A Science Fair Investigation Into Artificial Intelligence
Who will come out on top?
Can artificial intelligence really outperform a 7th grade student? We built a Raspberry Pi-powered AI bot using Claude Sonet and Code and put it to the test with three different challenges! We choose Claude as our LLM (Language Learning Model) of choice.
Can AI correctly identify everyday objects like nails, coins, markers, and household items?
Can AI count coins and calculate the total value of change accurately?
Can AI answer fact-based trivia questions better than a 7th grader?
Before running the experiment, I made a prediction about how AI would perform compared to a 7th grader based on my research.
If the AI is coded correctly then it will be more accurate than a 7th grader. AI can be programmed to do specific tasks and can do calculations with minimal mistakes, unlike a 7th grade student who is likely to make mistakes. AI is predictable to be more accurate because they will be able to do the same thing the same way every single time.
AI is predictable to be more accurate because it can do the same thing the same way every single time.
7th grade students are likely to make mistakes due to rushing, nervousness, or simply forgetting information.
AI can be programmed to do specific tasks with minimal mistakes when coded correctly.
Think you're smarter than AI? Test yourself! Ask the AI any question a 7th grader should know, then see if YOU can answer it better.
Here's how we built and tested our AI bot against a 7th grade student to see who's smarter!
"Lucky Block" housing printed on Bambu Labs X-1 Carbon using Gold and White PLA (16hr 43min print time)
4GB RAM with Raspberry Pi OS (64-Bit) and 64GB SD Card running Python3
Raspberry Pi Camera V 1.3 for object and coin identification tests
Freenove 5" touchscreen monitor displaying the frontend built with Python3 tkinter
Amazon Basics 2.0 USB Powered Speakers and SunFounder USB 2 Mini Microphone
Python3 with tkinter for frontend, Anthropic Claude Vision API for AI processing images, Claude Sonnet to create the scripts, and Claude Code to implement the changes.
After 7,833 participants took our 20-question trivia quiz, here's how humans compared to AI across five subjects!
After testing 7,833 participants from across the United States, the results are in!
After 7,833 participants from across the United States, AI achieved a perfect 100% while humans averaged 67%. AI is the champion! Not a single human got all 20 questions correct.
AI achieved a perfect 100% score on all 20 trivia questions. The highest human score was 19/20 (95%) - no one got a perfect score!
Across all 7,833 participants, humans averaged 67% correct. Math was the strongest subject (71%) while History was the weakest (62%).
I predicted AI would be more accurate than a 7th grader on knowledge-based tasks, and the data proves it. AI's consistency gave it the edge.
For trivia and knowledge-based questions, AI is smarter than a 7th grader! AI's perfect memory and instant recall give it an unbeatable advantage. However, this doesn't mean AI is smarter at everything - humans still excel at creativity, emotional intelligence, and real-world problem solving.
If we could improve this project, we would use a higher quality camera than the one that came with the Raspberry Pi so the images are clearer. We would also collect more data by having other people try the tests, which would make our results stronger and more reliable.
We want parents to feel confident about their child's participation in our science fair study. Here's everything you need to know:
This is a 7th grade science fair project. The goal is to compare AI performance against students on various knowledge-based tasks.
No personal data is stored on our website. Registration is handled securely this website. Only first names and grade levels will be used in the science fair presentation - no full names, emails, or identifying information will be shared publicly.
Parents are welcome to sit with their child during the quiz - we ask that you do not help your child so that the results are authentic.
The session lasts approximately 30-45 minutes. Students will answer trivia-style questions across various subjects appropriate for middle school level.
Three participants will be randomly selected to win a $50 Visa gift card. Winners will be contacted via the email provided during registration. No purchase necessary.
Participation is completely voluntary. Students may withdraw at any time during the session without any penalty or consequence.
This is NOT a test of your child's intelligence or abilities. It's a fun, low-pressure trivia game to see how humans compare to AI. There are no grades, no scores shared publicly, and no right or wrong way to participate. It's all about having fun with science!
This is an independent student science fair project. Gift card prizes are provided by the student's family. By registering, you consent to your child's participation. Results will be used for educational purposes only.
Questions? Send an email to momanddad@isaismarterthanaseventhgrader.com
Thank you to all 7,833 participants who helped make this science fair project a success!
AI achieved a perfect 100% while humans averaged 67%. Check out the full results in the Results section above!
Students from across the United States took our 20-question trivia quiz.
Not a single human got all 20 questions correct. The highest human score was 19/20.
This project wouldn't have been possible without everyone who participated. You helped prove that AI is smarter than a 7th grader at trivia!