GSV's AI News & Updates (11/08/24)

Google Project Jarvis, Perplexity $9B Valuation, Meta Search Engine, Google LearnAbout, Instagram "Adult Classifier", LinkedIn AI Recruiter, Open-World AI Simulation

Join me at the leading edge of all things AI in EDU at the second annual AI SHOW @ ASU+GSV, where 12,000+ PreK to Gray attendees will experience cutting-edge AI innovations through live EdTech demos, promptathons and hackathons, hands-on workshops, incredible thought leadership, and so much more — April 5-7 at the San Diego Convention Center. Register for FREE now, capacity is limited.

---

General 🚀

• Google’s Next AI Just Leaked, and It Can Take Over Your Computer: Project Jarvis is designed to automate various web-based tasks by taking control of a user's browser. It can perform functions such as researching topics, booking flights, and purchasing products without requiring manual input from the user. The AI prototype is capable of taking over a computer's web browser, enabling it to execute commands like typing text and clicking buttons autonomously.

• AI Startup Perplexity to Triple Valuation to $9B in New Funding Round: The company has experienced a 700% increase in revenue and usage since January 2024, currently handling about 100M queries weekly and generating around $50M in annual revenue. The funding round is led by Institutional Venture Partners (IVP). The company is facing copyright infringement lawsuits from major news outlets for using their content without compensation.

• Meta is reportedly working on its own AI-powered search engine, too: Meta’s team has been building a proprietary database and web crawler for about eight months, aiming to reduce reliance on Google and Microsoft.

• Meta’s former hardware lead for Orion is joining OpenAI: Kalinowski's hiring signals OpenAI's renewed commitment to robotics. The company is reportedly rebuilding its robotics team and aims to develop technologies that can be integrated into existing humanoid robots rather than competing directly with them.

• Tech Giants Are Set to Spend $200B This Year Chasing AI: Amazon and Alphabet exceeded earnings expectations, boosted by cloud growth, while Meta and Microsoft faced stock declines due to high spending and cloud capacity issues. Analysts generally remain optimistic, viewing these investments as necessary groundwork for future success.

  

• Microsoft says AI is on pace to be a $10 billion-a-year business: marking the fastest product category in the company's history to hit this milestone. Microsoft reports that 70% of Fortune 500 companies are using Microsoft 365 Copilot for some employees.

Education and the Future of Work 📚

• Google's new AI tool could be your new favorite learning aid - and it's free: Learn About is an AI-powered learning companion that functions similarly to a chatbot. Users can input topics, questions, or even upload images and documents to receive tailored responses based on their knowledge level. This tool aims to create a personalized learning experience by guiding users through challenging concepts with interactive resources like articles, videos, and images.

  

• Teachers: “These AI Resources Are Not Classroom-Ready.”: Educators found AI-generated materials to be text-heavy, lacking engagement, and insufficient for effective teaching. The materials were seen as presenting information passively rather than supporting active learning.

• AI, Privacy, and Power: The Ethical Battle Over Student Data in Higher Education: Current privacy protections, such as FERPA, allow universities broad discretion in sharing data with third-party vendors, sometimes without student consent, under the guise of “legitimate educational interest.”

• Admissions in the age of AI: detecting AI-generated application materials in higher education

• Even Microsoft Notepad is getting AI text editing now: Microsoft is introducing a new AI feature called Rewrite, which allows users to leverage generative AI to enhance their text editing experience by rephrasing sentences, adjusting tone, and modifying content length.

  

• Instagram Plans to Use AI to Catch Teens Lying About Age: Teen accounts, introduced by Meta in September 2024, are automatically set to private, restrict messaging to known contacts, and require parental approval for changes to settings. These features aim to enhance privacy and safety for users aged 13 to 17, providing a more secure online environment for those under 16.

  • Meta details 'adult classifier' tool for catching teens who lie about their age on Instagram: The adult classifier will estimate a user's age by analyzing several signals, including the accounts they follow, the content they frequently interact with, their profile information, and birthday posts from friends. If the tool suspects a user is under 18, it will automatically switch them to a teen account, regardless of the age they have listed on their profile.

• LinkedIn launches its first AI agent to take on the role of job recruiters: The AI can ingest informal notes and transform them into structured job descriptions, build pipelines of qualified candidates based on LinkedIn profiles, and even integrate with existing applicant tracking systems. This allows recruiters to leverage LinkedIn's vast dataset of over 1B users and 68M companies.

• T‑Mobile and OpenAI Join Forces to Revolutionize the Customer Experience: T-Mobile has committed to paying OpenAI approximately $100M over three years. IntentCX is designed to be the first intent-driven AI-decisioning platform in the telecommunications sector. It will utilize T-Mobile's extensive customer data to understand and anticipate customer needs, allowing for proactive problem resolution and personalized service.

• The doctors pioneering the use of AI to improve outcomes for patients: Faster diagnostics, more targeted treatment and better communication are among areas of healthcare already benefiting from artificial intelligence

• Meet the ‘super users’ who tap AI to get ahead at work: Adoption of AI at work is still relatively nascent. About 67 percent of workers say they never use AI for their jobs compared to 4 percent who say they use it daily.

Startups and Tools 🛠️

• Bret Taylor’s customer service AI startup just raised $175M: Sierra is dedicated to providing AI solutions that improve customer interactions. The platform allows businesses to customize AI personalities and integrates with various enterprise systems to automate customer service tasks. Taylor is the former co-CEO of Salesforce and current Chairman of OpenAI’s board.

• Buddy.ai is using AI and gaming to help children learn English as a second language: Buddy.ai employs a voice-based AI character that interacts with children in real-time, allowing them to practice speaking and listening skills in a stress-free environment.

• Decart’s AI simulates a real-time, playable version of Minecraft: Oasis is touted as the first playable open-world AI simulation that generates its environment dynamically based on player interactions. This allows for a unique gameplay experience each time a player engages with the game, as the world evolves in response to their actions.

Tech 💻 

• OpenAI has recently introduced a new feature called "Predicted Outputs", aimed at significantly reducing latency in AI-supported text processing: Predicted Outputs is a new tool from OpenAI that helps their AI models work faster when generating text - a way for the AI to get a head start on what it needs to say. It is particularly useful for faster responses (for editing code or updating text) and repetitive tasks.

• Google's 'Big Sleep' AI Project Uncovers Real Software Vulnerabilities: Google's 'Big Sleep' AI Project leverages large language models (LLMs) for proactive cybersecurity, simulating human vulnerability research to identify and trigger software weaknesses. Recently, it discovered a zero-day vulnerability in the widely used SQLite database engine.

• Google’s Gemini API and AI Studio get grounding with Google Search: This capability allows developers to integrate real-time data from Google Search into their AI applications, improving the accuracy and relevance of responses generated by Gemini models.

• More than a quarter of new code at Google is generated by AI

Safety and Regulation ⚖️

• Ars Technica on what a Trump victory could mean for US AI regulation:

  • AI Executive Order Rollback: Trump will likely roll back Biden's AI Executive Order from 2023, arguing that Biden's measures stifle innovation.

  • Trade Policies: Trump’s proposed tariffs, especially on Chinese imports, could restrict AI hardware supplies, potentially limiting access to GPUs. Export controls on AI tech to China may also intensify.

  • State-Level AI Regulation: If federal oversight declines, states like California and Tennessee may expand their own AI regulations to fill regulatory gaps.

• The Rise of AI in Defense:

  • Meta Permits Its A.I. Models to Be Used for U.S. Military Purposes: The Llama models will be accessible to various federal agencies and major defense firms, including Lockheed Martin and Booz Allen Hamilton. The integration of tech into military applications raises ethical questions and has historically faced opposition from employees within tech companies.

  • Anthropic Joins A.I. Giants to Provide Models to US Defense Agencies: Anthropic has teamed up with Palantir Technologies and Amazon Web Services (AWS) to supply its Claude AI models to U.S. intelligence and defense agencies. Claude 3 and 3.5 models will be integrated into Palantir's AI Platform (AIP).

• Researchers say an AI-powered transcription tool used in hospitals invents things no one ever said: OpenAI’s Whisper has been reported to fabricate entire sentences or phrases, including potentially harmful content such as racial commentary and fictitious medical treatments. Over 30,000 clinicians and 40 health systems are utilizing Whisper-based tools, which have been tailored for medical language by companies like Nabla.

• One in 20 new Wikipedia pages seem to be written with the help of AI

Other

• Netflix Bullish on Gen AI for Games After Laying Off Human Game Developers: Netflix appointed Mike Verdu as VP of GenAI for Games, emphasizing a new focus on generative AI technology in game development. He describes a "once in a generation inflection point" for game development and player experiences.

• Amazon’s Prime Video Launches AI-Generated TV Show Recaps, Promises They’re ‘Spoiler Free’: X-Ray Recaps utilizes generative AI to provide viewers with personalized, spoiler-free summaries of TV shows.

• The Alexa Skills revolution that wasn’t: Amazon’s “Skills” ecosystem was intended as an app alternative, letting developers build functions for Alexa. Despite early ambitions, Alexa remains primarily used for simple tasks (music, weather), failing to realize Amazon’s vision of a fully integrated “ambient computing” assistant.

• OpenAI acquired Chat.com: Chat.com now redirects users to its AI chatbot, ChatGPT. This acquisition reportedly cost around $15.5 million.