GSV's AI News & Updates (12/11/23)
Google releases Gemini, the EU's Landmark AI Act, Mistral AI raises $415M, Meta & IBM Create Industrywide AI Alliance, Mindsmith Pre-Seed, Amazon Q Hallucinations
We announced the inaugural ASU+GSV AIR SHOW: On April 13-15th we're launching ASU+GSV's inaugural AIR (AI Revolution) Show in San Diego, California. Done in conjunction with our 15th Annual ASU+GSV Summit, the AIR Show will be an immersive showcase of the world's most important AI in EDU companies. Tickets are complimentary for attendees.
General 🚀
Google introduces Gemini, the AI model it hopes will take down GPT-4: the highlights -
Google says Gemini beats GPT-4 in 30 out of 32 benchmarks.
There are 3 versions of Gemini: Ultra, the largest and most capable; Nano, which is significantly smaller and more efficient; and Pro, of medium size and middling capabilities.
Gemini’s clearest advantage comes from its ability to understand and interact with video and audio - Google hasn’t trained separate models for images and voice, unlike the approach OpenAI had to creating separate DALL-E and Whisper; it built one multisensory model from the beginning.
Gemini is a far more efficient model - it was trained on Google’s own Tensor Processing Units and is both faster and cheaper to run than Google’s previous models like PaLM.
Several demos illustrating Gemini’s ability to handle problems involving visual information - it drew images, created simple puzzles, solved math and physics homework, and offered game ideas involving a map of the world.
Separately, Google’s best Gemini demo was faked
E.U. Agrees on Landmark A.I. Act: After marathon ‘final’ talks which stretched to almost 3 days, EU lawmakers have clinched a political deal on a risk-based framework for regulating AI. The A.I. Act, sets a new global benchmark for countries seeking to harness the potential benefits of the technology, while trying to protect against its possible risks, like automating jobs, spreading misinformation online and endangering national security. It includes a total prohibition on the use of AI for:
biometric categorisation systems that use sensitive characteristics (e.g. political, religious, philosophical beliefs, sexual orientation, race);
untargeted scraping of facial images from the internet or CCTV footage to create facial recognition databases;
emotion recognition in the workplace and educational institutions;
social scoring based on social behaviour or personal characteristics;
AI systems that manipulate human behaviour to circumvent their free will;
AI used to exploit the vulnerabilities of people (due to their age, disability, social or economic situation).
OpenAI Agreed to Buy $51M of AI Chips From a Startup Backed by CEO Sam Altman: Documents show that OpenAI signed an LOI to spend $51M on brain-inspired chips developed by startup Rain in 2019. OpenAI CEO Sam Altman previously made a personal investment in Rain. Rain is working on a chip it calls a neuromorphic processing unit, or NPU, designed to replicate features of the human brain.
The Inside Story of Microsoft’s Partnership with OpenAI: Amidst the OpenAI chaos, Microsoft contemplated various strategies: ranging from stabilizing OpenAI's leadership to potentially absorbing OpenAI's talent into Microsoft. The saga culminated in Altman's reinstatement and a reconfiguration of OpenAI's board, underscoring Microsoft's crucial role in shaping the future of AI technology.
Celebrating the first year of Microsoft Copilot: new features that will roll out soon - Copilot will generate responses with GPT-4 Turbo, create images with the new DALL-E 3, multi-modal search, code interpreter and more.
Amazon’s Q has ‘severe hallucinations’ and leaks confidential data in public preview, employees warn: some hallucinations could ‘potentially induce cardiac incidents in Legal,’ according to internal documents.
Study gauges how people perceive AI-created content: when people knew a product’s source, they expressed a positive bias toward content created by humans. When people had no information about the source of the marketing or campaign copy, they preferred the results generated by AI.
Apple and Google avoid naming ChatGPT as their ‘app of the year’: and pick AllTrails and Imprint instead.
Education and the Future of Work 📚
Mindsmith Raises $550K Pre-Seed: Mindsmith is an AI-driven elearning tool for instructional designers that streamlines the process of development educational content. The company has found a way to improve upon an antiquated file format commonly used in L&D education called SCORM.
LEAD Introduces First-of-its-kind AI-powered Assessment For Schools: Offline-enabled and auto-synchronized, LEAD’s AI-powered assessments are powered by customizable inputs such as question type, composition, and difficulty level, etc., with options for teachers to review and edit as required.
Free New AI Tool to Help Americans Search and Compare Student Test Scores Across All 50 States: Zelma could democratize school performance data as schools struggle to recover from pandemic.
The Role of Artificial Intelligence in K-12 Professional Development: embracing the AI paradigm shift can truly help administrators contribute to building a future-ready teaching staff.
Generative AI in the Enterprise: a survey seeks to understand the real-world impact of these technologies in enterprises, revealing that 2/3 of respondents are already utilizing generative AI. While challenges like finding appropriate use cases and skills gaps remain, a majority (54%) anticipate the biggest benefit of AI to be enhanced productivity, not job reduction.
How GenAI Could Accelerate Employee Learning and Development: AI is emerging as a game-changer, offering tailored learning pathways, continuous content evolution, immersive simulation scenarios, and real-time skill gap analysis. Platforms like Sana Labs and Salesforce's Einstein GPT are at the forefront, creating dynamic, user-specific content, while AI tools like Workera are identifying and addressing skill gaps in real-time.
Gen Z driving early adoption of Gen AI: internet users aged 16+ are finding their way around this nascent technology. Most simply use generative AI for fun (58%), a third use it for work (33%) while a quarter (25%) turn to it for help with their studies.
Startups and Tools 🛠️
Mistral AI, a Paris-based OpenAI rival, closed its $415M funding round: “Since the creation of Mistral AI in May, we have been pursuing a clear trajectory: that of creating a European champion with a global vocation in generative artificial intelligence, based on an open, responsible and decentralized approach to technology”, says Mistral AI co-founder and CEO Arthur Mensch.
AssemblyAI raises a $50M Series C to build superhuman Speech AI models: AssemblyAI claims that 200k developers are building on their platform and handling more than 25M API calls a day. Its paying customer base grew 200% from last year to 4,000 brands.
Sarvam AI raises $41M in Series A funding: the five-month old startups has come out of stealth mode with Series A funding led by Lightspeed and Peak XV Partners and Khosla Ventures. Sarvam AI, which currently employs about 18 employees, is also focusing on building LLMs that use voice as the default interface in India. This strategy, combined with its emphasis on supporting local languages, aims to cater specifically to the Indian market’s requirements.
Replicate raises $40M Series B led by a16z: Replicate’s library of 25k+ open-source models have been used by more than 2M software developers. “There's been a larger interest in people switching to open source models because they don't want to be locked into a proprietary platform that might disappear at some point”.
Strut: the all-in-one AI workspace for writers. Strut combines tools to run your entire writing process - capture projects, notes, drafts in collaborative workspaces powered by AI.
Chatio.ai: AI chatbot for small businesses - answer 80% of support questions with AI.
WhatLetter: AI support for immigrant families and seniors to navigate paperwork without language barriers.
Searchie: your AI powered content assistant - helps you create engaging and insightful content for your files, automatically organizes and streamlines your content.
Tech 💻
Facebook researchers test AI's intelligence and find it is “unfortunately quite stupid”: the GAIA assessment, crafted by Meta's top researchers like "AI godfather" Yann LeCun, comprised 466 questions designed to be straightforward for humans but strenuous for AI. Humans scored 92%, GPT-4 managed just 15%, and its sibling, GPT4 Turbo, scored under 10%. Amidst these developments, rumors hint at OpenAI's work on a next-gen model, potentially named Q*, which could bring enhanced deductive reasoning.
Unpacking the hype around OpenAI’s rumored new Q* model: if OpenAI's new model can solve grade-school math, it could pave the way for more powerful systems.
Why Most Agent AIs Don't Work (Yet): Instead of creating a broad “this can do anything!” agents like AutoGPT, builders are building agents for specific types of goals in a particular domain.
An Opinionated Guide to Which AI to Use: ChatGPT Anniversary Edition
Evaluating and Mitigating Discrimination in Language Model Decisions
Regulation and Policy ⚖️
Beijing court’s ruling that AI-generated content can be covered by copyright: in contrast, US rulings have implied that AI creations cannot be copyrighted.
Meta, IBM Create Industrywide AI Alliance to Share Technology: the coalition, made up of more than 50 Founding Members of leading organizations across industry, startup, academia, research and government, are coming together to support open innovation and open science in AI.
Sports Illustrated Published Articles by Fake, AI-Generated Writers: Sports Illustrated reportedly used stories for product reviews that had authors it could not identify. Futurism found a picture of one author listed, Drew Ortiz, on a website that sells AI-generated portraits. Computer-generated writers … writing computer-generated stories?
Meta announces Purple Llama: Towards open trust and safety in the new world of generative AI
Beware Deepfake Scams: investing scams are starting to use more 'celeb-bait' AI deepfakery.
Other
The Mysterious Disappearance of Clyde AI: What Happened to Discord's AI Assistant?: a few months after the start of its limited experiment, Discord announced it was shutting down Clyde AI without explanations.
An Interview with Perplexity CEO, Aravind Srinivas: Aravind and his team are hot off the heels of a $500M funding round led by IVP.
Resources
The GSV BIG 10: weekly coverage corner for the top 10 stories, insights, and major plays in learning and skilling.
SO MUCH CONTENT!!!