Perhaps a manifest index/list of training data and learning method should be mandatory for “public” use of any model, sort of like food ingredient package labeling. The list of texts used and any weighting seems important to facilitate a public understanding / debate of model expectations and behavior.
This is not yet a given, I expect rule-making or copyright law expansion if courts rule otherwise.
Things also change. In the 80s lawyers carefully suggested white pages phone listings came under copyright but "sweat of the brow" was rejected by US supreme court in Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991)
Thank you. This seems a good start to a desperately needed credible "externalities report" to ground political and economic discussion.
The flood of anecdotal thumping seems to distract and undermine deliberation. The sad lack of peer reviewed data and impact analysis hinders thoughtful policymaking.
What is the core value proposition of twitter? It seems simple, short form informations delivery (eg “follow me on twitter”) is central. The “likes”, retweets, and follower counts are secondary but perhaps important metrics that drive engagement. A simple link plus QR for paper is needed. Moderation seems a challenge. Are there simpler alternatives to mastodon? is federation possible? Why is RSS not used for this?
I'm not so sure about that. Bluesky by large migrated from twitter aiming to escape the likes of Elon's politics, but they still brought over some of the most toxic cancel/brigade culture that twitter had to offer.
Not suggesting all of bluesky is bad, but it sure ain't the most pleasant crowd in it's own right either.
Bluesky, like Mastodon, is built on an open network on open source code. Run your own data server and moderation services that cater to your needs. Many are already doing it: Northsky, Eurosky, Blacksky, etc. If you want to live in a safe space isolated from the mainnet and its allegedly toxic users, you can do that too: Roundabout is built on ATProto but run on its own network, also Truth Social and Threads are forked versions of Mastodon but federated with nobody (Threads used to have opt-in Mastodon federation before Facebook decided they no longer want users to be able to do that).
Mastodon isn't all that complicated. But, obviously it's not where people went when people fled Xitter, by and large. The funny people and journalists mostly went to Bluesky and the normal people mostly went to Threads because they already had an Instagram account. The Nazis and child pornographers and the people who were comfortable being around Nazis and child pornographers remained on Xitter. But, a few nerds went to the fediverse and we are generally having a nice time there.
RSS is not a read-write medium. You can't realistically carry on a semi-realtime conversation over RSS.
Clearly they approve or they'd at least stop advertising their twitter handles on their websites and handouts. Most brands cannot ignore such a vast network effects and have made the decision to continue despite the changes.
How is product liability relevant here? If an AI company makes a model available, someone uses it, and it does something bad, who is at fault? The user, the data center, or the one who made the model?
If we want open weight models with a warranty disclaimer, then the user would be held liable. If we want to hold AI companies at least partially liable, that
seems a different, centralized model.
I think it’s more complex than that? What did it do? What did the user prompt it to do? What did the company train it to do? What did the harmed party do? There’s possibility for negligence at every level.
If you train a dog, rent it to someone, and the dog bites a third person, who is responsible? I think that’s the best analogue here.
All parties could share fault in that scenario, depending on what actually happened.
