Good Morning! Today we’ll look at the proposed licensing framework for powerful AI models put forward by Senators Richard Blumenthal and Josh Hawley. This proposal seeks to regulate companies working on advanced AI models, like OpenAI's GPT-4, requiring them to obtain licenses and undergo rigorous testing to ensure safety and transparency. Additionally, we’ll look at Google's new policy set to roll out in November 2023, compelling political advertisers to disclose the use of AI-generated content in ads, addressing growing concerns about the influence of AI in political campaigns. Then we’ll introduce Vite, a next-generation frontend tooling, designed for web development by providing developers with speed, simplicity, and efficiency in their projects.

— Forrest Knight & Meghanadh Vasireddy

Proposed Licensing Framework for Powerful AI Models

Senators Richard Blumenthal and Josh Hawley have proposed a new legislative framework that would require companies to obtain a license before working on powerful AI models like OpenAI's GPT-4. The proposal aims to create a new US government body to regulate artificial intelligence and restrict work on language models to companies granted licenses.

The senators' proposal would also require developing face recognition and other "high risk" applications of AI to obtain a government license. Companies would have to test AI models for potential harm before deployment, disclose instances when things go wrong after launch, and allow audits of AI models by an independent third party.

The framework proposes that companies should publicly disclose details of the training data used to create an AI model and that people harmed by AI get a right to bring the company that created it to court.

Blumenthal and Hawley will oversee a Senate subcommittee hearing next week about holding businesses and governments accountable when they deploy AI systems that cause harm or violate rights. Microsoft president Brad Smith and Nvidia's chief scientist William Dally are due to testify.

Read More Here

Google to Require Disclaimers for AI-Generated Political Ads

Starting in November 2023, Google will require political advertisers to prominently disclose when their ads contain AI-generated content. This new policy applies to all forms of content, including images, videos, and audio, and mandates a "clear and conspicuous" label for AI-generated content in election ads. The rule will apply to ads appearing on Google’s search engine and YouTube.

The Rise of AI-Generated Political Ads: Some political campaigns have already started leveraging AI to create ads. In April, the Republican National Committee released an attack ad containing AI-generated images targeting President Joe Biden’s bid for reelection. Florida Governor Ron DeSantis also released an attack ad that incorporates AI-generated images of Donald Trump and Anthony Fauci, the former chief medical advisor for the White House. These ads have raised concerns among lawmakers, including Representative Yvette Clarke (D-NY), who introduced a bill that requires disclosures for political ads containing AI-generated content. The Federal Election Commission is also considering restrictions.

Google's new policy is based on the company's existing Manipulated Media Policy and will take effect in November. Political advertisers must ensure that the disclosure is "placed in a location where it is likely to be noticed by users". Ads that utilize AI aspects will need to label them as such in a "clear and conspicuous" manner that is easily seen by the user, per the Google policy. The ads will be moderated first through Google's own automated screening systems and then reviewed by a human as needed.

As campaigns and digital strategists explore using generative AI-tools heading into the 2024 election cycle, Google is the first tech company to announce an AI-related disclosure requirement for political advertisers. The Federal Election Commission has not yet set rules on using AI in political campaign ads, but in August, it voted to seek public comments on whether to update its misinformation policy to include deceptive AI ads.

Read More Here

Vite - Next Generation Frontend Tooling

Vite is a build tool that aims to provide a faster and leaner development experience for modern web projects. It achieves this by leveraging modern tooling and language features to provide an excellent developer experience.

Vite offers on-demand file serving over native ESM, which means no bundling is required. This results in an instant server start and faster development times. Vite also offers Hot Module Replacement (HMR) that stays fast regardless of app size, making it easier for developers to get feedback on their changes in real-time.

Apart from these features, Vite also provides out-of-the-box support for TypeScript, JSX, CSS, and more. It comes pre-configured with Rollup build with multi-page and library mode support.

Vite has quickly become a developer's favorite due to its speed and simplicity. It sets up JS transpiling and bundling for you in an easy way, then provides a server with improved performance in development mode. Vite is still fairly new as a build tool, but it has already proven itself to be a fast development environment.

Read More Here

Hyperdimensional Computing: A New Frontier in AI

Hyperdimensional computing (HDC) is an emerging approach to artificial intelligence that has the potential to revolutionize the field. Instead of relying on traditional binary logic and linear processing, HDC uses high-dimensional mathematical vectors to represent and manipulate information, allowing for more nuanced and holistic processing.

Hypervectors, or high-dimensional vectors, can capture complex data patterns and relationships more accurately than traditional binary computing. Each dimension in a hypervector can represent different aspects or features of the data, allowing for more effective pattern recognition and reasoning. In a recent demonstration, researchers at IBM Research in Zurich used HDC with neural networks to solve a classic problem in abstract visual reasoning, a significant challenge for typical artificial neural networks (ANNs) and even some humans.

HDC offers several benefits over traditional computing, making it well-suited for a new generation of low-power, robust hardware. It is compatible with "in-memory computing systems," which perform computing on the same hardware that stores data, unlike existing von Neumann computers that inefficiently shuttle data between memory and the central processing unit. HDC has shown advantages such as energy efficiency and smaller model size compared to neural networks, although its learning capabilities have been considered sub-par in some cases.

Potential applications of HDC include text analysis, voice analysis, biosignal processing, and image classification. Researchers are currently exploring different ways to implement hyperdimensional vectors in AI systems and optimize the technology for practical use.

Read More Here

Book Review

Software Engineering at Google: Lessons Learned from Programming Over Time

Software Engineering at Google: Lessons Learned from Programming Over Time is a book written by Titus Winters, Hyrum Wright, and Tom Manshreck. The book is based on their experience at Google and presents a candid and insightful view of software engineering at Google. Here are some key takeaways from the book:

• Software engineering is programming integrated over time. It is the multiperson development of multi-version programs.

• Hyrum's law states that with a sufficient number of users of an API, it does not matter what you promise in the contract, all observable system behaviors will be depended on by somebody.

• When programming, clever is a compliment. When software engineering, it's an accusation.

• Testing is an integral part of software engineering. It is not just about finding bugs but also about preventing them.

• Code reviews are crucial for maintaining code quality and knowledge sharing.

• Documentation is essential for maintaining code quality and knowledge sharing.

• Technical debt is a reality in software engineering. It is important to manage it and pay it off over time.

• Communication is key in software engineering. It is important to communicate effectively with team members, stakeholders, and users.

Was this forwarded to you? Sign Up Here