AI News — what is changing around ideas worth trying?

I wanted to start a broader ideas worth trying thread around AI News.

A useful angle is evaluation, but it also connects with performance. Different tools and workflows can produce very different results even when the goal is similar.

It would be interesting to compare practical examples, experiments, setups, or lessons that show where AI News is genuinely useful and where it still needs improvement.
 
On AI News — what is changing around ideas worth trying?, the part about news and around stands out to me. I would test one small workflow first, keep the variables consistent, and compare the result before changing several things at once. That should make it easier to tell whether the change is actually helping.