Week 9: What's Happening in AI
EU Digital Services Act, CodeLlama and GPT-4 compete, brain-to-voice AI, Nvidia’s record earnings, and more.
Thanks for subscribing to the Responsible Innovation Labs Newsletter. We thought you may be interested in our new weekly series – check it out.
Welcome to this week’s edition of What's Happening in AI, covering the latest in policy, responsible AI, and where AI is going for the Responsible Innovation Labs community.
Below: EU Digital Services Act, CodeLlama and GPT-4 compete, brain-to-voice AI, Nvidia’s record earnings, and more. First, this week’s big story: Big Tech’s to-do list under new EU content rules and what’s next.
Why this matters: The EU Digital Services Act (DSA) sets rules for the responsibilities of digital service providers. ‘Micro and small companies’ will have obligations proportionate to size and ability. (more here)
What’s next: Full enforcement by February 17, 2024, with an August 25 start date for ‘very large’ companies
Consider: “[Fast growing companies] would benefit from a targeted exemption from a set of obligations during a transitional 12-month period” (more here)
Comments: Complying with the DSA (Google) and more from TikTok, Snap, and Meta
Policy
Americas:
Schumer to host AI forum with CEOs including Musk and Zuckerberg
SEC reforms private fund rules and brings its first NFT enforcement action
Corporate AI legal policies race to keep up (ex. Salesforce, Meta, Databricks)
NIST to standardize encryption algorithms that can resist attack by quantum computers
Europe:
Asia:
India’s PM Modi flags concerns over bias in AI; calls for ethical use
ICYMI: China to limit use of generative AI in online healthcare activities
Responsible AI
Health and science:
AI-powered biological software with Job Uszkoreit, Inceptive's Co-Founder
ChatGPT shows ‘impressive’ accuracy in clinical decision making
ICYMI: Scientific Advisory Board for independent advice announced by UN Secretary-General
Identity and security:
Workforce:
WhatsApp was able to support 50 billion messages a day with only 32 engineers
Behind the AI boom, an army of overseas workers in ‘digital sweatshops’
ICYMI: The Indian startup making AI fairer for workers (more here)
Resources:
Securing AI memo (Greylock)
Playbook for Universal Design (DTU Skylab)
Basics on board oversight of AI risks (Volkov Law Group)
3 trillion token open dataset for pretraining, Dolma (Allen Institute for AI)
Where AI is going
Compute and access:
Nvidia's knockout earnings give tech a giddy whiff of 1995 (more here)
Opinion: We must pass the Create AI Act (Stanford HAI)
Information and society:
Google’s SynthID, a watermark for AI images that you can’t edit out
Amazon, NYT, more are increasingly blocking OpenAI's GPTBot (more here)
Opinion: ISPs should not police online speech — no matter how awful (EFF)
Climate and efficiency:
IBM Research’s newest prototype chips use drastically less power
Intro: Liquid neural networks and reduced resource intensiveness
Releases and raises:
New tools and features: Google Workspace, Twilio, DoorDash
Raises: Arm's upcoming IPO, Hugging Face ($235M), Modular ($100M), EvolutionaryScale ($40M, biotech), Irreverent Labs (undisclosed), Ideogram ($16.5M seed)
What's happening in AI is shaped by our Charter themes and Responsible AI work.