Campfire Session — Can't Fake It: Oral Assessments in the Age of AI with Jon Bergmann

00:00 Welcome

Teddy Lane introduces Jon Bergmann and explains why Flint cares about assessing learning as it happens, not only the finished product.

01:09 How Jon failed at AI

Jon opens with his own failure. When he first brought AI into his classroom, he built Flint assignments that AI would assess, and students did them at home with another AI. They got perfect grades and learned nothing, and his end-of-year evaluations were the worst he had ever received.

He concludes that a finished product can be drafted in any voice, so the evidence of learning has to move from a product to a conversation.

05:15 The Mastery Flip framework

Jon's Mastery Flip has three parts: students watch a flipped video or read the night before, class time is analog with laptops away, and learning is checked through oral assessments.

06:13 The Mastery Viva in three stages

Students arrive with paper notes and ask Jon an interesting question about them. At the end of each lesson, they explain their reasoning on a conceptual question, which gives Jon about six formative conversations per student in a unit. During the unit test, each student comes to his desk for a one-on-one summative viva.

When Jon can't reach every student in a lesson, Flint fills the gap.

08:57 Why oral assessments work

Misconceptions surface early, a conversation is hard to outsource, thinking becomes visible, and feedback is immediate. Jon also finds that vivas build relationships with students faster. He shares sample questions for science, English and history, and notes that AI is a good tool for writing deeper questions.

10:04 Clips: end-of-lesson vivas

In clips from Jon's geology class, one student struggles to explain why alluvium covers a valley floor. Jon takes the next student, who answers well, to a harder question that applies the idea to something bigger.

13:33 Clip: a summative viva

A student answering why earthquakes and volcanoes cluster along plate boundaries tries to say everything she has learned at once. The viva shows Jon that she hasn't organized the content well enough to answer the question, and he catches her misconceptions.

15:26 Clip: a group viva

Three physics students draw a free body diagram of a bungee jumper as the rope starts to slow them down. Jon steps away to work with other students, then returns and has them explain their thinking.

18:32 Rehearsing with Flint before a viva

Students rehearse out loud with Flint, both to make up a viva they missed in class and to practice for test vivas. Jon sets these activities so students can only talk, not type, because typed answers could be pasted in from another AI. He can listen to each student's recording.

22:01 Vivas with large classes

Keep each viva to about two minutes and rotate through students. With large classes, Jon suggests seeing half the class on even days and half on odd days, and pulling students aside during work time. Flipping the class frees the time, group vivas reach more students, and students can rehearse first on Flint or another tool.

23:04 Getting started, grading, and students who freeze

To start, pick one concept students get wrong, write one question, and talk to a few students. Jon's students now come prepared, and one who aced his viva credited repeated practice with Flint. Jon will require at least one Flint practice before each major test.

He grades formative vivas by whether a student understands, sending unprepared students back to the textbook to try again in ten minutes. Students who freeze get better with practice. Jon's one-pager on the two-minute oral check is at jonbergmann.com.

27:00 Q&A: personalizing for each student

Teddy asks how Jon personalizes for each student. Jon connects concepts to what students care about, such as density explained through muscle for a group of athletes, and uses AI when he needs help making those connections.

30:29 Skills students need with or without AI

Jon says students need to express their thinking in writing. He describes writing a magazine article with AI as a critic that pushed back but never wrote a sentence for him, and warns that students who hand their thinking to AI while their brains are still developing will become weaker thinkers.

33:49 Authentic thinking and summative vivas

Sohan Choudhury asks what Jon does when a student struggles in a summative viva. Jon says he usually walks them through the right answer, and admits he hasn't settled on the best approach.

37:43 Talking with students about AI

Teddy asks how Jon explains to students why he teaches this way. Jon wrote his students an open letter about AI and talks with them about it often, telling them that relationships, connection and creativity are the skills AI won't replace.

41:18 Learning from failure in the open

Sohan asks whether Jon has told his students about his own failure with AI. The conversation turns to letting students see adults fail and learn, including Jon relearning calculus, and to how a teacher's enthusiasm for a subject carries over to students.

48:59 AI handles practice, teachers hold the conversation

Sohan notes that in Jon's approach, AI does the routine work, such as hook questions and practice, so the conversations between teacher and student get richer. Jon says AI has made his teaching better, because he spends more time connecting with students.

51:55 Human skills in an AI world

Sohan argues that AI may make people better at being human, as routine work shifts to AI. Jon worries that AI could leave students less capable if they use it to skip the thinking, and the group agrees that human skills and relationships matter more than ever.

59:20 Audience Q&A: language in math and science

An attendee asks how to teach precise language in math and science without turning class into an English lesson. Jon uses whiteboards so students can show their thinking, and doesn't let students get away with vague words. Teddy adds that students can learn concepts before terminology.

1:03:58 Students who think before they speak

Some students process out loud and others process internally. Jon suggests letting internal processors think it through on a whiteboard and come back when they're ready.

1:06:00 Students with social anxiety

Vivas are one-on-one conversations, not speeches in front of the class, and most are low stakes. Jon says many small, low-stakes practices are what help anxious students.

1:08:28 English language learners

Jon notes that Flint has been popular with his school's world language teachers. Teddy suggests an activity where Sparky listens in and translates, so a student can explain an idea in their own language.

1:10:32 Wrap-up

Teddy thanks Jon and everyone who joined.