Log in or create a free Rosenverse account to watch this video.
Log in Create free account100s of community videos are available to free members. Conference talks are generally available to Gold members.
Hands-on AI #1: Let’s write your first AI eval
This video is featured in the Evals + Claude playlist.
Summary
If you’re a product manager, UX researcher, or any kind of designer involved in creating an AI product or feature, you need to understand evals. And a great way to learn is with a hands-on example. In this talk, Peter Van Dijck of the helpful intelligence company will walk you through writing your first eval. You will learn the basic concepts and the tools, and write an eval together. This talk is hands on; you can follow along, and there will be plenty of time for questions. You will go away with an understanding of the basic building blocks of AI evals, and with the confidence that you know how to write one. And more importantly, you’ll build some intuition, some product sense, around how the best AI products today are built, and how that can help you use them more effectively yourself.
Key Insights
-
•
Evals consist of a task, a golden dataset with known correct outputs, and an evaluator that measures correctness.
-
•
Manual AI prompt testing is slow and inconsistent; automated evals accelerate and scale evaluation.
-
•
UX and product teams can and should learn evals as a practical, non-technical skill.
-
•
Creating your own golden dataset is essential and cannot be outsourced or fully automated.
-
•
Models are fixed once trained; improvements happen by refining prompts and context design, not retraining the model.
-
•
Evaluations measure task performance, not the underlying model itself, allowing comparison across models.
-
•
Outputting a confidence score from models is unreliable due to lack of internal memory and inconsistent scale interpretation.
-
•
Biases are baked into models during training via evals used in post-training refinement.
-
•
LLMs can be used to judge other LLM outputs to evaluate tasks with non-binary answers.
-
•
Effective eval work requires collaboration across data analysts, engineers, subject matter experts, and UX/product teams.
Notable Quotes
"Evals are like a way to define what good looks like."
"The model was baked and once it’s baked, it does not learn again until they bake a new one."
"You need to be looking at the data. Nobody wants to, but that’s core work."
"Without a golden dataset, you have to build the golden dataset yourself."
"We’re not teaching the model anything; we’re improving our prompts and context."
"Confidence scores from the model are not a good idea because the model has no memory."
"Biases are baked in through the evals used during model training and post-training."
"LLMs judging other LLMs might sound crazy, but if you do it right, it works."
"Evals are a product and UX skill; learning them lets you make these systems do what you want."
"There is a large and growing capability overhang in these models we haven’t discovered yet."
Or choose a question:
More Videos
"Fletcher Prevend took us to the next level by reporting design and user research directly into the CIO office—a huge differentiator."
Kristin WisnewskiMeasuring What Matters
October 23, 2019
"Thousands of parents came with their children to play in the pop-up urban play space downtown."
Sarah AuslanderIncremental Steps to Drive Radical Innovation in Policy Design
November 18, 2022
"Peter van D**e really pushed us to think about the bot as part of a conversation, not just a query-response machine."
Louis RosenfeldThe Rosenbot and the Rosenverse: An AMA with Lou Rosenfeld
June 5, 2024
"We can all become experts at stuff if we work hard enough – sometimes you have to fake it till you make it."
Megan BlockerGetting to the “So What?”: How Management Consulting Practices Can Transform Your Approach to Research
March 26, 2024
"Nearly 70% of organizations use mixed-method research approaches, mainly focused on usability testing and validation."
Caroline VizeThe State of UX: Five Lessons from 2021 to Accelerate Digital Experience in 2022
March 9, 2022
"Pairing is two people with different roles doing an activity together simultaneously, like a driver and a navigator."
Christopher Taylor Edwards Valerie RoskeDesign as a Team Practice, A Practical Guide to Cross-functional Collaboration
September 30, 2021
"People didn’t want explanations about how to use patterns; they felt designs were self-evident and wanted just examples."
Eniola OluwoleLessons From the DesignOps Journey of the World's Largest Travel Site
October 24, 2019
"Linking data quality efforts to AI initiatives can help secure attention and budget for necessary improvements."
Caroline JarrettGarbage in, garbage out? Measuring error rates to get ready for AI
January 8, 2026
"I myself am a full-time screen reader user. I have been a screen reader user all my life, as I am completely blind."
Sam ProulxTo Boldly Go: The New Frontiers of Accessibility
March 11, 2022