Collaborative Intelligence Week in Review - 12Apr2024
Best business article(s) I read this week…
Empowering AI Companions for Enhanced Relationship Marketing
The article presents an analysis of current chatbot and virtual assistant capabilities and describes how advances in these technologies could lead to “AI Companions” that develop and strengthen customer relationships.
If you don’t want to read the “book” (26-page journal article), perhaps you’ll like the movie:
Best technical article(s) I read this week…
Categorical Deep Learning: An Algebraic Theory of Architectures
I discovered this article by reading about Symbolica (see below), an AI start-up that emerged from stealth mode this week. It presents a novel framework based on category theory that details a unified approach to specifying and studying deep learning architectures. The framework provides a means of understanding and analyzing how AI systems model and manipulate data.
News items I found interesting…
What Business Leaders Really Think About Generative AI
This is an overview of a survey by INSEAD of 1200 executives regarding their uses of and attitudes toward generative AI. Click here for the entire survey.
Why Does ChatGPT Use ‘Delve’ a Lot?
AI Phrase Finder maintains an updated database of the words and phrases most used by ChatGPT and other LLMs. As they write, “You would think that a chatbot that crawls the open web would have developed one hell of a vocabulary. But judging by our dataset of 50,000 ChatGPT responses, it has an overwhelming preference for certain words.” They want you to eliminate these words from your writing, lest people realize or erroneously conclude that you’ve used an LLM to produce your content.
Digital Twins And Digital Threads: The Future Of Customized Health
An interesting view of one possible healthcare future wherein a digital twin of a patient is continually updated from a continuous flow of patient data. This combination will enable the testing of different treatment plans virtually before they are applied in real life and other applications of personalized medicine.
Coolest thing I saw…
Wearables are a CI category that I do not write about often. They fit into the “Human Oversight” category of applications. Applications in this category assign an intelligent machine a task and humans the task of monitoring or acting on the machine's outputs. In this case, the biosensor tracks hydration levels in your body by analyzing your perspiration. It alerts you via a smartphone interface when you need to hydrate. It tells you when you should drink, what you should drink, and how much you should drink to help ensure peak hydration.
A company that caught my eye…
This start-up, founded by a former Tesla senior engineer, is using category theory to build AI models. The company claims this approach will give AI inherent reasoning ability and interpretability. If successful, the approach will enable AI to be applied to many more problem domains.