Genuine question: Why do we train massive models to predict things that linear regression solved well 30 years ago?

Looking at papers where complex architectures barely outperform simple baselines, if at all.

What's your favorite example of when simpler = better?

#DataScience #ML #Statistics

Question for both researchers & practitioners:

What keeps you up at night about synthetic data?
- Data quality/realism
- Distribution drift
- Privacy guarantees
- Silent failure modes?

Share your perspective!

Bonus: Drop your synthetic data horror story 👻

#AI #DataScience #Research