@thsottiaux I have a great idea - could GPT do voice cloning to define its own voice? For example, we have Breeze TTS, so could we use its voice API in GPT-4 or other voice features to create a custom voice profile?
@thsottiaux For image generation models, there's a serious issue with false positive detections on guardrails. This problem is hard to reproduce, but I have clear records of it
@thsottiaux Core issue: This has been a problem for a long time now, whether on the web or in apps. When there's a long conversation window open, text display and output become extremely laggy - it seems like it freezes up for a moment, then suddenly spits out large chunks of text
@thsottiaux Another idea is, similar to Claude's interactive panels, could we work together with GPT in real time? Like I could directly select areas on the interactive panel and tell GPT how to improve that feature, or do it by using Ctrl+select multiple items
@Da7_Tech Da7em, I think I get where you're coming from - you just think Anthropic's models are better, but you don't agree with A's approach or way of doing things