IBL News

Author: IBL News

  • Biden Issued an Executive Order Directing Agencies to Develop AI Safety Guidelines

    Biden Issued an Executive Order Directing Agencies to Develop AI Safety Guidelines

    IBL News | New York

    U.S. President Joe Biden signed yesterday an Executive Order that establishes new standards for Generative AI safety, security, and privacy ahead of any legislation coming from lawmakers.

    Biden’s Executive Order responds to the global debate around the need for guardrails to counter the potential pitfalls of giving over too much control to AI systems.

    It builds on the work that led to voluntary commitments from 15 leading companies to reduce the risks of AI.

    The order will:

    • “Require that developers of the most powerful AI systems share their safety test results and other critical information with the U.S. government before companies make them public.
    • Develop standards, tools, and tests to help ensure that AI systems are safe, secure, and trustworthy. “The National Institute of Standards and Technology will set the rigorous standards for extensive red-team testing to ensure safety before public release. The Department of Homeland Security will apply those standards to critical infrastructure sectors and establish the AI Safety and Security Board. The Departments of Energy and Homeland Security will also address AI systems’ threats to critical infrastructure, as well as chemical, biological, radiological, nuclear, and cybersecurity risks. Together, these are the most significant actions ever taken by any government to advance the field of AI safety.”
    • Protect against the risks of using AI to engineer dangerous biological materials.
    • Protect Americans from AI-enabled fraud and deception by establishing standards and best practices for detecting AI-generated content and authenticating official content. “The Department of Commerce will develop guidance for content authentication and watermarking to clearly label AI-generated content. Federal agencies will use these tools to make it easy for Americans to know that the communications they receive from their government are authentic—and set an example for the private sector and governments around the world.”
    • Establish an advanced cybersecurity program to develop AI tools to find and fix vulnerabilities in critical software.
    • Order the development of a National Security Memorandum that directs further actions on AI and security, to be developed by the National Security Council and White House Chief of Staff.”

    “Without safeguards, AI can put Americans’ privacy further at risk. AI not only makes it easier to extract, identify, and exploit personal data, but it also heightens incentives to do so because companies use data to train AI systems. To better protect Americans’ privacy, including from the risks posed by AI, the President calls on Congress to pass bipartisan data privacy legislation to protect all Americans, especially kids,” said the White House.

    Now, the Biden-Harris Administration plans to pursue with Congress to pass bipartisan data privacy legislation by preserving AI development techniques.

    Meanwhile, Europe is on the cusp of passing the first extensive AI regulations.
    .
  • OpenAI Released Advanced Versions of DALL·E 3 and ChatGPT-4

    OpenAI Released Advanced Versions of DALL·E 3 and ChatGPT-4

    IBL News | New York

    OpenAI released and made it available its image model DALL-E 3 to ChatGPT Plus and Enterprise users. DALL-E 3 can create unique, crisper-in-detail images from a simple conversation, providing a selection of visuals for users to refine and iterate upon.

    This model can render intricate details, including text, hands, and faces. It also responds efficiently to extensive and detailed prompts, and it supports both landscape and portrait aspect ratios, as explained in this research paper.

    DALL·E 3 avoids any harmful imagery, including violent, sexual, or hateful content.

    This model is designed to decline requests that ask for an image in the style of a living artist.

    OpenAI offers the option for creators to opt their images out from training of their future image generation models.

    In addition, ahead of the upcoming OpenAI’s DevDay conference next week, where the company is expected to explore new tools with developers, the San Francisco–based research lab released a multimodal version of ChatGPT-4 that allows users to upload and analyze PDFs and various document types.

    The GPT-4 All Tools includes advanced data analysis, DALL·E 3, and built-in browsing capabilities without the need for plugins. These new features may make many third-party ChatGPT plugins obsolete.

    Microsoft’s Bing and Designer also added a more advanced version of DALL·E 3.

    This development pushes the boundaries of generative AI capabilities beyond text-based queries.

     

     


    In other news, OpenAI announced it built a new Preparedness team to evaluate, forecast, and protect against the risks of highly-capable AI—from today’s models to AGI.
    .

     

  • Class.com Expands Its Virtual Learning Platform to Microsoft Teams

    Class.com Expands Its Virtual Learning Platform to Microsoft Teams

    IBL News | New York

    Class.com, the learning platform exclusively built on Zoom since its release in late 2020, launched this month its virtual classroom solution on Microsoft Teams video conferencing platform.

    Class for Teams offers a similar experience to the existing Class for Zoom, with many of the same features and functionalities.

    “With the release of Class for Teams, we can bring Class to even more organizations who are already using Teams, expand the impact of Class, and improve teaching and learning for more individuals around the world,” said the company.

    Class worked together with Microsoft to build Class for Teams.
    .

  • “Open-Source AI Is Taking Over the World,” Says a Key Guide

    “Open-Source AI Is Taking Over the World,” Says a Key Guide

    IBL News | New York

    “Open-source AI represents the future of privacy and ownership of data.” This is the main conclusion of the 2023 State of Open Source AI Book.

    Another key finding is that “open-source AI is taking over the world.”

    In the last year, the open-source community has demonstrated its motivation by delivering quality products and creating different innovations in different fields.

    “This is just the beginning. Many improvements in multiple directions must be made in order to compare the results with centralized solutions,” says the report.

    Experts say that, as it happened with Linux, the world-class operating system, open source will dominate the future of LLMs and image models. Even Google acknowledged that they have no moat in this new world of open source AI.

    The consensus is that open source models are incredibly good at the most valuable tasks, as they can be fine-tuned to cover likely up to 99% of use cases when a product has collected enough labeled data.

  • The Video Editing App CapCut Introduced Its Tools for Business

    The Video Editing App CapCut Introduced Its Tools for Business

    IBL News | New York

    CapCut, the ByteDance-owned video editing app, introduced this month CapCut for Business targeting advertisers and marketers with features such as, AI ad scripts and AI-generated presenters, so they can be able to generate branded content.

    These tools — available across the CapCut app for desktop, mobile, and tablets — help advertisers to come up with script ideas based on their product or business description, as well as commercially licensed business templates, allowing to convert URLs of products or landing pages into videos.

    Tightly integrated with TikTok, CapCut has been a top consumer video editing app that regularly ranks in the top 20 in the iOS App Store.

    The company is positioning its editing app as a way for consumers to make compelling videos for social media, including TikTok, and for marketers to easily do so as well, without having to spend heavily on advanced video editing software.

    CapCut surpassed Splice to become the most profitable video editing app globally during the first half of 2023, pulling in a record high of $50 million, making it ByteDance’s second app to top $100 million globally.

     

     

     

     

     

  • A Majority of CEOs Prioritize Investments in Generative AI

    A Majority of CEOs Prioritize Investments in Generative AI

    IBL News | New York

    Nearly three in four global CEOs say that investing in generative AI is a top spending priority, despite uncertain economic conditions, according to a survey done by KPMG on 1,300 global managers, including 400 in the U.S.

    They expect to see a return on their investment in three to five years.

    They also look forward to increased profitability in new products, market growth opportunities, enhanced innovation, and aid cybersecurity efforts.

    Fewer CEOS — less than one-third — expect a faster ROI of one to three years.

    For CIOs, the focus is finding real value from implementation, as they have to find the proper foundational models and characteristics.

    “Increased disruption and structural changes to the economy are compounding risks, requiring CEOs to move forward with long-term growth strategies while remaining agile to take advantage of new opportunities and respond to unforeseen challenges,” said Paul Knopp, KPMG U.S. Chair and CEO.

  • American Federation of Teachers Partners with AI Identification Platform GPTZero

    American Federation of Teachers Partners with AI Identification Platform GPTZero

    IBL News | New York

    The American Federation of Teachers (AFT), the second-largest teacher’s union in the U.S., signed a deal with the identification platform GTPZero to detect when students use artificial intelligence to do their homework.

    The teacher’s union will be paying for access to more tailored AI detection and certification tools and assistance.

    “If we don’t guard against its perils upfront, we’re going to repeat the terrible transitions that happened with the industrial revolution,” AFT President Randi Weingarten told CBS MoneyWatch. “ChatGPT can be a really important supplement and complement to educators if the guardrails are in place.”

    “You can’t stop technology and innovation. You need to ride it and harness, it and that’s what we are talking to our members about,” she said.

    Founded in January by Princeton graduate Edward Tian, GPTZero is a 15-person company that says “it’s working with teachers to figure out where AI fits into education and empower students to use AI responsibly.”

    • Wired: Kids Are Going Back to School. So Is ChatGPT

     

     

  • Nvidia Announced an AI Agent Powered by GPT-4 That Can Teach Robots Complex Skills

    Nvidia Announced an AI Agent Powered by GPT-4 That Can Teach Robots Complex Skills

    IBL News | New York

    NVIDIA Research announced yesterday that it developed an AI agent called Eureka powered by GPT-4 LLM and generative AI. Eureka can teach robots complex skills by writing code that rewards robots for reinforcement learning.

    One of the 30 tasks is a robotic hand to perform rapid pen-spinning tricks for the first time as well as a human can.

    Eureka has also taught robots to open drawers and cabinets, toss and catch balls, and manipulate scissors, among other tasks.

    The Eureka research includes a paper and the project’s AI algorithms, which developers can experiment with.

    “Eureka is a first step toward developing new algorithms that integrate generative and reinforcement learning methods to solve hard tasks,”     said Anima Anandkumar, Senior Director of AI Research at NVIDIA and an author of the Eureka paper.

    The results from nine Isaac Gym GPU-accelerated simulation environments are showcased in visualizations generated using NVIDIA Omniverse.

    “It’s breakthrough work bound to get developers’ minds spinning with possibilities, adding to recent NVIDIA Research advancements like Voyager, an AI agent built with GPT-4 that can autonomously play Minecraft.

    NVIDIA Research comprises hundreds of scientists and engineers worldwide, with teams focused on topics including AI, computer graphics, computer vision, self-driving cars, and robotics.

  • LLMs Models Will Continue to Drive Real-World Breakthroughs, Says ‘State of the AI Report’

    LLMs Models Will Continue to Drive Real-World Breakthroughs, Says ‘State of the AI Report’

    IBL News | New York

    LLMs models will continue to drive real-world breakthroughs, especially in the life sciences, with meaningful steps forward in both molecular biology and drug discovery.

    This is one of the conclusions of The State of AI Report, a classic research study produced by AI investors Nathan Benaich and the Air Street Capital and reviewed by AI practitioners.

    Other key findings include:

    • GTP-4 is beating every other LLM, validating the power of proprietary architectures and reinforcement learning from human feedback.

    • Efforts are growing to try to clone or surpass proprietary performance through smaller models, better datasets, and longer context.

    • Compute is the new oil, with NVIDIA printing record earnings and startups wielding their GPUs as a competitive edge. As the US tightens its restrictions on trade restrictions on China and mobilizes its allies in the chip wars, NVIDIA, Intel, and AMD have started to sell export-control-proof chips at scale.

    • Generative AI startups raised over $18 billion from VC and corporate investors, while other tech industry valuations are on a slump.

    • Safety concerns are prompting action from governments and regulators around the world.
    .

    > Download the Report

  • Stanford, MIT, and Princeton Researchers Rate 10 LLM on How Transparent They Are

    Stanford, MIT, and Princeton Researchers Rate 10 LLM on How Transparent They Are

    IBL News | New York

    Stanford University, MIT, and Princeton researchers ranked ten major AI models on how openly they operate after applying a newly created scoring system.

    Included in the index are popular models like OpenAI’s GPT-4 (which powers the paid version of ChatGPT), Google’s PaLM 2 (which powers Bard), and Meta’s LLaMA 2. It also includes lesser-known models like Amazon’s Titan Text and Inflection AI’s Inflection-1, the model that powers the Pi chatbot.

    Three systems in the list (Meta, Hugging Face, and Stability AI) develop open foundation models (Llama 2, BLOOMZ, and Stable Diffusion 2, respectively), and the other seven developers built closed foundation models accessible via an API.

    Llama 2 led at 54%, GPT-4 placed third at 48%, and PaLM 2 took fifth at 40%. See the ranking in the table above.

    The Stanford ‘Foundation Model Transparency Index’ paper research featured 100 indicators, including the social aspects of training foundation models (the impact on labor, environment, and the usage policy for real-world use) in addition to technical aspects (data, computing, and details about the model training process).

    “The indicators are based on, and synthesize, past interventions aimed at improving the transparency of AI systems, such as model cards, datasheets, evaluation practices, and how foundation models intermediate a broader supply chain,” explained one of the authors in a blog post.

    “Transparency is poor on matters related to how models are built. In particular, developers are opaque on what data is used to train their model, who provides that data and how much they are paid, and how much computation is used to train the model.” 

    The researchers released a repository with all of their analysis on our GitHub repository as well.

    The New York Times reported that several AI companies have already been sued by authors, artists, and media companies, accusing them of illegally using copyrighted works to train their models.

    So far, most of the lawsuits have targeted open-source AI projects or projects that disclosed detailed information about their models.