Synthetic Media & Generative AI News and Discussions

[weatheriscool]

YouTube to roll out labels for 'realistic' AI-generated content

Source: CNN

YouTube will soon require disclosures on videos that contain content generated by artificial intelligence and could mislead viewers.

The Google-owned platform has long prohibited technically manipulated content that could mislead viewers and “pose a serious risk of egregious harm.” But with a new policy update announced Tuesday, YouTube will now require creators to add labels when they upload content that includes “manipulated or synthetic content that is realistic, including using AI tools.”

The policy is meant to help prevent users from being confused by synthetic content amid a proliferation of new, consumer-facing generative AI tools that make it quick and easy to create compelling text, images, video and audio that can often be hard to distinguish from the real thing.

The new labels will be required only on AI-generated or otherwise synthetic content that is “realistic.” That includes, for example, videos that “realistically depict an event that never happened, or content showing someone saying or doing something they didn’t actually do,” YouTube Vice Presidents of Product Management Jennifer Flannery O’Connor and Emily Moxley said in a blog post.

-snip-

Read more: https://www.cnn.com/2023/11/14/tech/you ... index.html

[Powers]

^

[wjfox]

[spryfusion]

[weatheriscool]

A peek into the future of visual data interpretation: A framework for assessing generative AI's efficacy

https://techxplore.com/news/2023-11-pee ... ative.html
by Nathi Magubane, University of Pennsylvania

In the last year, large language models (LLMs) have come into prominence for boasting a suite of ever-expanding capabilities including text generation, image production, and, more recently, highly descriptive image analysis. The integration of artificial intelligence (AI) into image analysis represents a significant shift in how people understand and interact with visual data, a task that historically has been reliant on vision to see and knowledge to contextualize.

Now, new AI tools present a paradigm that allows more and more people to interact with images by generating descriptions that could not only assist the visually impaired but could also inform lay audiences about the contents of a scientific figure.

Associate professor Chris Callison-Burch, assistant professor Andrew Head and Ph.D. candidate Alyssa Hwang of the Department of Computer and Information Science in the School of Engineering and Applied Science at the University of Pennsylvania have developed a framework for gauging the efficacy of vision-based AI features by conducting a battery of tests on OpenAI's ChatGPT-Vision ahead of its release earlier this month.

[wjfox]

[Yuli Ban]

[caltrek]

Artificial Intelligence: Stemming the Tide of Fake Facts
November 21, 2023


Introduction:

(Eurekalert) AI expert Stefan Feuerriegel looks at the danger of automated fake news campaigns and points out ways that we can counter them.

Pretty much at the push of a button, AI tools are now able to generate convincing texts, images, voices, and even videos. Can we still trust our eyes, our ears, and our common sense going forward?

Stefan Feuerriegel: The picture of Pope Francis that went viral this past spring illustrates just how convincing such artificially generated content is. Ordinary people were scarcely equipped to recognize whether the picture was real or not. Of course, gifted artists were able to hoodwink the public and experts with counterfeit paintings in bygone days. But in the case of the picture of the pope, it was one person who quickly spun out dozens of versions of this photo. This highlighted what’s possible with AI today – producing these images is easy and you don’t need special skills. At the moment, AI programs still don’t get some individual details quite right: sometimes the background is a bit off, and AI has a habit of rendering hands with too many or too few fingers. However, engineers are developing and optimizing the technology at a rapid rate. Over the next few months, the improvements will be apparent.

What makes AI-generated fake news particularly dangerous?

Unfortunately, not only are fakes with error-free text and authentic-looking pictures already very convincing, but AI makes it possible to personalize misinformation and tailor it to the religion, gender, or political convictions of individual consumers and sow rage or hatred in each target group. You can build a bot that no longer just posts a message, but writes to people personally on Facebook or Twitter and responds to replies and carries on conversations. This extends even to fake calls, where scammers or other bad actors deliberately generate a voice that sounds like a person’s family member.

Read more here: https://www.eurekalert.org/news-releases/1008816

[raklian]

This is cool! See the thread for the rest of the parade. The Xenomorph one is my favorite!

[Tadasuke]

So far, generative AIs caused me to block many accounts on Twitter, which post a lot of images from such AIs. I don't like them. Following 200 good human artists is enough for me. I don't need a bazillion AI-generated repetitive pictures. I just don't. However, I have been thinking for a long time about an eventual future, when nearly every new comic page looks like the best looking ones today (without sacrificing author's time, money or health) and everything is as long and detailed as someone likes.

And also spambots. Yes, spambots. There are clearly more of them now. They like my comments and even post comments to my comments. Everything connected to sex, even when I write about CPUs, even when I write negatively about porn or about generative AI. Not the future I wanted, but it's what we homo-sapiens have built concertedly and we are responsible for this reprehensible, disgraceful and blameworthy World Wide Web.

Generative AIs haven't made WWW easier searchable or navigable. There is just more text, more images, more videos, more ads.

[firestar464]

You're not wrong, but it does democratize access to decent depictions of stuff

[wjfox]

[Cyber_Rebel]

Talk about exponential growth, can only imagine next year since it seems like it was only yesterday that I even heard of Pika labs:



Pika Labs new generative AI video tool unveiled — and it looks like a big deal
A big upgrade to AI video

(tomsguide)

Generative artificial intelligence company Pika Labs has unveiled its latest model Pika 1.0. This builds on earlier versions and is a significant step up in AI video generation.

Dubbed an "idea-to-video" model, it can produce content in a range of styles and allows for editing existing video clips by painting over objects, people, or even whole scenes.

In a promotional video for Pika 1.0, you see an example of clothing being changed on the fly, the style of the video clip being updated, and even real people such as Elon Musk being depicted as a cartoon character.

This is where synthetic media really takes off, especially if we can tailor even standard 20-30 min videos to our liking not too far out from now.

[Powers]

Talk about exponential growth, can only imagine next year since it seems like it was only yesterday that I even heard of Pika labs:



Pika Labs new generative AI video tool unveiled — and it looks like a big deal
A big upgrade to AI video

(tomsguide)

Generative artificial intelligence company Pika Labs has unveiled its latest model Pika 1.0. This builds on earlier versions and is a significant step up in AI video generation.

Dubbed an "idea-to-video" model, it can produce content in a range of styles and allows for editing existing video clips by painting over objects, people, or even whole scenes.

In a promotional video for Pika 1.0, you see an example of clothing being changed on the fly, the style of the video clip being updated, and even real people such as Elon Musk being depicted as a cartoon character.

This is where synthetic media really takes off, especially if we can tailor even standard 20-30 min videos to our liking not too far out from now.

I heard about it at least one month ago. As I said before it's very easy to lag behind.

[wjfox]

[wjfox]

[wjfox]

[firestar464]

No human is that clean lmao

[wjfox]

Fears UK not ready for deepfake general election

2 hours ago

Artificial Intelligence (AI) has already been used to disrupt elections around the world - and there are fears among senior politicians and the security services that the UK will be next.

Former Justice Secretary Sir Robert Buckland is urging the government to do more to tackle what he sees as a "clear and present danger" to UK democracy.

The Conservative MP, who now chairs the Northern Ireland select committee, is particularly concerned about the rise of deepfakes - realistic audio and video clips of politicians appearing to say things they did not say.

The threat posed to democracy by AI-generated misinformation does not belong to some dystopian vision of the future, he argues.

"The future is here. It's happening. Unless the policymakers [in the UK] are showing some leadership on the need for a strong and effective domestic set of guardrails - plus international work - then we are going to be behind the curve."

https://www.bbc.co.uk/news/uk-politics-67518511

[wjfox]