@WisprFlow I did file a support request the first time I saw intent inversion. I got back a basic confirmation that model changes had made rewriting “more aggressive”. Since then I have only seen the product get worse with larger and more semantically questionable rewrites.
I am an indirect investor in @WisprFlow through a VC firm and I have enjoyed using the product for voice input ... until recently. Multiple times I've seen rewriting prompts that users cannot customize invert intent. "do not" became "do", "can you assure" became "I assure". In an agentic context, this is dangerous. There is no such thing as one perfect voice input setting for all possible contexts. Beyond @superwhisper, who else provides smart/customizable voice input for agentic contexts?
After three years of exceptional growth and profitability, Swoop is independent again. https://t.co/q9B7Whvrlh Another quiet #unicorn created in Boston.
When it comes to copyright, LLM vendors will try to shift all risk to their customers. They'll argue that the LLM itself is transformative change, but that LLM outputs may not be. #GenerativeAI#chatgpt4
Learn how our strategic partnership with @Get_Writer is putting our employees in the driver seat for creating strategic and innovative content, and how our alignment will shape the future of healthcare communications and engagement.
https://t.co/4e8rqEOkqu
Excited about our strategic partnership with @Get_Writer. Differently from other leaders working in AI-assisted writing, @may_habib understands the power of platforms. https://t.co/VUjAvDc74j #GenAI
@durjoy What's old is new again: well-designed traditional lace-up leather shoes have a reinforced section over the toes to prevent creasing and allow for that mirror polish look.