UI Changes address superficial and subjective aesthetic fixes, and do not address true usability issues.
These changes interrupt workflow while providing minimal impact to overall usability. “Making things look better” does not always make things better, this is a prime example. These changes appear to try to massage the UI into looking more “human,” but - correct me if I’m wrong - the use cases for the actual users are intending to interact with AI.
A few things with negatively trending change:
As a user, parsing past conversations with GPT to reference something specific is now more cumbersome.
Primary feedback and copy actions have moved from the top of a response, to the bottom. They are only discoverable on hover.
Text is visually blended in this SMS-style conversation with GPT - making it difficult to understand the flow of conversation. GPT is BAD at understanding context, meaning it’s more of a burden on the user to control that narrative. With the more blended visual style, it’s more difficult to do that.
Usability problems exist throughout the platform. If this is a first step towards addressing those issues, I support it and understand it. However, I dont see structural changes to the UI and the components that make up the conversational interface. This is necessary to actually improve the usability.
So, if you’re making style changes before structural changes: you’re playing with fire. Failing to understand the needs and impacts of structural changes before making stylistic changes will ultimately bastardize your UI.
If your design leaders are not vocalizing this, then get new design leadership. If they are, listen to them before it’s too late.