Mark Ghuneim

AI risk regulation frameworks:

Singapore’s Model AI Governance Framework: Soft law focused on explainable, transparent, and human-centric AI decision-making.

The Draft EU AI Act: Hard law that classifies AI systems into unacceptably risky, high-risk, and low-risk uses, and implements risk regulation for high-risk AI systems.

NIST’s AI Risk Management Framework (U.S.): Soft law that emphasizes enterprise risk management and an iterative approach to risk management.

This is my current area focus if you’re digging here please get in touch.

I am excited about the potential of this medium to offer meaningful and captivating experiences. To further explore the possibilities of generative storytelling and to contribute to the growth of this emerging field.

There are volumes to be written on the shift in browsing and publishers new challenges around IP – but this is not that. A fastshort observation about the browsing version of ChatGPT. It often surfaces urls – new to me and refreshing resources that have been unaware of.

Today when asking for a news summary it referred to this alt view of hckrnews.com which was nice as my other alt view hack.ernews.info stopped working.

The Copyright Office has issued a new policy statement regarding the registration of works containing material generated by artificial intelligence (AI). Providing guidance on how to submit applications for copyright protection for works that contain AI-generated material, and how to correct previously submitted or pending applications.

Key points

• Individuals who use AI technology in creating a work may claim copyright protection for their own contributions to that work.
• The Standard Application must be used, and in it, applicants must identify the author(s) and provide a brief statement in the ‘‘Author Created’’ field that describes the authorship that was contributed by a human.
• Applicants should exclude AI-generated content that is more than de minimis in the application, by providing a brief description of the AI-generated content under the ‘‘Material Excluded’’ heading in the ‘‘Limitation of the Claim’’ section.
• Applicants who fail to update the public record after obtaining a registration for material generated by AI risk losing the benefits of the registration.
• If necessary, the examiner will correspond with the applicant to obtain additional information about the nature of the human authorship included in the work.
• The supplementary registration process may be used to correct errors in a copyright registration or to amplify the information given in a registration, including the exclusion of AI-generated content
• The policy statement reflects a major change in the way copyright law addresses works containing AI-generated material.

The Copyright Office will host public listening sessions throughout the spring of 2023, with the following dates:

Literary Works on Wednesday, April 19, from 1:00 p.m. to 4:00 p.m. ET

• Visual Works on Tuesday, May 2, from 1:00 p.m. to 4:00 p.m. ET
• Audiovisual Works on Wednesday, May 17, from 1:00 p.m. to 4:00 p.m. ET
• Music and Sound Recordings on Wednesday, May 31, from 1:00 p.m. to 4:00 p.m. ET

Later in the year, the Copyright Office plans to publish a notice of inquiry soliciting public comments on a wide range of copyright issues arising from the use of AI.  The Copyright Office will continue engaging with the public through informational webinars during the summer.

Abstract: This paper explores the alignment of digital interfaces with our existing innate encoding and decoding of information.

The focus is on “front door” moments in technology, where new interfaces or technologies fundamentally change the way we interact with information and the world around us.

Top 10  “front door” moments in technology and information theory, along with a short description and the year the technology became broadly available:

1. Web browsers (1993) – The development of the first web browsers, including Mosaic, provided users with a way to access and navigate the web.
2. Search engines (1995) – The emergence of search engines, such as Yahoo and AltaVista, made it easier for users to find information on the web.
3. Site maps (1998) – The development of site maps, which provided a visual representation of a website’s structure and content, made it easier for users to find what they were looking for.
4. Social media (2003) – The emergence of social media, such as MySpace and LinkedIn, provided new ways for users to access and share information.
5. GPS data (2009) – Foursquare’s use of GPS data to put a “doorway and face” on location information provided users with a new way to access and share information about their location.
6. Mobile devices (2007) – The emergence of smartphones, such as the iPhone and Android, required the development of new and innovative interfaces that could accommodate smaller screens and limited processing power.
7. Big data (2010s) – The emergence of big data and the development of tools and techniques for processing and analyzing large amounts of information provided new ways to access and process information.
8. AI and machine learning (2010s) – The development of AI and machine learning has enabled new approaches to information retrieval and processing, with significant advances in recent years.
9. Blockchain (2009) – The emergence of blockchain technology and its potential to create new and secure ways to exchange and verify information is an example of a new “front door” to information exchange.
10. Text and voice to AI UI (2010s) – The development of natural language processing and speech recognition technologies has enabled new ways for users to interact with AI systems in a more intuitive and natural way, opening up new “front doors” for effective communication and collaboration.

 (c) Mark Ghuneim 202

Generative Entertainment Milestones  In chronological order.  

Milestones

1. 1956: The first computer-generated music is created by an IBM computer in Australia.
2. 1966: The Yule Log TV, a holiday program featuring a looping video of a burning fireplace, debuts on WPIX-TV in New York City.
3. 1979: The first video game with procedural generation, Rogue, is released.
4. 1982: The Weather Channel begins broadcasting, featuring generative content such as weather maps and graphics.
5. 1985: Max Headroom, a television series featuring a computer-generated character, debuts on HBO.
6. 1998: Brian Eno releases “Generative Music 1,” a music album generated by a software program that creates music in real time.
7. 2011: “Hip Hop Beats to Study/Relax to” YouTube channel is created, featuring 24/7 live streaming of lo-fi hip hop music with generative visuals.
8. 2016: “Radio Garden,” a web-based platform that allows users to listen to live radio stations from around the world, features a generative interface that shows a spinning globe and allows users to navigate to different locations.
9. 2017: “The Infinite Now,” a short film created by filmmaker Armand Dijcks and photographer Ray Collins, features generative visuals that create a sense of infinite movement in a series of still images.
10. 2022: Genuary is held for the second time, featuring a range of generative art and creative coding challenges. (c) Mark Ghuneim 2023

Moments

1. “Nothing, Forever” – a generative storytelling experiment using Twitch as the platform.
2. “AI Dungeon” – an interactive storytelling game that uses AI-generated text to create dynamic storylines.
3. “Minecraft” – a popular video game that allows players to create and explore procedurally generated worlds, and has also been used to host virtual events such as the Travis Scott Concert.
4. “Google DeepDream” – a software that uses deep learning algorithms to transform images into dream-like, hallucinogenic visuals.
5. “Infinite Jest” – a novel by David Foster Wallace that features numerous generative elements, such as a filmography of made-up movies and a detailed timeline of future events.
6. “AIVA” – an artificial intelligence program that composes original music in a range of styles and genres.
7. “The Infinite Album” – a music album created by musician Bryce Dessner that features generative elements that change each time the album is played.

As we venture into uncharted territory in the realm of synthetic media, it’s important to provide clear definitions and vocabulary to help others understand and engage with the vision. (As someone who sees around corners, I learned this the hard way many times!!)

Looking back at the history of generative entertainment can provide a framework for moving forward in this new generative AI landscape beyond sorting, completion, summary etc and start to look at the tooling ++++ around creative uses

MusicLM: Generating Music From Text 

“We introduce MusicLM, a model generating high-fidelity music from text descriptions such as “a calming violin melody backed by a distorted guitar riff”. MusicLM casts the process of conditional music generation as a hierarchical sequence-to-sequence modeling task, and it generates music at 24 kHz that remains consistent over several minutes. Our experiments show that MusicLM outperforms previous systems both in audio quality and adherence to the text description. Moreover, we demonstrate that MusicLM can be conditioned on both text and a melody in that it can transform whistled and hummed melodies according to the style described in a text caption. To support future research, we publicly release MusicCaps, a dataset composed of 5.5k music-text pairs, with rich text descriptions provided by human experts.”

To get a real idea of this model check the different examples : Project page

Audio Generation From Rich Captions :  The audio is generated by providing a sequence of text prompts. These influence how the model continues the semantic tokens derived from the previous caption.

Long Generation 

Story Mode  : The audio is generated by providing a sequence of text prompts. These influence how the model continues the semantic tokens derived from the previous caption.

Text and Melody Conditioning : By adding melody embeddings to the conditioning, we can generate music that respects the text prompt while following the provided melody.

Painting Caption Conditioning   (!!!)

Generation Diversity

Test the diversity of the generated samples while keeping constant the conditioning and/or the semantic tokens.