My first blog post: On the Conversational AI Audio Pipeline(wallacecorp.ai)
wallacecorp.ai
My first blog post: On the Conversational AI Audio Pipeline
https://www.wallacecorp.ai/content
3 comments
Hey folks! Long-time lurker here. I posted my first piece of content on what I've learned building in the conversational AI space. I'd love your thoughts on what I got wrong, if/how it was useful, general comments, etc.
Thank you for taking the time.
nit, the primary text font weight is a little light for reading, or maybe it is anti-aliasing, the chars look like they are missing pixels on a 4k screen
I saw a full-duplex video model that was conversational, and it spooked me. What are your thoughts on for how this plays out for societal well-being after having experienced them more deeply than myself?
> voices have personas, like "romantic and gentle"
You might find this interesting while pondering my question
https://www.npr.org/2025/07/18/g-s1177-78041/what-to-do-when...
I saw a full-duplex video model that was conversational, and it spooked me. What are your thoughts on for how this plays out for societal well-being after having experienced them more deeply than myself?
> voices have personas, like "romantic and gentle"
You might find this interesting while pondering my question
https://www.npr.org/2025/07/18/g-s1177-78041/what-to-do-when...
Thank you for the feedback, I will increase the font weight.
I'd say not great for societal well-being. I'd say it's preying on the lonely. Even with a thinly personified system prompt, some models will suddenly start calling you "babe" and be generally hyper-flirtatious.
The most optimistic light you could shine on it is: let's make an assistant that responds like a human, tells you when you're being weird, tells you when you're wrong, etc.
Loved that TED talk, thanks for sharing!
I'd say not great for societal well-being. I'd say it's preying on the lonely. Even with a thinly personified system prompt, some models will suddenly start calling you "babe" and be generally hyper-flirtatious.
The most optimistic light you could shine on it is: let's make an assistant that responds like a human, tells you when you're being weird, tells you when you're wrong, etc.
Loved that TED talk, thanks for sharing!