Back to science use cases

Science6th, 7th, 8thYear 7, Year 8, Year 9

How to turn feedback into a conversation in middle school science

How to get every student to review and improve their own lab answer, through a back-and-forth with Sparky instead of a comment they skim.

Title card reading 'How to turn feedback into a conversation in middle school science', next to three students' cards from a pendulum lab: a student whose claim had no data and who changed it using the table, a student who had data but no reasoning and added it, and a student who has not changed their claim yet.

"I created a bot using Flint that supports students in reviewing and improving their answers. This process fosters ownership of learning and helps students develop reflective skills—turning feedback into a conversation rather than one-way critique."

John SalgadoJohn Salgado, Middle School Science Teacher

Why this matters

In a 7th grade science class, one student writes that a heavier pendulum swings faster without checking the data. Another cites the right trials but never says why they matter. Both get the lab back days later with a comment in the margin, and the next lab answer has the same gap.

Feedback a student has to answer

In a study of 12 fifth and sixth grade classes, students who got comments without a grade did best on later tasks, and students who got a grade with the same comments did no better than a grade alone (Butler, 1988).

In Flint, each student talks through their own answer with Sparky, Flint's AI tutor. Sparky asks one question at a time, uses what the student is into outside class, and never rewrites or scores the answer. No teacher can have that conversation with 30 students about every lab answer.

One pendulum question, two students revising their answers

Every student gets the same question and the same table: the time for 10 swings with a 50 g and a 200 g bob, on 25, 50 and 100 cm strings. Everything after the question is one student's own chat.

Sparky asks about the student's own answerSparky uses the student's interest
M

Maya

First answer: Claim, no dataInterest: Basketball

Understanding check

Does a heavier bob make a pendulum swing faster? Give your claim, evidence from the table, and reasoning.

Maya

Yes. Heavier pendulums swing faster because gravity pulls harder on heavier things.

Sparky

Test your claim. Race a heavy bob against a light one on the same string.
An interactive Sparky made in the chat, titled Pendulum race. A 200 g bob and a 50 g bob hang on the same 50 cm string. The heavy bob took 14.3 seconds for 10 swings and the light bob took 14.2 seconds. A slider sets the string length from 25 to 100 cm, and a button runs the race again.
InteractivePendulum race

Maya

14.3 and 14.2 seconds. They basically tie.

Sparky

Drop a basketball and a tennis ball together. Which lands first? Does your claim still hold?

Maya

They land together. I'll change my claim: a heavier bob doesn't swing faster.

Changed her claim to match the data

J

Jordan

First answer: Data, no reasoningInterest: Drums

Understanding check

Does a heavier bob make a pendulum swing faster? Give your claim, evidence from the table, and reasoning.

Jordan

No. 50 g vs 200 g: 10.1 vs 9.9 s, 14.2 vs 14.3 s, 20.0 vs 20.2 s.

Sparky

Right trials. Why do 10.1 and 9.9 count as the same?

Jordan

0.2 seconds is about how long it takes to click a stopwatch.

Sparky

Put that in your reasoning. On a metronome, what happens to the beat when you slide the weight up?

Jordan

It slows down. So string length changes the time, and mass doesn't. I'll add that.

Added his reasoning, then found what does change the time

Every fix is in the student's own words. Sparky can make interactives like the pendulum race inside a student's chat when testing a claim beats reading about it. Maya and Jordan are examples we wrote for this page. Sparky learns a student's interests from the chat and can save them as memories, which students can see and delete in Settings.

See who fixed their answer and who is stuck

Sparky grades each finished chat against your rubric. The activity page shows who needs help right now and what the class still misses, and you can open any session to read the student's chat from the first answer to the last.

Flint's activity page for an activity called Pendulums: What Changes the Swing Time? Needs attention shows a live alert that one student has repeated the same claim three times without citing a trial from the table. Activity analytics lists what the class does well, an area of improvement where 7 students cite trials without explaining why they support the claim, and a follow-up with a button to create a follow-up activity. The Sessions table shows two submitted students graded Proficient and one still in progress.

Example activity page

After a pendulum lab in Science 7

The school and students are made up for this page. In this example, most of the class now says mass doesn't change swing time, and the skill to practice next is explaining why the data supports the claim.

Set up the activity for free

The pendulum activity is in Flint's public library with the data, question, guidelines and rubric written. Open it, select Try this activity to copy it into a free Flint account, and move it to a class. To test it as a student first, select Try session myself in the activity settings.

To use your own lab, attach its question and data in Guidelines for Sparky and keep these lines:

Example guideline

Ask the student what they're into outside class, then ask for a claim, evidence and reasoning. Don't rewrite or score their answer. Ask one question at a time about the part that's missing or doesn't match the data. When a claim doesn't match the data, make an interactive the student can test it with.

For more science activities, see Flint for science.

Other use cases

See all use cases →

Learning feels different when it fits you.