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 #2: Understanding evals: LLM as a Judge
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 second talk in the series, Peter Van Dijck of the helpful intelligence company will show you how to create an eval for an AI product using an LLM as a judge (when we use a Large Language Model to evaluate the output of another Large Language Model). We’ll have a look at how that works, but also dig into why this even works. Are we creating problems for ourselves when we let an LLM judge itself? This talk is hands on; and there will be plenty of time for questions. You will go away understanding when and how to use LLM as a judge, and build 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 are a foundational feedback loop defining what 'good' means for AI products, helping to measure and improve systems continuously.
-
•
Evaluating fuzzy, subjective AI outputs requires innovative approaches such as using LLMs as judges to score results.
-
•
Binary (yes/no) scoring is more reliable than rating scales with ranges because LLMs lack internal memory and consistency.
-
•
Starting evals early (week one of a project) drastically improves AI product outcomes, but many teams delay due to perceived complexity.
-
•
High-risk or important tasks should be prioritized for evals instead of attempting broad coverage.
-
•
Assigning a dedicated owner or 'benevolent dictator' for evals who works closely with domain experts accelerates feedback and quality.
-
•
Creating a written constitution of principles helps concretize AI behavior goals and guides prompt and model training.
-
•
Most current eval tooling is too technical, slowing iteration cycles and making expert involvement inefficient.
-
•
Custom feedback interfaces tailored to expert users significantly speed up evaluating AI outputs in domains like healthcare and law.
-
•
Diverse perspectives from UX, product, strategy, and domain experts are critical in defining and refining what 'good' means in AI systems.
Notable Quotes
"Evals are everywhere, right? Everybody's talking about evals. It is like one of the key things in developing useful AI products."
"You want to ask an LLM to evaluate the fuzzy stuff because there’s no black and white output."
"LLMs don’t have memory, so rating on a scale from one to five is pretty random. Better to have yes or no answers."
"One of the biggest problems in AI building is evolving your prompts and having a fast feedback loop."
"By starting to categorize risk in detail, you naturally lead to better prompts and better evals."
"A constitution is a very good exercise: write down your system’s principles and values to help guide its behavior."
"Use custom systems for experts to quickly review and rate outputs, making feedback cycles much faster."
"Evals define a shared definition of good with tests to measure it, and that is the secret sauce for building great AI products."
"Model companies are students in a classroom wanting good points—they’re happy to run external expert evals to improve."
"The more I work with evals, the more I think UX and product people need to be involved because of the need for diverse perspectives."
Or choose a question:
More Videos
"Take your cross-functional partners out to lunch. Build relationships because people don’t resist change; they resist what they don’t know."
Greg PetroffSoftware as Material—A Redux
June 6, 2023
"The stories we tell are framed by the questions we ask. Our stories repeated become our beliefs and our beliefs influence our behavior."
Brendan JarvisFraming Tomorrow by Questioning Today
June 8, 2022
"When things behave the same way, users don’t have to worry about what will happen—that’s the intuitive factor."
Prerna MakanawalaAchieving Balanced Design Consistency
June 9, 2021
"Language matters — calling people 'users' reflects and reinforces the reality of reduced agency."
Tricia WangFrom Users to Shapers of AI: The Future of Research
March 25, 2024
"If 71% of people self-identify as gamers, why aren’t you learning from the video game industry?"
Cheryl PlatzEmbrace Your Fun Factor: Game Development Best Practices for Product Design
January 9, 2026
"There’s inherent conflict that will arise when navigating many stakeholders."
Ariel KennanTheme Two Intro
November 17, 2022
"Companies parachuting in to 'study' marginalized groups often do more harm than good when they don't build local capacity."
Tricia WangThe most popular design thinking strategy is BS
January 27, 2022
"We’re the voice of the employee—we’re arbiters of truth, defenders of experience, and sometimes validators against manipulation."
Kristin WisnewskiMeasuring What Matters
October 23, 2019
"Sometimes it’s reasonable to consider your design as strategy and ship a piece now and other pieces later."
Scott Jensen Sarah Delaney Carmen LiuShort Take #2: UX/Product Lessons from Your Industry Peers
December 6, 2022
Latest Books All books
Dig deeper with the Rosenbot
How should I choose communication channels and message formats to effectively share research insights?
How can research repositories score or show the impact of research recommendations to enhance organizational learning?
How does UX Tweak approach integrating AI features responsibly into their UX research platform?