Iris, the AI Tutor
Iris is a tutor your students can ask questions of at any hour, inside the exercise they are working on. It has the exercise and their latest work as context, so a question does not have to be explained before it can be answered, and it is built to guide rather than to answer: Artemis will not let it hand over a complete solution however you configure it.
You decide two things per course — whether Iris is available at all, and how much help it gives.
Turning Iris On
Iris in the course management opens the settings, which are separate from the rest of the course configuration. The switch at the top decides whether every student in the course can use Iris; nothing below it applies while it is off.
Save Changes applies what you have set. Reset to Default puts the support level back to Moderate and clears the guidelines, and touches nothing else: whether Iris is enabled, and the administrative settings, are left as they are. It is a way to undo your tuning, not a way to switch Iris off.
Level of Instructional Support
This is the setting worth thinking about. It decides how much Iris gives away, on a scale of three, and the page shows you an example conversation at each so the choice is not abstract.
Low answers with counter-questions and nothing else, sending the student back to their own reasoning. It suits an advanced course where working the problem out unaided is the point of the exercise.
Moderate, the default, asks guiding questions and adds targeted hints. It suits a course where students have some grounding but benefit from a nudge and a pointer.
High explains step by step, with analogies and small code examples. It suits an introductory course where students are new to programming and an unexplained hint helps nobody.
Reading the three example conversations against each other is the quickest way to pick one. The same student question — a sorting algorithm that does not work — draws a counter-question at Low, a walk-through invitation at Moderate, and at High the actual off-by-one error named with the fix.
Additional Guidelines
Additional Guidelines is free text, up to 2048 characters, that sets expectations Iris follows in this course. It is for the things a general tutor cannot know: which language version the course uses, which of two conventions you teach, where students should be sent for background.
Use Java 21 syntax in any example. Refer students to the lecture on design patterns rather than to outside tutorials.
Keep it to instructions about how to answer. It is not a place to put course content, and it is not a way to make Iris stricter than the support level already makes it.
What Students See
Enabling Iris adds an Iris tab to the panel beside the problem statement on programming and text exercises, with the current exercise already set as the chat's topic. Other exercise types have no tab, and neither does an exam exercise. Iris can also answer about a lecture, and about the course as a whole.
Before anything else happens, each student is asked to choose whether they want to use cloud AI at all — a choice they make once, for themselves, and which you cannot make for them.
That means enabling Iris does not oblige anyone to use it, and a course cannot depend on every student having done so. See Iris for the student's side of it, which is worth reading before you set the support level.
Rate Limits and Variants
Two further settings sit on an Admin tab of the same page and are not yours to change: the pipeline variant Iris runs, and a rate limit of so many requests per timeframe. The page shows the effective limit for your course, so if students report Iris refusing to answer, that figure is the first thing to check with your administrator.