2º IA, Robots & Automation - 1st April

News:

  • AI researcher Margaret Mitchell was fired by Google. Now she's the chief AI ethics scientist at one of its startup rivals. (The Information)
  • Norway-based 1X, which is developing humanoid robots, raised a $23.5M Series A2 led by the OpenAI Startup Fund with participation from Tiger Global and others (Will Knight / Wired)
  • An AI chatbots comparison test finds ChatGPT is the most verbally dextrous, Bing is best for getting information from the web, and Bard is surprisingly limited (The Verge)
  • LLMs will remove scholarly, creative, and economic software creation barriers, marking the industry's Gutenberg moment to unleash the next great technology wave (Irregular Ideas)
  • A look at the community of users who “jailbreak” GPT models to generate unfiltered content and see themselves as fighting back against OpenAI's closed policies (Chloe Xiang / VICE)
  • Q&A with Nvidia CEO Jensen Huang on AI's iPhone moment, the DGX Cloud service, the US' export controls, Nvidia's position in the stack in an LLM world, and more (Ben Thompson / Stratechery)
  • Replika restores erotic roleplay for users who signed up before February 1, after complaints from some who considered themselves “married” to chatbot companions (Anna Tong / Reuters)
  • Zoom partners with OpenAI to expand its Zoom IQ assistant with features to generate whiteboards from text prompts, recap meetings, summarize threads, and more (Emma Roth / The Verge)
  • A mid-March Midjourney update seemingly fixed one of the AI image generator's major failings: its inability to depict lifelike human hands with five fingers (Pranshu Verma / Washington Post)
  • A look at Modulate's ToxMod and other AI-based tools that help gaming companies moderate voice chats in online games, amid concerns of accuracy and user privacy (Sarah E. Needleman / Wall Street Journal)
  • Porn creators in Germany face fines and imprisonment as regulators use AI to spot porn not protected by age verification tech on social media and messaging apps (Matt Burgess / Wired)
  • A look at the non-consensual deepfake porn economy, with video sites easily accessible via top Google search results and some accepting Visa and Mastercard (Kat Tenbarge / NBC News)
  • Europol details how LLMs can be used to fuel fraud, cybercrime, and terrorism, and claims criminals are already misusing ChatGPT to carry out illegal activities (Katyanna Quach / The Register)
  • To compete with GitHub Copilot, Google partners with Replit, which used AI to make coding tools and will now rely on Google's LLMs for its Ghostwriter product (Dina Bass / Bloomberg)
  • Discounting AI's short-term risks, from phishing to fraud to propaganda, because artificial general intelligence is not here yet leaves society ill-prepared (Gary Marcus / The Road to AI We Can Trust)
  • Cerebras open sources seven GPT-based LLMs, ranging from 111M to 13B parameters and trained using its Andromeda supercomputer for AI, on GitHub and Hugging Face (Mike Wheatley / SiliconANGLE)
  • Plugins turn ChatGPT into an aggregator, signaling a major shift in ambition that puts OpenAI well on its way to becoming the next major consumer tech platform (Stratechery)
  • Sources: Google Brain and DeepMind have been forced to work together on a project known as Gemini to compete with OpenAI's GPT-4, after Bard's stumble (Jon Victor / The Information)
  • The UK government publishes recommendations for AI and tasks regulators with issuing practical guidance to organizations on implementing them in their sectors (Ryan Browne / CNBC)
  • As Microsoft, Meta, Google, Amazon, Twitter, and others cut their responsible AI teams, experts worry about potential abuses, disinformation, and hallucination (Financial Times)
  • Report: Google AI engineer Jacob Devlin quit to join OpenAI in January 2023 after complaining Bard was being trained on ChatGPT data; Google denies the charge (Sean Hollister / The Verge)
  • Some signatories of the open letter to pause AI training walk back their positions, others turn out to be fake, and many experts disagree with its proposal (Chloe Xiang / VICE)
  • Synopsys claims to launch the first full-stack AI-powered EDA suite that covers all stages of chip design, including architecture, design, and manufacturing (Anton Shilov / AnandTech)
  • BuzzFeed quietly started publishing fully AI-generated SEO-driven travel posts by non-editorial staff on March 14; a spokesperson calls the posts an experiment (Futurism)
  • Q&A with Vinod Khosla on how AI could “free humanity from the need to work”, seeing the AI future early, his OpenAI investment, AI and geopolitics, and more (Reed Albergotti / Semafor)
  • FreedomGPT, a chatbot from Austin VC firm Age of AI, is free of ChatGPT's safety filters and ethical guardrails, and praises Hitler, uses the N-word, and more (Pranav Dixit / BuzzFeed News)

Startups raising funds

  • Perplexity AI, an SF-based answer app, raised $25.6m in Series A funding. NEA led, and was joined by Databricks Ventures, Elad Gil, Bob Muglia (former CEO of Snowflake) and Paul Buchheit. Link
  • FedML, a Sunnyvale, Calif.-based collaborative AI startup, raised $6m. Camford Capital led, and was joined by Plug & Play Ventures, AimTop Ventures, Acequia Capital and LDV Partners. Link
  • Irrigreen, an Edina, Minn.-based robotic irrigation startup, raised $15m in seed funding. Ulu Ventures led, and was joined by with Sage Hill Investors, Burnt Island Ventures, MFV Partners, Anorak Ventures, Echo River Capital, Tamiami, Catalyst Innovation Lab and Sum Ventures. Link
  • Deep Render, a London-based video compression by AI startup, raised $6.3m (+2,7 grant) in Series A funding co-led by IP Group and Pentech Ventures. Link
  • Parloa, a German contact center automation startup, raised €20m in Series A funding. EQT Ventures led, and was joined by Newion and Senovo. Link
  • Fixie, a Seattle-based automation platform for large language models used by enterprises, raised $17m in seed funding. Redpoint Ventures led, and was joined by Madrona Venture Group, Zetta Venture Partners, SignalFire, Bloomberg Beta and Kearny Jackson. Link
  • Stratyfy, a New York-based provider of machine learning solutions for lenders, raised $10m. Truist Ventures and Zeal Capital Partners co-led, and were joined by Mendon Venture Partners, The 98, FIS, and Barry J. Glick. Link
  • Jigso, an Israeli AI observability platform, raised $7.5m in seed funding from General Catalyst, Entree Capital and Jibe Ventures. Link

Read the news of digital businesses by email each day at www.multiversial.es