Armilla Review - Advancing AI Governance: Initiatives and Challenges Across Sectors

Welcome to your weekly review. This week’s newsletter looks at initiatives advancing AI Governance and challenges across sectors. In a comprehensive framework for responsible AI innovation, the UK government outlines principles guiding regulators, emphasizing safety, transparency, and fairness. Meanwhile, the Biden administration appoints leaders for the US AI Safety Institute amid funding concerns. The proliferation of deepfake porn prompts debates on legislative measures and industry accountability, while accusations of AI text generation shake trust in scientific publishing. Hugging Face introduces customizable AI Chat Assistants, and OpenMoE sheds light on mixture-of-experts language models. Amidst these developments, Apple's aggressive AI startup acquisitions signify its commitment to maintaining a competitive edge in the rapidly evolving tech landscape.
February 14, 2024
5 min read

Advancing AI Regulation: A Comprehensive Framework for Responsible Innovation

The UK government's AI regulation white paper outlines principles to guide regulators in ensuring responsible AI innovation, including safety, security, transparency, fairness, and accountability. Regulators are aligning their work with these principles, addressing specific risks across various sectors such as societal harms, bias, data protection, and trust in online content. Efforts include updating guidance on AI's impact on employment rights, funding AI education, and addressing copyright concerns. Additionally, the UK advocates for international collaboration on AI governance through bilateral partnerships, multilateral initiatives, and support for AI projects in developing regions. Looking forward, the government plans to engage experts, collaborate with regulators, promote AI opportunities, and support industry while continuing to foster international collaboration on AI governance.

Source: UK Government

New York Proposes AI Rules for Insurers: Implications, Challenges, and Future Directions

The New York Department of Financial Services has proposed rules for insurance carriers regarding the use of artificial intelligence (AI) and alternative data in underwriting and pricing. These rules entail establishing governance protocols for AI systems, employing fairness testing of predictive models, and addressing concerns about systemic biases and inequality. The move mirrors similar initiatives in the European Union and Colorado and follows high-profile incidents involving AI in insurance claims processing. Armilla's Head of AI Policy, Phil Dawson, suggests that these regulations are just the beginning of a broader wave of AI insurance regulation across many states, emphasizing the need for insurers to assess third-party AI models and data sets. Despite the potential of AI to revolutionize the insurance industry, regulators and industry experts remain vigilant about associated risks such as data breaches, algorithmic bias, and security vulnerabilities, urging the development of robust governance frameworks to mitigate these risks.

Source: InsuranceNewsNet

Biden Names Leaders for the US AI Safety Institute Amid Funding Concerns

The Biden administration appointed Elizabeth Kelly and Elham Tabassi as director and chief technology officer, respectively, of the US AI Safety Institute (USAISI) at NIST. NIST, tasked with AI risk management, faces funding uncertainties despite bipartisan requests for $10 million. Chowdhury, an expert, views NIST as pivotal for AI standards but notes funding challenges, describing USAISI's creation as an "unfunded mandate." Nonetheless, optimism persists, with Chowdhury highlighting the institute's placement within NIST as a promising signal for its scientific focus.

Source: VentureBeat

California Proposes Landmark AI Regulations to Address Extreme Risks

A proposed bill in California seeks to regulate the risks posed by advanced artificial intelligence (AI) technology, aiming to prevent potential catastrophes such as rogue weapon systems and cyberattacks. The legislation mandates rigorous testing and emergency shutdown mechanisms for major AI models before their deployment, alongside measures to enhance cybersecurity and establish regulatory enforcement. While the bill focuses on extreme risks and aims to foster innovation, it also includes provisions to promote AI development among smaller entities through initiatives like CalCompute. This initiative, if enacted, could set a precedent for AI regulation nationwide, given California's influence as a hub for major AI companies.

Source: ABC News

Battling the Proliferation of Deepfake Nudes: Challenges and Potential Solutions

AI-generated sexually explicit images, like those of Taylor Swift, have sparked renewed calls for action against deepfake porn. While celebrities are frequent targets, everyday individuals, including minors, are also victimized. The absence of robust laws to address deepfake porn and the challenges in regulating online platforms contribute to its persistence. Proposed solutions range from legislative measures to industry accountability and victim-centered approaches, but debates persist about their efficacy amidst concerns about technological limitations and the scale of the issue. Despite challenges, raising awareness and exploring legal avenues remain crucial steps in combating the harms of AI-generated porn.

Source: Yahoo News

Accusations of AI Text Generation Shake Trust in Scientific Publishing

Lizzie Wolkovich recounts her experience of being accused of using ChatGPT to write a manuscript, despite not having done so. The incident highlights flaws in the peer-review process and raises concerns about the potential for AI to corrupt scientific integrity. Wolkovich calls for clearer standards regarding the use of AI in manuscript writing and better handling of accusations of AI-generated text by scientific journals.

Source: Nature

Hugging Face Launches Free Customizable AI Chat Assistants to Compete with OpenAI

Hugging Face introduces third-party customizable Hugging Chat Assistants, allowing users to easily create personalized AI chatbots similar to OpenAI's custom GPTs. Unlike OpenAI's subscription-based model, Hugging Chat Assistants are free and offer a variety of open source language models for users to choose from. While praised for its user customizability and cost-effectiveness, Hugging Chat Assistants currently lack certain features like web search and logo generation available in GPTs.

Source: VentureBeat

Unveiling OpenMoE: Exploring Open Mixture-of-Experts Language Models

OpenMoE, a collection of fully open-source and reproducible Mixture-of-Experts (MoE) based large language models (LLMs), has been introduced to provide insights into MoE LLMs' cost-effectiveness and routing mechanisms. The study reveals that MoE-based LLMs can offer a more favorable cost-effectiveness trade-off compared to dense LLMs, showcasing potential benefits for future LLM development. Additionally, an in-depth analysis of routing mechanisms within OpenMoE models uncovers three significant findings, including Context-Independent Specialization, Early Routing Learning, and Drop-towards-the-End, prompting the proposal of strategies to address these issues and enhance future MoE LLM designs.

Source: arXiv

OpenAI CEO Seeks Trillions to Revamp Semiconductor Industry

OpenAI CEO Sam Altman is aiming to revolutionize the semiconductor industry with a multitrillion-dollar investment plan, according to reports. Facing limitations in AI chip supply that hinder OpenAI's growth, Altman seeks to bolster global chip-building capacity and is engaging with potential investors, including the UAE government. The ambitious endeavor, which could require $5 trillion to $7 trillion, marks a significant effort to address the growing demand for AI infrastructure and reshape the competitive landscape in the tech sector.

Source: CNBC

Apple's AI Acquisition Blitz: Secures 32 Startups in 2023

Apple has emerged as the frontrunner among major tech companies in acquiring AI startups, with a record-breaking tally of 32 acquisitions in 2023, according to The strategic move underscores Apple's commitment to bolstering its AI capabilities across its product ecosystem. By investing in promising AI startups, Apple aims to stay ahead in the AI arms race, ensuring access to top talent and innovative technologies to maintain its competitive edge in the rapidly evolving tech landscape.

Source: Stocklytics