Keeping AI in check: global efforts to rein in artificial intelligence

Welcome to The Dispatch! We are the newsletter that keeps you informed about AI. Each weekday, we scour the web to aggregate the many stories related to artificial intelligence; we pass along the news, useful resources, tools or services, guides, technical analysis and exciting developments in open source.

In today’s Dispatch:

• From Nature: a comprehensive breakdown of current global efforts to regulate AI

• Researchers from Cornell have discovered an alarming new AI-powered attack that can steal keystrokes and passwords just by listening to your typing sounds

• A look into Microstructure Imaging (MICSI): a new AI-powered technology for MRI scans. The tech allows faster scan times, higher patient throughput and improved diagnostic potential. Current wait timesfor an MRI in the U.S. can be over a year long.

Plus: Disney picks a bad time to make AI news, Amazon Web Services CEO Adam Selipsky talks AI, trending tools and more.

Rules to keep AI in check: a guide to the ways the US, the European Union and China are reining in artificial intelligence

The story: Nature does a deep dive on current approaches to AI regulation being taken by the US, EU and China - as well as some of the challenges in effectively regulating the complex, rapidly evolving technology.

• The United States so far lacks comprehensive federal laws governing AI, instead issuing non-binding principles and guidelines. Some states have enacted their own rules, like Illinois' law requiring disclosure of AI in employment interviews.

• The European Union is taking a precautionary approach with its proposed AI Act (still evolving, final version expected to be published this year), which would ban certain high-risk uses of AI and require due diligence for others based on their level of risk. Companies like Microsoft have welcomed this regulatory framework, but it has been criticized widely by consumer protection groups and others.

• China sees controlling AI as vital for maintaining ideological control and social order. The government walks a fine line between wanting economic gains from AI while strictly limiting its democratizing potential. They have been quite proactive in issuing top-down regulations and restrictions when it comes to AI and internet companies. Through agencies like their Cyberspace Administration (CAC), the government has implemented rules aimed at controlling the spread of harmful content and aligning AI with government interests.

• Globally, attempts at international agreements on AI ethics are underway, but significant disagreements remain between nations. While companies like Google have proposed voluntary standards, experts believe enforcing regulations, especially around transparency, will prove challenging. The varying priorities and values of different countries indicate that meaningful coordination on AI governance might be unlikely in the near future without substantial global effort.

Takeaways: Most efforts on regulating AI are still in very early experimental stages. Monitoring the effects of China’s more direct regulations will be a continuing storyline in AI. Most nations will continue to develop regulations based on their own experimental frameworks and the results of similar such efforts from abroad. It’s also worth noting that any catastrophic high-profile harms caused by the technology will likely spur an instant diversion from current regulatory paths.

Disney creates task force to ‘explore AI and cut costs’

The story: Walt Disney has created a specialized task force to explore how AI can be utilized throughout its expansive entertainment enterprise, with sources claiming the aims of the task force are to keep Disney relevant and cut costs. The task force comes amid a backdrop of Hollywood professionals striking over the potential misuses of the technology.

More details:

• Disney initiated the task force before the recent Hollywood writers' strike; the news breaking at this time is coincidence.

• Disney has 11 jobs currently open that call for AI or machine learning proficiency, covering most of the company's divisions from Imagineering to Disney TV - even advertising.

• AI's potential in Disney's theme park businesses include offering and enhancing unique interactions like Project Kiwi – a machine-learning driven replica of Baby Groot from "Guardians of the Galaxy".

Takeaways: Legacy companies continue to recognize the role AI will have in staying competitive. By expanding AI applications across verticals from movie production to theme parks, Disney is positioning itself to stay relevant. However, balancing AI with workforce concerns, especially in a historically human-centric domain like Hollywood, demands careful navigation. The onus is on Disney and its peers to integrate AI in a manner that complements human talent rather than displacing it.

From our sponsors:

UniTel Voice - the virtual phone system that lets entrepreneurs run & grow their business from anywhere.

Pick Your Number - choose a toll free, local or vanity number that perfectly matches your business, or keep your existing number. After that, connect your greetings to any department or employee extensions, and run your business from anywhere on any device!

Try it risk free for 30 days!

Nvidia announces AI Workbench, a new dev tool for building gen AI models on PCs

The story: Nvidia has just announced ‘AI Workbench’ at the SIGGRAPH conference. This new tool is aimed at aiding developers to create generative AI models on their PCs or workstations by providing a centralized interface to major AI code repositories.

Breakdown of the paper:

• AI Workbench offers a unified interface, connecting directly to repositories like HuggingFace, Github, and NVIDIA NGC.

• The platform facilitates importing and customizing model code directly on local machines.

• Nvidia AI Enterprise 4.0 was also announced. It focuses on secure integration and deployment of AI models with features aiding end-to-end support and optimization of production deployments.

Takeaways: Nvidia's AI Workbench represents the company’s latest strategic move to simplify and streamline generative AI development process for individual developers. With the simultaneous introduction of Nvidia AI Enterprise 4.0, there's an emphasis on security and scalability for business AI deployments. The support from major AI infrastructure providers indicates an industry trend towards integrated development environments for AI.

Social media/YouTube:

• Scanning a menu into ChatGPT’s Code Interpreter to help you decide what to eat [Reddit]

• AI-generated trailer for faux horror film “Magic 8” [Reddit]

• Unity Weta Tools - complete toolchain for 3D creation, simulation, and rendering, for any size of production across game and film [X]

• Nvidia’s new CALM AI explained [YouTube]

• Dr. Stephen Wolfram (creator of the #1 ChatGPT plugin) on the Impact of ChatGPT [YouTube]

Did you know?

Clearbit founder Alex MacCaw has built 101.school. It uses GPT-4 to create multi-week courses and send regular lessons via email. You can choose any topic you’re interested in - or create your own course.

Most online learning resources fall short of the structure of a traditional curriculum. At this phase, 101’s courses are high-level overviews for the most part, but this is an interesting first step towards something more comprehensive.

Trending Tools & Services:

• Picogen: Unofficial API for Midjourney

• Aura: All-in-one digital protection for the whole family, powered by artificial intelligence

• Relume: Create a marketing website in minutes using systems you already love

• Stori: Create and publish on-brand content across multiple platforms