GSV's AI News & Updates (11/15/24)
AI "Scaling Ceiling", OpenAI Operator Agent, Perplexity Ads, ChatGPT and Chegg, Stepful $31.5M Series B, Bhanzu raises $16.5M, North American Compact for AI
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 🚀
OpenAI, Google and Anthropic Are Struggling to Build More Advanced AI: Three of the leading artificial intelligence companies are seeing diminishing returns from their costly efforts to develop newer models. The belief that larger models and more data inevitably lead to better AI performance (scaling laws) is under scrutiny. Experts suggest that breakthroughs may require new training approaches and innovation beyond merely increasing scale.
OpenAI completed initial training for its new AI model, Orion, in September 2024. Orion underperformed in key areas, such as answering coding questions it wasn’t trained on, and fell short of the significant leap seen from GPT-3.5 to GPT-4.
Google's upcoming Gemini update and Anthropic’s Claude 3.5 Opus have also not met internal expectations, with delays and underwhelming performance relative to costs.
OpenAI and others seek new path to smarter AI as current methods hit limitations
OpenAI Nears Launch of AI Agent Tool to Automate Tasks for Users: “Operator” is set to launch in January 2025, designed to autonomously perform complex tasks such as coding, travel booking, and workflow organization with minimal user intervention.
ChatGPT App for Windows Now Available to Everyone: Paid subscribers can use voice mode to carry on a natural, back-and-forth conversation with ChatGPT. The app also syncs with your overall ChatGPT activity, so you can continue a conversation or view a past chat from the desktop app, mobile app, or website.
China’s Baidu joins Meta in race to make AI-integrated smart glasses: The glasses will include voice interaction, photo and video capture, and integration with Baidu services like Maps and its encyclopedia. They will also track health metrics and provide reminders for tasks.
Perplexity brings ads to its platform: This initiative involves the integration of sponsored follow-up questions that will appear alongside search results for U.S. users. These ads will be clearly labeled as "sponsored" to distinguish them from the main AI-generated content.
Education and the Future of Work 📚
How ChatGPT Brought Down an Online Education Giant: The launch of ChatGPT provided students with a free and instant alternative to Chegg's subscription-based model, leading to a loss of over 500,000 subscribers and a 99% decline in stock value since early 2021. Chegg collaborated with OpenAI to launch Cheggmate, integrating GPT-4 with Chegg’s proprietary content to generate instant answers and quizzes. However, the project failed to gain traction and was quietly shelved. Students reported dissatisfaction with Chegg’s AI-generated solutions compared to ChatGPT’s more comprehensive responses.
Harvard Medical School Alum Donates $6 Million for AI Health Care Education: The fund will support the expansion of the Artificial Intelligence in Medicine Ph.D. track and provide various programs and opportunities for students and researchers to integrate AI into clinical practice.
OpenAI releases “A Student's Guide to Writing with ChatGPT”: This guide aims to offer valuable insights into how students can use ChatGPT effectively while maintaining academic integrity and developing critical thinking skills.
Infosec Institute Launches AI-Powered Cybersecurity Training Tool: The tool customizes training modules based on individual employee roles and their specific vulnerabilities, ensuring that the content is relevant and engaging.
Research: How Gen AI Is Already Impacting the Labor Market: Research on freelancing platforms (July 2021–July 2023) shows gen AI tools like ChatGPT and image-generating AI caused immediate declines in demand for automation-prone jobs (31% decrease in Writing, 21% decrease in Software Development, 17% decrease in Graphic Design).
Growth of AI adoption slows among U.S. workers: Slack said its most recent survey found 33% of U.S. workers say they are using AI at work, an increase of just a single percentage point. Slack also found that globally, nearly half of workers (48%) said they were uncomfortable telling their managers they use AI at work. Among the top reasons cited were a fear of being seen as lazy, cheating or incompetent.
Startups and Tools 🛠️
Stepful raises $31.5M to address healthcare staffing shortage with AI-powered training: Stepful targets hourly workers and offers a cost-effective, fast alternative to traditional pathways like trade schools or community colleges, completing training in four months for an average cost of $2,500. AI monitors student progress and sends personalized motivational messages to those falling behind. If AI interventions fail, human coaches provide support, contributing to a 75% graduation rate.
Bhanzu Raises $16.5M to Expand India’s Premier Math Platform Globally: Founded in 2020 by Neelakantha Bhanu Prakash, who is recognized as the world's fastest human calculator, Bhanzu specializes in transforming math education for children aged 5 to 16 through experiential and personalized learning approaches.
ElevenLabs has recently launched Projects, a comprehensive tool designed for creating, editing, and generating long-form audio content with precision: This platform is particularly beneficial for creators, authors, and storytellers looking to transform written content into high-quality audio formats like audiobooks and voiceovers.
Tech 💻
AI groups rush to redesign model testing and create new benchmarks: Advances in AI models have surpassed the capabilities of current standardized benchmarks. These outdated tests fail to capture the complexity of reasoning and planning required for next-generation AI systems, especially autonomous agents. Companies like OpenAI, Microsoft, Meta, and Anthropic are creating internal benchmarks tailored to their systems.
We can all be AI engineers – and we can do it with open source models: Tools for building AI applications have advanced significantly, reducing barriers to entry.
Safety and Regulation ⚖️
OpenAI Says US Allies Should Partner on AI to Compete With China: OpenAI suggests forming a "North American Compact for AI" that would involve the United States and neighboring countries. They also propose establishing "AI Economic Zones" to accelerate permitting processes and bring nuclear reactors back online, while recommending that the US backstop costly energy infrastructure projects. This proposal comes at a time of increasing technological competition between the US and China. Both countries recognize the strategic importance of AI in shaping future economic and national security landscapes.
EU AI Act: Draft guidance for general purpose AIs shows first steps for Big AI to comply: Transparency requirements begin on August 1, 2025. Systemic risk assessments and mitigations are required by August 1, 2027, 36 months after the law’s entry into force.
Anthropic, feds test whether AI will share sensitive nuke info: Anthropic is working with the Department of Energy's nuclear specialists to ensure its models don't help people make weapons.
AI Startups, Investors Eagerly Await Less ‘Red Tape’ Under Trump: Practitioners expect more acquisitions and less oversight in a Trump administration expected to repeal Biden’s executive order. But Elon Musk’s influence is one question mark.
Not even Spotify is safe from AI slop: Artists have discovered unauthorized AI-generated albums appearing on their Spotify pages, leading to concerns about metadata violations and algorithm exploitation.
Meet 'Daisy' an AI granny that wastes phone scammers' time: Daisy's voice and character are crafted to engage scammers more effectively, appealing to their tendency to target vulnerable individuals. The AI keeps scammers on the line with long-winded stories and fictitious personal information, sometimes prolonging calls for up to 40 minutes.
Other
Spotify’s CTO on AI Generated Music, Podcasts, and Recommendations
NASA’s AI Earth Copilot will take your questions about our planet: The AI tool is designed to make it easier to access and understand NASA’s database of information.
AI protein-prediction tool AlphaFold3 is now more open: This move enhances its functionalities, including modeling protein interactions with biomolecules, and addresses previous criticisms regarding limited access, promising significant advancements in drug discovery and structural biology.
Painting by A.I.-Powered Robot Sells for $1.1 Million: The portrait depicts the British mathematician Alan Turing as the god of artificial intelligence. Its creator is a robot named Ai-Da that resembles a woman with a bob haircut.
