Generative AI Weekly - Mar. 28, 2024

March 28, 2024

For this week:

  • Amazon Adds $2.75 Billion To Anthropic Investment
  • OpenAI Courts Hollywood in Meetings With Film Studios
  • Like we discussed a few months back - OpenAI is moving into monetizing GPTs
  • DBRX launches a new open-source model that is outperforming previous models. Notable that it is outperforming GPT 3.5…
  • A Q1 trends deck (must add email) from Octahedron (VC) which has a lot of great AI (and other) insights. Include macro-economic musings, sector specific insights, and a bunch on AI.
  • Similar themes as we have discussed in AI helping workers coming from a new paper on legal work. "We found that access to GPT-4 only slightly and inconsistently improved the quality of participants’ legal analysis but induced large and consistent increases in speed. AI assistance improved the quality of output unevenly—where it was useful at all, the lowest-skilled participants saw the largest improvements."
  • great read on how AI agents will integrate with existing apps and platforms. Feels like this assessment is right on: "1) The AI agent runs on my smartphone where it has access to my data. 2) It asks for my approval before using “tools” such as: search, book, purchase, message, etc. 3) Agent tools are implemented as app extensions. i.e. app developers include an extension a bit like a Share Sheet, or widgets, only it’s an on-device agent-accessible API for running tools"
  • Some counter-signals on sentiment around AI - and I think all appropriate!
  • Ethan Mollick's book Co-Intellgience is coming out soon - with "access to exclusive bonus material, including special GPTs to help apply the book’s lessons" if you pre-order.
  • Accenture's Q2 earnings note "We also had over $600 million in new GenAI bookings taking us to $1.1 billion in GenAI sales in the first-half of the fiscal year, expanding our early lead in GenAI, which is core to our clients reinvention. We now have over 53,000 skilled data and AI practitioners against our goal of doubling our data and AI workforce from 40,000 to 80,000 by the end of fiscal year 2026."
  • And lastly - a paper I am digging into this week on "simulating cultural evolution in populations of LLMs". How cool!!