Comment on Artists lose first copyright battle in the fight against AI-generated images

Thekingoflorda@lemmy.world ⁨1⁩ ⁨year⁩ ago

• Shayeta@feddit.de ⁨1⁩ ⁨year⁩ ago

  I propose the following solution: Everyone is free to use any publicly available data to train AI. Any data generated by AI is automatically in the public domain and cannot be copyrighted.

  source

  • kogasa@programming.dev ⁨1⁩ ⁨year⁩ ago

    No, that doesn’t work at all. That gives all the power to people with billions of dollars to train and run the best proprietary models, at the expense of the people who created the data.

    source

    • Dkarma@lemmy.world ⁨1⁩ ⁨year⁩ ago

      Welcome to the internet where you can literally experience everything that AI was trained on right now for free.

      That’s how the internet works. You put your stuff out there and people and software experience it.

      THIS IS NOTHING NEW

      source

      • kogasa@programming.dev ⁨1⁩ ⁨year⁩ ago

        Yeah? Where did you get the impression I’m in favor of a completely unregulated marketplace where corporations are free to harvest everything from everyone for free?

        source

        • -> View More Comments
  • Dkarma@lemmy.world ⁨1⁩ ⁨year⁩ ago

    That’s literally how it works now. There are no legal restrictions on training ais and courts have rules ai generated works are not copyrightable.

    source
• ParsnipWitch@feddit.de ⁨1⁩ ⁨year⁩ ago

  They will also very likely make less people pursue drawing, (graphic) design and painting as a skill.

  source

  • kromem@lemmy.world ⁨1⁩ ⁨year⁩ ago

    The same was said about Photoshop decades ago.

    Also reminds me of this.

    source

    • ParsnipWitch@feddit.de ⁨1⁩ ⁨year⁩ ago

      To compare this with what happened when Photoshop came up is not a fitting comparison at all.

      Digital art pushed more people into learning how to draw (with traditional and digital media) leading to a surge in courses, books, workshops etc. Digital drawing made it affordable for more people to get into drawing. And it also encouraged them to learn traditional drawing to improve their skills and expand their portfolio.

      This is not comparable to what is happening now with AI image generators.

      Photography versus traditional media realism paintings like still lifes and portraits is a better analogy. But photography only touched one specific area of drawing/painting not all of it. And in this case it really did lead to a skill becoming incredibly rare.

      source

      • kromem@lemmy.world ⁨1⁩ ⁨year⁩ ago

        I have a feeling you may not have seen the difference between how digital artists are using generative AI in workflows and users just creating generative AI images from a prompt with no additional work.

        The idea that digital art capabilities are going to disappear because diffusion models can generate digital art may be a bit too binary of a consideration from the reality.

        You are right, that photography did replace a lot of still life work, such as in magazine ads, etc. But it only reduced the market for the skill set, and many people still produce ‘photorealistic’ art today.

        I’d agree the market for drawing and especially prototyping is going to be made more efficient, but I’m skeptical it’s going to be entirely reduced as you put forward by your analogy.

        Making something like a movie poster already went from making 10 mock-ups by hand from a team of artists to making a hundred mock-ups compositing using asset libraries and moving forward will likely end up in a place where there’s 1,000-10,000 versions which are each run though virtual focus groups to create a selection set for the client.

        But the final product will absolutely still involve digital artists, and if anything the component that’s mostly being replaced is the asset library, along with around a 10x or more time savings on an individual artist’s generation.

        That will either result in a 10x increase in variations or 1/10th the staffing or somewhere in between, but as mentioned parallel advances in AI mean that significantly increased output is very likely going to have significantly increased return, so you may even ultimately see slightly larger digital art teams from today as time goes on.

        There’s a bizarre assumption that modern labor output represents a demand cap and as such efficiency in supply means less people making the same amount of things.

        That’s almost never historically been the case during industrialization and unless the role becomes entirely obsolete, scaling up productivity with a new tool will bias towards increased output not decreased suppliers - outside of decreased demand for suppliers who have eschewed the new technology and efficiencies.

        source

        • -> View More Comments
• commie@lemmy.dbzer0.com ⁨1⁩ ⁨year⁩ ago

  they will likely destroy the livelyhood of thousands of people and probably give more power to big corporations.

  i doubt it.

  source
• nutsack@lemmy.world ⁨1⁩ ⁨year⁩ ago

  in the long run it is not good for humanity. it leads to sameness. there’s less incentive to make something new when an ai trained on existing work can make a thing that you know already works for most people.

  source