Angelina YangFounder and Host, TwoSetAI · AI builder for visibility and discoverability

Can we eliminate humans from AI workflows?

24 Dec 2025 · 1 min

Watch on YouTube

0:45

A speaker (Angelina Yang) discusses whether AI can eventually remove humans from feedback/annotation loops. She argues it may be possible in the distant future—especially with pervasive sensing (wearables, ambient devices)—but not soon because current models still fail on important real-world tasks (e.g., identifying scar tissue on brain scans or dental lesions), so reliability remains a barrier.

Chapters

Can we eliminate human in a loop one day? I think it's possible sometime in somewhat distant future, we can capture the human input without the human having to go into a specific place to give that input. Especially with wearables and a lot of devices around us, we can capture that input significantly higher without the human having to go and actually describe it. So I think it could be possible, but I don't think it's anytime soon because I'm looking at some of the machine outputs, and like you said, the demos are very impressive. But you look in reality the models, for example, identifying scar tissue on brain scans, and it's not good. Or lesions on teeth, and it's not good, it's not reliable. So. So I think it's going to take some time.