In this section, Jessica Noss and Dylan Holmes, both teaching assistants in 6.034 Artificial Intelligence, discuss how a student-centered ethic shapes assessment and grading in the course. Professor Winston has also written about assessment in the course in an article entitled “Skills, Big Ideas, and Getting Grades out of the Way.”
Assessing Students’ Broad Understanding
In 6.034 Artificial Intelligence, our goal with quizzes is to test students’ understanding, not to trick them or stress them out. Our quiz policy exemplifies this principle in several ways.
We give students four short quizzes over the course of the semester, and we keep the quiz material fairly consistent from year to year so that students can use past quizzes to prepare for the quizzes in the current semester.
As for material, we emphasize fundamental conceptual understanding rather than cleverness or speed—we strive to write quiz questions that are as straightforward as possible. To calibrate the difficulty of each quiz, we make sure that the teaching assistants can complete the quiz in around half the time that we give students. That way, we know students will have enough time to complete the quiz without rushing. Lastly, by making all quizzes entirely open book and open notes, we ensure that we’re not testing students on memorization.
Prioritizing understanding over cleverness is part of the student-centered ethic that shapes the design of the course, as well as our interactions with students.
Grading with the Big Picture in Mind
As teaching assistants, our primary objective is to help students learn the material and to become inspired, and we go out of our way to make sure they have every opportunity to do so.
Our grading system supports the idea that we want to test students’ broad knowledge: We don’t want to split hairs about individual point differences, and we want students to be measured against their own understanding, not their peers’ performance.
For each quiz, we come together and decide what it would take for students to demonstrate different levels of understanding. We use the resulting thresholds to convert raw scores to a five-point scale. If a student is above the threshold for thorough understanding, that student gets a five on the exam; if a student is above the threshold for adequate understanding, that student gets a four, and so on.
This five-point system alleviates stress about individual point differences. For example, students often understand the material but make trivial calculation errors that cause them to lose points. In 6.034 Artificial Intelligence, if students are able to demonstrate a thorough level of understanding, they are guaranteed a five and aren’t penalized for insignificant mistakes.
We also offer a second-chance policy on quizzes via our final exam, which is optional. The sections of our final exam function as make-ups for each of the four quizzes. If students do poorly on a quiz, they have an opportunity to demonstrate understanding on the corresponding section of the final. Students may choose to do as many (or as few) of the sections of the final as they wish in the three-hour exam period, and we only replace their quiz scores with their corresponding final scores when it would improve their grade. As Professor Winston has written about in his article, “Skills, Big Ideas, and Getting Grades out of the Way,” the teaching staff members believe the final should serve as another opportunity for students to demonstrate mastery of the material. Our policy eliminates minor point differences and the pressure to perform well on a single quiz. Instead, we are able to focus on helping students learn the material, not specifically on when they do.
Promoting Student Wellness With a Relaxed Submission Policy
"[W]e try to keep in mind that students are always doing their best, even if it doesn’t look like our best."
We understand that students lead busy lives and sometimes stay up late to complete their programming assignments. Accordingly, we make our assignments due at 10:00 pm rather than midnight so that students are not incentivized to stay up as late.
Furthermore, we have a forgiving late policy: If a student turns in an assignment only a few minutes late, we don’t penalize the student at all, and if a student turns in an assignment several days late, we only gradually take off more points. In this way, our policy alleviates the pressure to choose between finishing an assignment and getting adequate rest.
Using the Student-Centered Ethic to Guide Communication
In 6.034 Artificial Intelligence, we go the extra mile to empathize with and respect our students. As a rule, we try to keep in mind that students are always doing their best, even it if doesn’t look like our best. After all, students have different individual skills and challenges; our position as teaching assistants requires us to meet students as they are and to find the best ways to help them improve.
One incident in particular characterizes this approach. We had a situation where a student was frustrated about a particular part of the course. The student sent us angry emails saying “This is really unfair. I have no idea how you expect us to do this. You should change your grading policy.”
We could have dismissed the student’s complaint, saying “I’m sorry. That’s just how we’ve decided to run the course.” Instead, in keeping with our student-centered ethic, we took the time to explain how we had designed our policy, and we volunteered to help the student succeed however we could. The student was surprised and grateful for our response, and in fact later sent us a kind message expressing appreciation for our patience and our willingness to help. Simply by showing that we cared about the student, we were able to bring about a 180-degree change in the student’s attitude and show that we were not out to get them.
We also employ an empathetic approach when helping students in office hours. When a student approaches us with questions, we avoid standardized explanations that might bore or further confuse the student. Instead, we make sure to assess the student’s background and baseline understanding to tailor our explanations. Our principle is that we can teach more effectively when we build on what the student currently understands.
These key interactions shape the culture of the course. It is through interactions like these—validating student concerns and tailoring explanations to students’ knowledge—that we put our empathetic ideals into practice.
