GSV's AI News & Updates (09/13/24)

ChatGPT o1, World Labs, Agentforce, AceUp $22.5M Series A, AllHere Bankruptcy, CA AI Infrastructure, Apple Visual Intelligence, Deepfake Abuse, Classroom of the Future

Take ASU+GSV’s AI Community Pulse Survey to contribute insights on AI in education! (for K-12 and Higher Ed practitioners and educators)

---

---

General 🚀

• OpenAI releases new reasoning model, ChatGPT o1: Formerly code-named Strawberry, the model excels at complex tasks like physics, chemistry, biology, coding, and mathematics, outperforming previous models significantly. However, it has longer response times and isn't recommended for factual queries.

  

  • OpenAI’s new guidance on how to prompt o1: (1) use brief, clear instructions, (2) don't instruct the model to "think step by step" or "explain your reasoning", (3) employ triple quotation marks, XML tags, or section titles to clearly separate different parts of the input, (4) when providing extra context or documents, include only essential context to avoid overcomplicating responses.

  • You can click the dropdown to see it “think” in real time - OpenAI’s o1 models write out their chain of thought as they work through your queries.

    

• OpenAI is in talks to raise $6.5B at a $150B valuation: This valuation is significantly higher than the previous $86B. The funding will be used for computing power and operating expenses and may allow employees to sell shares in a tender offer later this year.

  • OpenAI surpasses 11M paying subscribers and ~$225M revenue a month

• 'AI godmother' Fei-Fei Li raises $230M to launch AI startup: World Labs aims to create AI systems that can construct complete worlds with physics, logic, and rich detail, moving beyond current language-based generative AI. The company has secured $230 million in funding and is valued at $1 billion, despite being a year away from having a product.

• Salesforce Unveils Agentforce: A suite of autonomous AI agents designed to enhance business operations in sales, service, marketing, and commerce by providing customizable, pre-built solutions that operate independently.

• Mistral releases Pixtral 12B, its first multimodal model: a 12-billion-parameter multimodal AI model capable of processing both images and text. Pixtral 12B comes in the wake of Mistral closing a $645M funding round led by General Catalyst that valued the company at $6B.

• TIME releases its annual AI 100 list

Education and the Future of Work 📚

• AceUp Raises $22.5M in Series A to Bring AI-Powered Coaching for Team Transformation: The company offers a business performance analytics platform alongside coaching services, using AI to assess organizational strengths, weaknesses, and growth opportunities. AceUp serves around 100 companies including major corporations (LVMH, L'Oréal, IBM, and SpaceX) and charges $20 per employee per month for basic services, with expert-led solutions for C-suite executives costing up to $1,595 per month.

• AllHere Files for Bankruptcy: LAUSD paid AllHere roughly $3M of the $6M contract before the chatbot was taken offline.

• Clever’s 2024 Classroom of the Future Report: AI optimism on the rise, with the percentage of teachers who believe AI will make their job harder dropping dramatically from 48% to 18% since 2023.

• Oregon ED Deploys AI-Powered Career Exploration Tool for Students: They announced the release of “Sassy,” a free AI-driven tool that aims to make career exploration more fun and less daunting for middle and high school students.

• Universities Across the U.S. Are Testing A.I. as a Fundraising Tool: Universities and nonprofits are using an A.I. avatar to boost donor engagement amid staffing shortages.

• San Diego Supercomputer Center and CENIC Work to Expand AI Education Infrastructure in California: This collaborative program, which is part of the National Research Platform (NRP), aims to provide AI and machine learning resources to K-12 schools, public libraries, community colleges, and universities, with nine CSU campuses and one community college currently participating.

• Oracle AI Agents Help Organizations Achieve New Levels of Productivity: Oracle has introduced over 50 AI agents across its Fusion Cloud Applications Suite to automate tasks, provide insights, and boost productivity in various business functions.

• Klarna shuts down Salesforce as service provider, Workday to meet same fate amid AI initiatives: Klarna's CEO announced the company is discontinuing its use of major SaaS providers like Salesforce and Workday as part of a broader initiative to streamline operations using AI. This move, coupled with workforce reductions, aims to create a more efficient and cost-effective tech stack for the company.

Startups and Tools 🛠️

• Three-quarters of founders in the latest Y Combinator cohort are working on AI startups

• NotebookLM: Acts as a personalized AI collaborator that helps users organize, synthesize, and gain insights from their research materials and notes. Powered by Google's most capable model, Gemini 1.5 Pro.

• Serra: The search engine for recruiters - identify and source top inbound & outbound candidates with Serra's AI-powered search engine.

• Smartcat: One AI translation platform for content in any language.

• Writing with AI: 5 examples of how professional writers are using ChatGPT in innovative ways to enhance their creative process, rather than relying on it to write for them.

• Apple adds Google Lens-like Visual Intelligence to the iPhone 16 camera: Visual Intelligence can recognize objects in images on your iPhone, providing relevant information and performing actions on your behalf (eg. take photo of a restaurant and Apple can pull up the hours)

Tech 💻 

• LLMs are a dead end to AGI, says François Chollet: Despite significant funding for LLMs, some experts view them as unlikely to lead to AGI, advocating instead for redirecting research towards more promising architectures while keeping breakthroughs open and accessible.

• Loopy: An end-to-end audio-conditioned video diffusion model designed to generate more natural and expressive talking head videos from static images.

Safety and Regulation ⚖️

• Major AI Vendors Pledge to Combat Deepfake Abuse: Adobe, Cohere, Microsoft, Anthropic and OpenAI have pledged to combat nonconsensual deepfakes and child sexual abuse material through responsible data sourcing, safeguards in development processes, and dataset management, though the commitments are self-policed and not universally adopted across the industry.

• Facebook and Instagram are making AI labels less prominent on edited content: Labels for AI-generated content remain front and center, but labels for content ‘only modified or edited by AI tools’ will appear behind a menu.

• YouTube is making tools to detect face and voice deepfakes: It plans to launch a pilot program for the voice detection tool by early next year.

• Will California flip the AI industry on its head?: SB 1047 aims to regulate AI, and the AI industry is out to stop it.

• Google’s AI Will Help Decide Whether Unemployed Workers Get Benefits: Nevada is launching an AI system to analyze unemployment appeals, aiming to speed up decisions. While officials promise human oversight, experts worry about bias and accuracy.

• Google AI Overviews rollout hits news publisher search visibility: AI Overviews now being offered for 17% of queries in UK and US.

  • Senators Believe AI Summaries May Be an Antitrust Violation

• NaNoWriMo Organizers Said It Was Classist and Ableist to Condemn AI. All Hell Broke Loose: Last weekend several authors cut ties with National Novel Writing Month after the nonprofit issued a statement seemingly greenlighting the use of AI in writing.

Other

• Audible to Start Generating AI Voice Replicas of Select Audiobook Narrators: Narrators who opt in will be able to construct their voice replicas for free. Afterward, they will be compensated for any audiobooks created using their AI voice.

• New Data Hub Shows How Waymo Improves Road Safety: 84% fewer airbag deployment crashes, 73% fewer injury-causing crashes, 48% fewer police-reported crashes. Out of the 23 most serious Waymo crashes in 22M miles driven, 16 were because a human rear-ended a Waymo car.

• Apple’s new AI is magically mediocre: Apple Intelligence is not that scary, not that advanced, and definitely not finished.

Watch 🎥

• This video is entirely AI-generated…