A
“Even easy things are hard.”
Astute AI copyright observer Michael Weinberg raises some good questions about the Common Pile, an AI training dataset billed as being composed of only “openly licensed text”:
On one hand, this is an interesting effort to build a new type of training dataset that illustrates how even the “easy” parts of this process are actually hard. On the other hand, I worry that some people read “openly licensed training dataset” as the equivalent of (or very close to) “LLM free of copyright issues.”
Does an AI Dataset of Openly Licensed Works Matter?
[michaelweinberg.org]
Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.
Most Popular
Most Popular
- All the news from Nintendo’s July 2025 Direct showcase
- Epic just won its Google lawsuit again, and Android may never be the same
- Inside the LG G5’s shocking last-place finish at the 2025 TV Shootout
- 8BitDo’s wireless Nintendo 64 controller is now available
- Sharge added a rechargeable battery to its sleek retractable charger
