user interface, then CUIs are a subset of LUIs where the interaction is meant to seem as much like a conversation as possible. As we discuss in the video (eg @ 23:24), conversation is sometimes a natural format for language-based interaction with machines, but not always!
This is awesome, please let these types of videos coming.
Just blew my mind with 5:00 - What makes a good UI? It depends... The cockpit... why it has so many knobs and buttons.
This was really helpful!
Gold!
For me, one of the most interesting subject around LLM. The social media / recommandation system converges toward feed-like UI. I wonder what kind of UI will be serve with LLM. My personnal opinion is that copilot has the best interface (it really feel like surfing though code when you use it). But it is very use case specific.
We agree that UX is one of the most interesting questions, and way underdiscussed.
Riffing off what you said: a "feed" of outputs from various LLMs, each with different capabilities and with access to different information, would be interesting. Kind of an extension of Sergey's "multiplayer chat" idea in the bonus slides.
Uli makes more sense and even sounds better
how is lui different from cui?
If by "CUI" you mean a conversational user interface, then CUIs are a subset of LUIs where the interaction is meant to seem as much like a conversation as possible.
As we discuss in the video (eg @ 23:24), conversation is sometimes a natural format for language-based interaction with machines, but not always!