Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI

Amazon Finally Releases Its Own AI-Powered Image Generator (techcrunch.com) 16

During a keynote at its re:Invent conference today, Amazon debuted the Titan Image Generator, which can create new images or customize existing images via a text description. It's now available in preview for AWS customers on Bedrock, Amazon's AI development platform. TechCrunch reports: Amazon says that Titan Image Generator was trained on a "diverse set of datasets" across a "broad range of domains" and can be optionally fine-tuned on custom datasets, and includes built-in mitigations for toxicity and bias. (Barring testing, the jury's out on just how effective those mitigations are, of course.) The company declined to say exactly where those datasets came from however -- and whether it obtained permission from or is compensating all the creators of the images used to train Titan Image Generator. [...] Sivasubramanian did claim onstage, however, that Amazon will protect customers accused of violating copyright with images generated by Titan Image Generator -- in keeping with its AI indemnification policy. That's surely music to the ears of AWS customers worried about regurgitation, or when a generative model spits out a mirror copy of a training example.

Images created with Titan Image Generator will also come with a "tamper-resistant" invisible watermark by default -- an attempt to mitigate the spread of AI-generated misinformation and abuse imagery, Sivasubramanian says. (Deepfakes from the Gaza war and AI-generated child abuse images are the latest illustrations of how major the threat's become.) It's not clear exactly what sort of watermarking technique Amazon's using and which tools beyond Amazon's own API will be able to detect it; we've reached out to Amazon for clarification. Sivasubramanian noted watermarks are a part of the voluntary commitment around AI that Amazon signed with the White House in July.

This discussion has been archived. No new comments can be posted.

Amazon Finally Releases Its Own AI-Powered Image Generator

Comments Filter:
  • by Mr.Intel ( 165870 ) <mrintel173 AT yahoo DOT com> on Wednesday November 29, 2023 @07:03PM (#64042379) Homepage Journal

    I just checked and a single model plan is $10k per month. Even if AI were my business, no thanks...

    • by zshXx ( 7123425 )
      You are obviously not the target customer. Also, amazon tends to launch stuff at higher cost and then reduce prices as costs goes down with scale (see s3 cost history for example)
      • by tlhIngan ( 30335 )

        I'm sure a graphic artist is cheaper than $10K/month. Remember, that's $120k/year.

        • They will be, but they won't be able to produce as much output.

          If you have a whole art department you might be able to cut a significant percentage of your staff if you have the rest of them use a service like this.

          i.e. It's not for you.

  • by MIPSPro ( 10156657 ) on Wednesday November 29, 2023 @07:08PM (#64042399)
    Imagine going to school and honing your art skills and these things come along... ooof. Nobody who needs custom art is going to ignore it. It's too effective and too cheap and the results are usually amazing. Yes, someone might be needed to resize or change the aspect ratio maybe upload it somewhere or even make a few manual edits (not really needed since most of the image bots also have edit modes). However, the skill level needed is so much lower than someone who can do photorealistic digital painting by hand. Artists and anyone who writes a lot of text/copy are about to get some pink slips. I don't think any amount of strikes or labor action is going to turn the clock back, either.
    • by alvinrod ( 889928 ) on Wednesday November 29, 2023 @07:55PM (#64042511)
      The most talented or skilled artists will still have reliable business. The AI generated works can't be copyrighted and some people will absolutely want something to which they can own and enforce copyright on, hence the need to employ a human. The other side of all of this is that having had an actual person create the work makes it far more difficult for someone else to sue over supposed infringement on the part of an AI.

      The colleges that offer these degrees will probably be hit hard. There's not much reason to offer or keep around degree programs that have little demand.
      • by HiThere ( 15173 )

        IIUC, turning the output of the AI into something that is copyrightable requires only a trivial amount of editing. Perhaps that will change.

      • by tlhIngan ( 30335 )

        some people will absolutely want something to which they can own and enforce copyright on

        That's a big point.

        If a prompt can produce some art, someone else can reproduce it with the same prompt as well. The only reason things come out different is that the AI generates a few random numbers to perturb the model so it doesn't generate the same image over and over again.

        But in general, the same prompts should generate the same images. If people use AI generated backgrounds for stuff, you're going to see it bei

    • You'll want a human for things that need to follow a design and remain consistent. Not all work requires that, so there will be losses.

      That random animal on your programming textbook is a one-off. Someone made that and maybe receives royalties from it being there. It doesn't reference anything; it has nothing to do with the contents of the book. It's just meant to be eye-catching - which AI does perfectly well today. That artist's money may dry up. Meanwhile, for architectural renderings, mere approximates

      • You'll want a human for things that need to follow a design and remain consistent.

        Alas, this is actually what AI does really well, especially the newer AI systems. If you tell AI to create a new work that resembles or is based on an existing work, the results are amazingly consistent. Engineering is a different issue, of course, but aesthetic consistency isn't actually that hard.

        More and more often, I see AI-generated art that has consistent use of perspective, lighting, and shadows. It's obvious that the AI systems are modifying existing images and probably heavily infringing on copy

    • The pay for digital artists working for others is already abysmal. Truck drivers literally make more than unionized illustrators.

      And to make matters worse, the sort of image manipulations AI is designed to replace are the soul-crushing mindless illustrations that most artists do only because it's steady work. Anything edgy, thought provoking, or opinionated/editorial will still require a human being.

      So, from the perspective of an actual, working artist, this only takes away the jobs we wouldn't want

      • And if you know anything about the art market, no one wants to buy something mechanically generated.

        I really don't know much, but I instinctively want to believe that real people want art created in traditional ways by real human beings. The thing is, I just don't know how much of the money artists make overall is generated in by the kind of fine art you and I are talking about versus commercial art modalities like graphic design or 3D modeling & design. The latter I'm guessing will take a harder hit.

  • It was clearly something the world was waiting for, we were sitting here with bated breath wondering when Amazon would finally do it.

    • Impatient, entitled pricks say "finally!" whenever something good happens.

      It's a real warning sign to avoid these people.

Congratulations! You are the one-millionth user to log into our system. If there's anything special we can do for you, anything at all, don't hesitate to ask!

Working...