Collaborative Intelligence Week in Review - 21Jun2024
Best business article(s) I read this week…
When AI Comes for Knowledge Workers
This article from Boston Consulting Group (BCG) imagines a future “where every business forecast, analysis, and decision can be made by AI.”
Transforming Consulting Through Generative AI
This Harvard Business Review Analytics Services white paper was sponsored by Salesforce. It describes how consulting firms are using generative AI, and then using the insights they’ve gained to inform their generative AI-focused client work.
Best technical article(s) I read this week…
Knowledge Graphs in the Digital Twin
The authors explore how the combination of knowledge graphs and digital twins could lead to more scalable and cost-effective development of simulations based on digital twins. As I’ve written before, while each collaborative intelligence technology has value, much more value can be derived from combining them.
Retrieval-Augmented Generation–Enabled GPT-4 for Clinical Trial Screening
This is an interesting use case for fine-tuned LLMs. The authors describe how they built customized GPT models for screening patients with symptomatic heart failure. The models matched or outperformed clinicians. Also of interest is their use of these models to extract the information required from unstructured clinical notes in Electronic Health Records (EHRs), a difficult challenge to overcome.
Other items I found interesting/valuable…
How AI and Digital Twins are Enhancing Access to Healthcare
This is a video presentation by Dileep Mangsuli, an Executive Director of Siemens Healthineers. (Disclosure: I have completed consulting engagements for Siemens Heathineers and their precursor organizations.)
Generative AI’s killer enterprise app just might be ERP
According to the author, the combination of generative AI and ERP systems could lead to enhanced profitability, customized ERP apps, and other things. It’s another creative and valuable application of generative AI.
NVIDIA Releases Open Synthetic Data Generation Pipeline for Training Large Language Models
Companies attempting to build custom LLMs often struggle to amass the data required to produce high-quality results. NVIDIA’s new Nemotron-4 340B family of models can help. According to the company, “Through a uniquely permissive open model license, Nemotron-4 340B gives developers a free, scalable way to generate synthetic data that can help build powerful LLMs.”
Forrester survey finds that US ad agencies are leading GenAI adoption
The full report is behind a paywall, but the summary has a few interesting data points regarding how generative AI is being deployed in creative agencies.
Coolest thing I saw…
This is a LinkedIn post with a video of a working robotic submarine built from Lego parts.
A company that caught my eye…
Former OpenAI Chief Scientist Ilya Sutskever announced he has formed a new company that aims to develop safe superintelligence technology.