Looks Good to Me: Catching the AI Data Errors Nobody Questions
Details
Free Event | Thursday, October 29 | 2–3:30 p.m. CT
AI is awesome at handing over good-looking results. When it goes off
the rails, you notice. But when the results look perfect, that's when
you should worry.
Benjamin Schatz, Director of Data & Analytics at Craig Frames, builds
data platforms that put the entire warehouse in executives' hands,
and they run their own requests without going through him. So how
does he know the AI isn't just telling them what they want to hear?
Ben will break down why AI gets it wrong no matter how many times you
tell it to "make no mistakes," and the checks he uses to catch it:
adversarial agents, provenance tracking, request auditing, and user
prompt training.
Let's talk about what you're running into and how to keep it from causing
problems.
Add to Outlook Calendar:
https://bit.ly/4yG2eZA
Add to Google Calendar:
https://bit.ly/4i77UWt
About the Speaker: Ben Schatz
Ben Schatz is a self-described data nerd who loves figuring out how
things work. After leading analytics orgs at Google and Qualcomm, he's
now Director of Data & Analytics at Craig Frames, making good
numbers go up and bad numbers go down. And he's currently on the
Board of Directors at Minnestar.
He cares more about the big levers than the individual gears and is
focused on making insights actually lead to action. Outside of work,
he's a motorcyclist, a DIY'er, and a chronic tinkerer.
Agenda: Thursday, October 29
- 2 p.m. | Doors Open
- 2:15 p.m. | Talk Begins
- 3:30 p.m. | Event Wrap-Up
All our meetups are free. Food included. Whether you're an AI or data pro,
a tech enthusiast, or just curious, this one's for you. Grab your spot and
bring a friend.
Location:
Nagios Enterprises, 1295 Bandana Blvd N, Suite 310, St. Paul, MN 55108
(Enter through the main Atrium entrance and head to the third floor,
Suite 310.)
Free parking available in any non-reserved or "Nagios" spot; overflow
parking in the ramp.
Twin Cities Tech is a community hosted by Nagios Enterprises
(www.nagios.com).