Human or Machine? Low-Latency Audio Detection of Humans at Scale

I want to say as an engineer that cold calls aren't effective but I feel like it's just due to the numbers of how many calls you can make per hour.

If you can increase the number of dials like in this post it really makes more sense from a time perspective. We also know that regulations on phone calls are going nowhere anytime soon lol.

Interesting that data quality matters so much than data scale. kinda tracks with some of the news I've heard out of the foundation labs where they've stopped scraping the whole web and instead focused on small, high quality, curated data sources

most interesting thing was how the model already beats human performance without an AI. I suspect in lots of low-latency applications AI will outperform humans very rapidly just due to raw compute GPUs have