GSV's AI News & Updates (08/16/23)
Uplimit's AI Teaching Assistant, One Model Raises $41M, Vimeo's One-Take Video, NVIDIA Announcements, Language Model Political Biases, Bioterrorism
You are receiving this weekly update on the intersection of AI and education, as part of our GSV community – we hope this will be a helpful resource as we navigate this evolving and exciting intersection. Please feel free to send over any relevant resources you may come across (claire@gsv.com) – we’d love to learn and hear from you!
General 🚀
NVIDIA CEO Jensen Huang’s major announcements at SIGGRAPH: highlights include (1) the next generation GH200 Grace Hopper Superchip platform, (2) NVIDIA AI Workbench (a new unified toolkit that introduces simplified model tuning and deployment on NVIDIA AI platforms) and (3) a major upgrade to NVIDIA Omniverse with generative AI and OpenUSD.
Vimeo launches one-take video creation: create wow-worthy content in minutes with an AI-powered script generator, teleprompter, and a new way to edit video by editing text.
Amazon taps generative AI to enhance product reviews: they will leverage generative AI to help customers better understand what is being said about a product, without necessarily having to read through dozens of individual reviews. Amazon will also highlight key product attributes as clickable buttons (eg. ease of use, performance, etc).
How Mars, Colgate-Palmolive, Nestle & Coca-Cola are Exploring Generative AI
Anthropic raises $100M from South Korea's SK Telecom: Anthropic, a startup competing with OpenAI in building AI foundation models, is among the most well-funded AI firms, having raised $450M from investors including Google and Spark Capital.
AI language models are rife with different political biases: researchers conducted tests on 14 large language models and found that OpenAI’s ChatGPT and GPT-4 were the most left-wing libertarian, while Meta’s LLaMA was the most right-wing authoritarian.
Education and the Future of Work 📚
AI in Education: the leap into a new era of machine intelligence carries risks and challenges, but also plenty of promise.
Uplimit—formerly CoRise - offers AI bot teaching assistant: the company’s goal is not to use AI to lessen the involvement of its human teachers. Instead, it aims to augment their expertise in a way that scales way, way up. Uplimit’s AI assistant, dubbed Cobot, answers students’ questions, analyzes their programming code, and even provides moral support. The company is also deploying AI to aid instructors as they put together their course plans.
An Iowa school district is using AI to ban books: the Mason City School District is bringing in AI to parse suspect texts for banned ideas and descriptions since there are too many titles for human reviewers to cover on their own. So far, the AI has flagged 19 books for removal.
Bringing AI Literacy to High Schools: Stanford education researchers collaborated with teachers to develop classroom-ready AI resources for high school instructors across subject areas.
The $900,000 AI Job Is Here: salaries rise as employers such as Netflix and Walmart seek candidates with artificial-intelligence skills.
AI Could Save School Districts Time and Money—If They Use It Correctly
Startups and Tools 🛠️
One Model lands $41M to bring AI & data science-powered insights to HR: One Model is an “enterprise people data orchestration platform” - a toolkit for extracting, modeling and governing HR data as well as delivering that data to various applications and services. It performs basic tasks like identifying areas where a company has a shortage of skills or talent and projecting future workforce needs; calculating the cost of turnover and headcount, attempting to create a plan that measures and reduces this cost over time.
One Model’s monitoring dashboard, which unifies HR data in one place. MindPal: your personal repository and second brain, dump all your files and chat with them like a second brain - chat with YouTube, a website, PDF, PowerPoint and more.
Fynt: AI powered finance teams - supercharge your financial department with automated GPT-powered financial insights, reconciliation, reporting and more.
Wondercraft: turn existing content into captivating podcasts.
Metaphor: AI search engine that allows you to describe what you’re searching for and then locates it for you (tweets, wiki, news, papers and more).
Tweetify It: transform any long-form content (text, blog-posts, websites) into engaging and personalized short media posts for Instagram, Twitter, LinkedIn and more.
Tech 💻
Meta releases synthetically generated Photorealistic Unreal Graphics (PUG) datasets: they used the Unreal Engine, a powerful game engine well-known in the entertainment industry, to create photorealistic interactive environments from which they can easily sample images with given specifications. These PUG datasets are for model evaluation and benchmarks (license does not allow training generative models).
Stanford Smallville is officially open-source: Smallville is among the most inspiring AI agent experiments in 2023. 25 AI agents inhabit a digital Westworld, unaware that they are living in a simulation. They go to work, gossip, organize socials, make new friends, and even fall in love. Each has unique personality and backstory.
Scale AI launches its Test & Evaluation platform for LLMs: Scale’s hybrid approach to test & evaluation combines automated evaluations, a best-in-class evaluation platform, and an expert network to assess an LLM’s helpfulness vs harmfulness. Scale was selected to develop the platform to evaluate leading Generative AI Systems at DEFCON 31, as announced by the White House’s Office of Science and Technology Policy (OSTP).
Model Evaluation: quantitative and qualitative evaluation of model performance
Red Teaming: identification of model weaknesses and vulnerabilities
DoctorGPT: a Large Language Model that can pass the US Medical Licensing Exam. This is an open-source project with a mission to provide everyone their own private doctor. It works offline, it's cross-platform, & your health data stays private.
Regulation and Policy ⚖️
DOD Announces Establishment of Generative AI Task Force: led by the Chief Digital and Artificial Intelligence Office (CDAO), Task Force Lima will assess, synchronize, and employ generative AI capabilities across the DoD.
U.S. Bans Future Investments in Chinese AI, Semiconductor, and Quantum Computing
Stanford’s Bootcamp to Educate Congress on AI: Hosted by Stanford’s Institute for Human-Centered AI (HAI), the event offered a crash course on AI’s benefits and risks for information-starved staffers staring down the possibility of legislating a fast-moving technology in the middle of a gold rush. Attendees included aides for Rep. Ted Lieu (D-Calif.) and Sen. Rick Scott (R-Fla.).
Poll: Americans distrust AI giants: 72% prefer slowing down the development of AI compared to just 8% who would rather speed it up.
Why the Great AI Backlash Came for a Tiny Startup You’ve Probably Never Heard Of: a literary analytics project called Prosecraft has shuttered after backlash from the writing community. It's a harbinger of a bigger cultural tide shift.
AI Governance Is Not Just About The Models: 3 different “levels” of AI Governance: Organizational, Use Case and Model.
AI Causes Real Harm. Let’s Focus on That over the End-of-Humanity Hype: Effective regulation of AI needs grounded science that investigates real harms, not glorified press releases about existential risks.
Other
ChatGPT could make bioterrorism horrifyingly easy: In a recent exercise at MIT, it took just one hour for ChatGPT to instruct non-scientist students about four potential pandemic pathogens, including options for how they could be acquired by anyone lacking the skills to create them in the lab, and how to avoid detection by obtaining genetic material from providers who do not screen orders.
AI comes for YouTube’s thumbnail industry: creators like MrBeast pay up to $10,000 for a single YouTube video thumbnail, but the rise of AI tools has some designers anxious about the microeconomy’s future.
Unconfuse Me with Bill Gates and Sal Khan (Episode 2): Bill Gates chats with Sal about why tutoring is so important, how his new service Khanmigo is making the most of ChatGPT, and how we can keep teachers at the center of the classroom in the age of AI.
Jan Leike on OpenAI’s massive push to make superintelligence safe in 4 years or less:
Resources:
The GSV BIG 10: weekly coverage corner for the top 10 stories, insights, and major plays in learning and skilling.