Collaborative Intelligence Week in Review - 29Mar2024

Mar 29, 2024

Mayo Clinic’s Healthy Model for AI Success
This article uses examples from healthcare, but its theme is applicable across industries. It focuses on what the authors term “AI Enablement” and touches on data management strategies and generative AI. (Full disclosure: I’ve worked with one of the authors.)

MemGPT: Towards LLMs as Operating Systems
LLMs have a limited amount of contextual memory, which is why they tend to drift when producing lengthy outputs. They also can suffer from what is known as “catastrophic forgetting” (forgetting something very important that occurred earlier, so they fail to produce things that are on point). The authors have introduced an approach that enhances the LLM with virtual memory tiers, enabling them to maintain context over longer sequences. For those of you so inclined, here is the associated GitHub repository.

Generative AI and the Risk Of Inbreeding
Explores the impact on future versions of generative AI of their being trained on the rapidly expanding corpus of text and images produced by generative AI rather than by humans.
8 Google Employees Invented Modern AI. Here’s the Inside Story
This lengthy feature from Wired Magazine describes how a Google team came together and eventually developed the transformer architecture that has revolutionized generative AI.
3 insights from nonprofits about generative AI
This is a short article that reveals the top three insights from Google’s survey of more than 4000 not-for-profit organizations on how they’re using AI and what they’ve discovered.
US Digital Twin market forecast to grow to $150B by 2030
Grand View Research has released a new research report on digital twins. I was surprised by the rapid growth they forecast.

VIAM
The company provides a software platform for quickly building intelligent devices, such as robots. The platform has been available since May 2023. They just received $45M in Series B funding.

High-Impact AI: Strategies & Tactics for Leaders