Collaborative Intelligence Week in Review - 12Jul2024
Best business article(s) I read this week…
GenAI: Too much spend, too little benefit?
This research note from Goldman Sachs has more of an investment focus than a business focus. Still, it poses some thought-provoking questions and provides a bevy of informative data/analytics.
Best technical article(s) I read this week…
GenSQL: A Probabilistic Programming System for Querying Generative Models of Database Tables
MIT researchers have developed a generative model that sits on top of SQL, allowing people to detect anomalous data, fill in missing values, and generate realistic synthetic data for database tables. Their goal is to develop a chatbot that people can talk to about any database, with the bot replying based on answers derived from GenSQL queries.
The paper discusses the shortcomings of current AI agent benchmarks and evaluation practices and proposes solutions to improve their usefulness in real-world applications. I particularly like their focus on jointly optimizing cost and accuracy.
News items I found interesting…
Digital Twins Give Olympic Swimmers a Boost
The Summer Olympics will begin soon. Read how swimmers use digital twins to improve their techniques and race plans and reduce their times. One of the authors, guided by her digital twin, shaved 0.44 seconds off her 200M breaststroke and set an American speed record, breaking the previous record by 0.29 seconds.
20 Emerging Technologies That Will Change the Future
This is a nearly 90-minute YouTube video that explores such things as humanoid robots, programmable materials, and hypersonic weapons. I think it’s well done, but it is long.
How Generative AI is Changing the Mortgage Process
This article provides a specific example of the benefits and challenges of incorporating generative AI into a customer-facing process.
AI Breakthrough in Detecting New Particles at the Large Hadron Collider
Most readers will not be interested in this, but I am. One of my ML course instructors led the data mining team that participated in the discovery of the Higgs Boson in 2012. Researchers are now training AI to detect anomalous patterns in post-collision particle jets, possibly pointing to new physics.
The Economist Special Report: Will AI transform education?
I put this at the bottom of the list because a subscription is required for access. The authors argue that “visions of AI in the classroom misunderstand much about education.” The article begins with a 1913 quote from Thomas Edison, who forecasted that “books will soon be obsolete in schools” because he thought motion pictures would replace them. It’s Part 3 of a five-part report on education.
Coolest thing I saw…
I’ve done a lot of work in healthcare. As telemedicine’s growth continues, there are many opportunities for health tech companies (wearables and other devices) to provide reliable patient information to support these interactions and to support more frequent (or continuous in the case of wearables) patient monitoring. This device, expected to receive FDA approval soon, provides “a complete medical checkup in one minute.”
A company that caught my eye…
The company was voted most likely to succeed at VentureBeat’s 2024 Innovation Showcase. The company’s core product, AI Hub, ingests semi-structured, unstructured, and binary data and enables companies to query it securely.