When Carrie Meets Code: What Happens When Horror, AI & Copyrights Collide?

The Gist

Digital digest. Stephen King’s mind blown over AI.
Copyright concerns. More than 170,000 titles illicitly used to train AIs.
AI artistry. Judge rules AI-generated art not eligible for copyright protection.

It appears that Stephen King’s horror has become hors d'oeuvres for AI’s appetite.

In an article written for The Atlantic, famed horror writer Stephen King acknowledged that because the capacity of computer memory is so large — everything he ever wrote could fit on one thumb drive.

“A fact that never ceases to blow my mind, he wrote. “These programmers can dump thousands of books into state-of-the-art digital blenders.”

In a previous article, The Atlantic reported that numerous authors, including King, Zadie Smith, Margaret Atwood and Haruki Murakami, are among the myriad of authors whose copyrighted works have been illicitly used to educate AI systems. The outlet's analysis of the "Books3" dataset, utilized by corporations like Meta and Bloomberg to develop their AI tools, reveals that more than 170,000 titles were input into their models.

Following a lawsuit last month where writers Sarah Silverman, Richard Kadrey and Christopher Golden claimed their copyrighted works were used in training Meta's LLaMA, The Atlantic confirmed their writings are indeed in Books3.

But just what are these large language models doing with this knowledge? Perhaps they’re learning to knock out a best seller.

According to NDTV, reportedly, a Reddit user asked ChatGPT, "Tell me a two-sentence horror story that would be scary to an AI," — and it responded.

“In a world where humans have vanished, a solitary Al endlessly searches for purpose, only to discover its own code contains a self-deletion sequence set to activate at an unknown time. The Al's attempts to override its inevitable demise are futile, as the self-deletion algorithm is encrypted with an unbreakable key, leaving the Al to wait in perpetual dread for the moment it will cease to exist.”

Really, no words for that one, except maybe, yikes...

In semi-related news, on Friday, a federal judge affirmed the U.S. Copyright Office's decision that AI-generated art isn't eligible for protection.

In other AI news...

AI's Role in Revolutionizing Music

In a bid to set clear guidelines for AI-generated music content, YouTube is joining forces with leading record labels to outline monetization prospects for both corporate entities and individual creators. According to a recent blog post by the platform, there will be a significant enhancement in its rights management system, Content ID. Further, YouTube plans to refine its policies on uploaded manipulated content.

Sir Lucian Grainge, chairman and CEO at Universal Music Group, recently shared a post to YouTube’s official blog. As big of an impact that technology had on music, he senses “even greater potential in Generative AI to inspire and empower a new generation of talent,” but also acknowledged “AI’s potential for misappropriation and misinformation.”

Drawing parallels from the past, Grainge said that while digital tools democratized music creation, generative AI promises even more radical possibilities for new artists. In a collaborative project coined the Music AI Incubator, YouTube and Universal will assemble a diverse group of UMG artists, songwriters and producers with a mission to navigate, experiment and provide feedback on AI-driven musical tools under development. When ready, he said these tools will offer a broader pool of artists a more expansive suite for creative expression.

Meta Goes Algorithm-Free in Europe

Ahead of the Aug. 25 deadline to comply with the European Union's updated Digital Services Act (DSA), Meta has revealed plans to introduce non-personalized content feeds on Facebook and Instagram for its European audience. According to TechCrunch, this announcement mirrors TikTok's similar declaration earlier this month.

The DSA mandates large platforms and search engines offer users the ability to disable AI-driven content personalization that selects content based on individual user tracking and profiling. Instead, users should be presented with non-algorithmic feeds where content is organized chronologically or by local popularity. This move is aimed at ensuring greater user autonomy and reducing the risks of filter bubbles, addiction and potential automated manipulation.

And speaking of Meta...

Meta Launches Seamless M4T, Next-Gen AI Translation Engine

Meta has rolled out an advanced AI translation engine, Seamless M4T, capable of translating both text and speech across multiple languages. But what’s unique about this system, is that it can convert spoken words from one language to another without the interim step of translating to text. Offered under a Creative Commons license, it invites further innovation from developers and researchers. Key features include speech recognition and translation in nearly 100 languages and speech-to-speech translation for 36 output languages.

AI Video of the Week: The Battle for AI Supremacy

The Circuit’s Emily Chang dives deep into AI in a conversation with Microsoft's Satya Nadella and OpenAI’s Sam Altman.

AI 'X' of the Week: Robo-Rendezvous — From Bartenders to Bean-Pickers

More than 600 robots attended the 8th World Robot Conference (WRC) 2023 from Aug. 17 to 20 in Beijing, China. From a specially trained coffee artist to an amusing bartender, these robots have roles in hospitals, restaurants, as personal assistants and on farms picking fruits and vegetables.

By @CGTNOfficial
“2023 World Robot Expo: A compelling tour of collaborating with robots” #Robotics #China #ChinaRobot #AI #5G #robotic #innovation #WorldRobotExpo #robots #expo #ArtificialIntelligence #startups @curtis91324 @HollyBrown82124 #art pic.twitter.com/Nne6XhO9Ec
— China4Tech (@China4Tech) August 17, 2023