From the Editor's Desk

Our Guide to the Summer 2024 Issue

Top Takeaways: As business uses of artificial intelligence, LLMs, and other algorithmic applications expand, so, too, does the possibility of unintended negative consequences for users and other stakeholders. The authors introduce two auditing frameworks — Ethical Matrix and Explainable Fairness — that can help organizations identify these potential risks and address critical questions about who could be harmed by algorithmic systems and how. They also discuss applying red teaming and benchmarking to difficult-to-audit LLMs, before applying all four of the approaches to a real-life example to demonstrate how an algorithmic audit could have prevented a very public chatbot failure.

Top Takeaways: A significant reason why machine learning projects fail to deliver business value is data scientists’ failure to adequately understand the business context. Development teams can avoid mistakes when they put aside any reticence to ask basic questions and engage with colleagues on the business side. The authors advise gaining input from all involved stakeholders and suggest some specific types of queries that might help machine learning developers get to the heart of the problem at hand.

Top Takeaways: Advanced analytics, such as predictive and prescriptive models to support business decisions, remain the primary drivers of data science value in the enterprise. How might the flashy, fluent, but not entirely reliable generative AI LLMs contribute to traditional analytics practice? The authors describe some experimental prompts that show potential for labeling data and explaining model predictions, and share guidance on monitoring and verifying that output.

Continued here

TradeBriefs: Newsletters for Decision-Makers!

Our advertisers help fund the daily operations of TradeBriefs. We request you to accept our promotional emails.

Want the newsletters, without the promotional mailers?
Get an (ad-free) subscription to TradeBriefs Premium for just $1 per month.