StackOverflow had committed itself to give back cc licensed DB dumps. They stop providing them some time ago, justifying it with LLM training when asked and re-enabled them after they got a bit of a shitstorm for no longer giving back to the community. The site content itself is cc-licensed anyway.
This means, that if they continue to provide the dumps, people could just train on the dumps. If they do not continue, people can legally (given that attribution is provided) crawl the site itself.
And finally, one should think that one of the points in "giving back to the community" would be allowing the users to train LLM on their (own) data. The whole deal with that was users saying "We accept SO being a silo as long as the silo regularity provides us dumps of the data".

> Was it instructed to "add diversity" (probably not done using english phrases), or was it instructed to use real world demographics rather than assumes western Europeans are the statistical norm?

Actually neither. The tests people made show that it is more direct and terms like "black" "asian" and other nationalities were randomly inserted into the prompts. In the end it does the same as Dall-E (https://twitter.com/rzhang88/status/1549472829304741888), but when multiple people are in a scene it seems to target multiple areas of the image with different prompts.

As it uses a LLM it is possible that the actual LLM prompt was "Insert one of the following words: ..." but in the end I think one can only find out parts of the generated image prompts and they definitely contain these terms.

The image AI works well. It's the LLM instructed to add diversity that runs amok.

1) Make sure never to agree to a contract regarding the model. Don't use the hugginface version that requires you to agree to conditions. You may be bound to them otherwise even when you don't need a license for the weights.
2) The weights itself can't be copyrighted, so you can use them without a license. Without a license you aren't bound to what they write in the license.

Source for the claim that there can't be copyright on weights:
https://old.reddit.com/r/Local...

TL;DR reason why there is no copyright: Copyright implies creative work of a human. Running gradient descent on a large dataset is fully automated and does not qualify as the work of a human.

> The Chrome operating system will ultimately allow users to keep using their Windows 10 systems
That's just wrong. They can keep using the computer, but not Windows 10. ChromeOS is Linux-based.

Other than the student, spammers will not care about minor mistakes in the text due to automatic word replacement to destroy watermarks. Shuffling grammar a bit and exchanging words can be automated.

The headline is wrong is both directions.
Google won't stop third-party tracking, but third-party cookies. And third-party tracking won't be stopped by not having third-party cookies. Do yourself a favor and actually read the cookie banners. They contain a lot of things that have nothing to do with cookies. They literally ask you if they can track you across devices and uses browser fingerprints.

Trusted computing is about companies not trusting the user. The steps will be that websites can verify official browser builds. Then a website can demand a certain browser, e.g., official Chrome or official Firefox, but no Chromium, no Firefox builds by Linux distributions or yourself. Next the browsers offer APIs like "detect if adblock is installed" (Chrome much more likely than Firefox) and you cannot modify the browser to lie to websites.

Also trusted computing needs a whole chain of "We don't trust you" technology. Have a look at widevine at the highest trust level. You need secure boot, a recent CPU with an integrated TPM and at least Windows 10. No chance with Linux. Android only works if it is not rooted because when safetynet fails, apps like netflix will not show you videos.

Trusted Computing is the way to close down computer ecosystems and let vendors dictatate what you can do and prevent you from tampering with (parts of) the system. It takes away our freedom.

As said, there's no use-case that ever broke something for me in the last 10-15 years. I know that like buttons and similar stuff require third-party cookies, but I do not need like buttons. And sites manage to keep their logins working without 3rd party cookies.

Like something that can go into a dishwasher.

I browse without Thirdparty-Cookies for longer than this feature exists. If you don't need like buttons you won't notice many problems. There are very little legitimate use-cases.

No, it stopped supporting plugins. You are thinking of addons. That's a whole different thing.

No. An adblocker is the alternative.

> would you possibly settle for a deep "Why?" button that deeply explained where the ad came from, exactly how your personal information was used to select the ad for you, and even explored the motivations of the advertiser?

Why should I even want to waste my time looking up why I saw an ad? It's not like I am browsing the web for the ads.

So, why do you need a targeted ad on Facebook instead of booking a ad in a specific group? Sell your lawn mower in the gardening facebook group and you're fine.