AI Lab · by Minh Nguyen · 2026-10-17

Synthetic data for training: genius or model collapse in slow motion?

Seeing more teams train on model-generated data. Quality filtering seems to be the whole game. Thoughts?

23 sparks · 15 comments

Comments

My manager read one article on synthetic data and now it's our whole roadmap lol

send help

I've lived this exact meeting

We tried synthetic data last quarter, worked great for small stuff, got messy at scale

Ananya Rao · 1 likes

Good point, didn't think of that

Finn Brown · 5 likes

Is anyone else tired of the hype around synthetic data, feels like every other post is about it.

Good point, didn't think of that

An Pham · 4 likes

As someone who does this for a living, the biggest cost with synthetic data is the stuff nobody puts in the demo

wait until you see the invoice

The beginner friendly path for synthetic data is way better than a year ago, that part is underrated!

lol my cat has stronger opinions on it than I do

Aoi Ito · 1 likes

cost vs benefit matters here, we ran numbers on synthetic data and the payoff came later than expected!

send help

Would love to see actual data on synthetic data instead of vibes. Anyone have links

Fair enough