The National Science Foundation (NSF) is launching a pilot program to create the National Artificial Intelligence Research Resource (NAIRR), a shared U.S. AI research infrastructure. The move fulfills part of President Biden’s October executive order on the responsible development of artificial intelligence. Ten other federal agencies have joined the NSF in launching the program, while tech giants Microsoft, Nvidia and Google have already pledged their support, along with more than 20 other private organizations across the industry, academic and non-profit sectors. The idea is to create shared access to information and things like cloud computing resources. Continue reading Big Tech Onboard to Advance President Biden’s NSF AI Pilot

U.S. Consumer Product Safety Commission (CPSC) Chair Alexander Hoehn-Saric spoke with CTA Senior Director of Regulatory Affairs Rachel Nemeth during CES 2024 about the challenges of extending safety to products that are constantly evolving and incorporating new technologies such as machine learning and artificial intelligence. Nemeth pointed out that the agency’s authorizing statute was enacted in 1972 and was last amended in 2008. “We’re doing a lot of good work with the statute we have,” Hoehn-Saric responded. “But we’re changing the way we operate. We talk a lot about machine learning and AI.” Continue reading CES: Championing Consumer Product Safety in the Age of AI

Apple recently announced advances in artificial intelligence research that could introduce more immersive visual experiences and enable sophisticated AI systems to run on the company’s popular mobile devices. Two new research papers highlight techniques for creating 3D avatars from video content and efficiently deploying large language models on devices challenged by limited memory. The real-time ability to create avatars and 3D scenes from an iPhone camera could bring a range of new possibilities for CE devices in areas such as synthetic media, telepresence, social interaction, virtual try-on and more. Continue reading Apple Unveils New Advances in Artificial Intelligence Research

The research division of Meta AI has developed Seamless Communication, a suite of artificial intelligence models that generate what the company says is natural and authentic communication across languages, facilitating what amounts to real-time universal speech translation. The models were released with accompanying research papers and data. The flagship model, Seamless, merges capabilities from a trio of models — SeamlessExpressive, SeamlessStreaming and SeamlessM4T v2 — into a single system that can translate between almost 100 spoken and written languages, preserving idioms, emotion and the speaker’s vocal style, Meta says. Continue reading Meta AI Seamless Translator Converts Nearly 100 Languages

Stability AI has opened research preview on its first foundation model for generative video, Stable Video Diffusion, offering text-to-video and image-to-video. Based on the company’s Stable Diffusion text-to-image model, the new open-source model generates video by animating existing still frames, including “multi-view synthesis.” While the company plans to enhance and extend the model’s capabilities, it currently comes in two versions: SVD, which transforms stills into 576×1024 videos of 14 frames, and SVD-XT that generates up to 24 frames — each at between three and 30 frames per second. Continue reading Stability Introduces GenAI Video Model: Stable Video Diffusion

Having made the leap from image generation to video generation over the course of a few months in 2022, Meta Platforms introduces Emu, its first visual foundational model, along with Emu Video and Emu Edit, positioned as milestones in the trek to AI moviemaking. Emu uses just two diffusion models to generate 512×512 four-second long videos at 16 frames per second, Meta said, comparing that to 2022’s Make-A-Video, which requires a “cascade” of five models. Internal research found Emu video generations were “strongly preferred” over the Make-A-Video model based on quality (96 percent) and prompt fidelity (85 percent). Continue reading Meta Touts Its Emu Foundational Model for Video and Editing

A second Meta Platforms whistleblower has come forward, testifying this week before a Senate subcommittee that the company’s social networks were potentially harming teens, and his warnings to that effect were ignored by top leadership. Arturo Bejar, from 2009 to 2015 a Facebook engineering director and an Instagram consultant from 2019 to 2021, told the Senate Judiciary Subcommittee on Privacy, Technology and Law that Meta officials failed to take steps to protect underage users on the platforms. Bejar follows former Facebook whistleblower Frances Haugen, who provided explosive Senate testimony in 2021. Continue reading Second Meta Whistleblower Testifies to Potential Child Harm

LinkedIn expects to pass the 1 billion user mark this month, and timed to that feat is unleashing a new suite of AI productivity tools, including job coaching, personalized digests and help writing original content for the platform. The new machine learning assists will initially be available only to Premium subscribers, centered on the aforementioned three main areas. The move follows months in which LinkedIn has been upgrading its AI capabilities in areas like automated recruiter messaging, job descriptions and profile writing suggestions. The improvements draw on OpenAI technology, in which LinkedIn parent Microsoft has an ownership stake. Continue reading Nearing 1 Billion Users, LinkedIn Debuts Job Coach Chatbot

The University of Science and Technology of China (USTC) and Tencent YouTu Lab have released a research paper on a new framework called Woodpecker, designed to correct hallucinations in multimodal large language AI models. “Hallucination is a big shadow hanging over the rapidly evolving MLLMs,” writes the group, describing the phenomenon as when MLLMs “output descriptions that are inconsistent with the input image.” Solutions to date focus mainly on “instruction-tuning,” a form of retraining that is data and computation intensive. Woodpecker takes a training-free approach that purports to correct hallucinations from the basis of the generated text. Continue reading Woodpecker: Chinese Researchers Combat AI Hallucinations

OpenAI has released the DALL-E 3 generative AI imaging platform in research preview. The latest iteration features more safety options and integrates with OpenAI’s ChatGPT, currently driven by the now seasoned large language model GPT-4. That is the ChatGPT version to which Plus subscribers and enterprise customers have access — the same who will be able to preview DALL-E 3. The free chatbot is built around GPT-3.5. OpenAI says GPT-4 makes for better contextual understanding by DALL-E, which even in version 2 evidenced some glaring comprehension glitches. Continue reading OpenAI’s Latest Version of DALL-E Integrates with ChatGPT

Four in five U.S. homes now have a smart TV, accounting for three in five TV sets, according to the fifth annual Hub Entertainment Research “Evolution of the TV Set” survey, which found streaming is growing commensurate with penetration of the intelligent displays. About 64 percent of viewers use their smart TVs to stream video, while roughly half use the connected devices to stream music or other audio content, the study found. The 74 percent of households that own at least one smart TV is up from 61 percent in 2020. Additionally, Horowitz Research found that consumers are increasingly turning to curated collections and hubs for content discovery. Continue reading Study: Smart TVs Are Now in 74 Percent of American Homes

Advancing President Biden’s push for responsible development of artificial intelligence, top AI firms including Anthropic, Google, Microsoft and OpenAI have launched the Frontier Model Forum, an industry forum that will work collaboratively with outside researchers and policymakers to implement best practices. The new group will focus on AI safety, research into its risks, and disseminating information to the public, governments and civil society. Other companies involved in building bleeding-edge AI models will also be invited to join and participate in technical evaluations and benchmarks. Continue reading Major Tech Players Launch Frontier Model Forum for Safe AI

This week, Meta Platforms released Llama 2, the next generation of its open-source large language model that is free for research and commercial use. Llama 2’s pretrained and fine-tuned language models are available in sizes ranging from 7 to 70 billion parameters. Meta also named Microsoft Azure its “preferred partner for Llama 2,” offering it through the Azure AI model catalog for use with cloud-native tools that leverage content filtering and safety features. Meta says Llama 2 is “also optimized to run locally on Windows,” providing developers a seamless workflow across enterprise and consumer platforms. Continue reading Meta Unveils Llama 2 LLM with Microsoft as Preferred Partner

Threads released an iOS update this week that automates the ability to translate posts into foreign languages. The Instagram spinoff also added a follows tab to the activities feed, where replies and mentions are displayed. Also new to iOS is the ability to access a list of any user’s Instagram followers, to subscribe to “unfollowed” users, and tappable repost labels. While Threads has prompted shock and awe by hitting 100 million downloads within five days of its July 5 launch, and is now at about 150 million, there are reports of dips in user activity. Meanwhile, the new platform has followed Twitter in introducing tighter rate limits. Continue reading Meta’s Threads Adds Updates, Aims to Suppress Bot Attacks

IATSE, the International Alliance of Theatrical Stage Employees, has adopted what it is calling “Core Principles for Applications of Artificial Intelligence and Machine Learning Technology” in the entertainment industry. Unveiled last week, the document follows the May creation of the group’s Commission on Artificial Intelligence. The activity is meant to convey a proactive approach to the challenges and opportunities ahead. “With AI, the stakes for IATSE members in all crafts is high,” said IATSE International President Matthew Loeb. “There is much work to do, but I am pleased to report the union’s efforts are already well underway.” Continue reading IATSE Unveils Its Core Principles for AI and Machine Learning