@along_the_road what’s the alternative scenario here?
You could push to remove some public information from common crawl. How do you identify what public data is _unintentionally_ public?
Assume we solve that problem. Now the open datasets and models developed on them are weaker. They’re specifically weaker at identifying children as things that exist in the world. Do we want that? What if it reduces the performance of cars’ emergency breaking systems? CSAM filters? Family photo organization?
“These were mostly family photos uploaded to personal and parenting blogs […] as well as stills from YouTube videos"
So… people posted photos of their kids on public websites, common crawl scraped them, LAION-5B cleaned it up for training, and now there are models. This doesn’t seem evil to me… digital commons working as intended.
If anyone is surprised, the fault lies with the UX around “private URL” sharing, not devs using Common Crawl
@abaci @mojo 100% worth it for me.
I switched from Google to #DDG to #Kagi. Even if the results were exactly the same, Kagi is FAST. The whole experience is very snappy.
Beyond that:
- All your DDG bangs work, and you can add custom bangs
- They have some neat AI summarization features.
- You can manually boost/penalize/block domains
- Lenses focus your search on particular kinds of sites
@esaru @bmaxv @technology concur that this reduces privacy for users of Jitsi’s hosted service. It also has some concrete benefits for Jitsi - they get to outsource account validation and security. Perhaps they were struggling to contain abuse.
I certainly think that a Generative AI model is a more significant harm to the artist, because it impacts future, novel work in addition to already-published work.
However in both cases the key issue is a lack of clear & enforceable licensing on the published image. We retreat to asking “is this fair use?” and watching for new Library of Congress guidance. We should do better.
For example, image search has been contentious for very similar reasons.
@donuts would you please share your thinking?
I certainly agree that you can see the current wave of Generative AI development as “scraping and stealing people’s art.” But it’s not clear to me why crawling the web and publishing the work as a model is more problematic than publishing crawl results through a search engine.
👉🏼“AI stole my book/art” is not that different from “show me in the search results, but only enough that people click through to my page”
👉🏼 “AI is taking all the jobs” is not that different from “you outsourced all the jobs overseas”
👉🏼 “the AI lied to me!” is not that different from “that twitter handle lied about me!”
The main difference is scale, speed and cost. Things continue to speed up, social norms and regulations fall behind faster. #ai
@SSUPII @throws_lemy @technology concur.
Skynet is a red herring.
The real issue is that #AI is putting more stress on long-standing problems we haven’t solved well. Good opportunity to think carefully about how we want to distribute the costs and benefits of knowledge work in our society.
@SemioticStandard Kagi. I used DDG for a long time, and Kagi is strictly better. Specifically, it’s very snappy and I trust the privacy guarantees even more since I’m a paying customer.
@kent_eh exactly.
The alternative is “if you want your content to be private, share it privately.”
If you transmit your content to anyone who sends you a GET request, you lose control of that content. The recipient has the bits.
It would be nice to extend the core technology to better reflect your intent. Perhaps embedding license metadata in the images, the way LICENSE.txt travels with source code. That’s still quite weak, as we saw with Do Not Track.