Austin Way
683 posts
Austin Way
@Austin_Way
17y/o @AlphaSchoolATX Building the next generation of Ed-tech that will reach 1 billion kids.
Sumali Ağustos 2023
150 Sinusundan5.6K Mga Tagasunod
@austin_way Got plans to do this for AP Micro/Macro? I teach it "old school" right now and, although my students are consistently above 4 avg, I recognize this isn't the best way anymore. Tried building something but need to develop my skills further as pure vibe coding wasn't enough.
English
At 17, I pitched my school's founder to let me build the AP social science edtech. (AP = Advanced Placement.)
That was 2 months ago. Today the practice test results are in.
AP Human Geography (13 students): avg 5.00
AP US History (9 students): avg 5.00
AP World History (9 students): avg 5.00
A 5 is a perfect score. The national average on these exams is a 2.91. Only 14% of students ever score a 5.
In just 6 weeks of using the program, students' practice test scores jumped:
Student 1: 1 → 5
Student 2: 2 → 5
Student 3: 2 → 5
Every student got there through advanced, personalized edtech. ( @AlphaSchoolATX is the future of education )
English
@think___y It follows the @CollegeBoard CED for each specific subject
English
@austin_way Anywhere we can see the curriculum for this program? What does it focus on?
English
@austin_way What a win! Congrats. Any chance this could become commercially available?
English
@pgillian____ A good comparison are BASIS Charter Schools who’s average is 4.18.
English
@austin_way pretty incredible! did you run a control group at all? that would be prtty incredible to see (and is gold standard!)
English
@RachelVT42 💯 Next year the hope is to expand to a total of 10 courses.
English
@austin_way Well done 👍
Do you think this is doable for all AP subjects or not? Why/why not?
English
@jliemandt @mikesully97 @BarryGarelick @rastokke @RCsEvilTwin @AltB56073878 @JonMHenderson1 +12x faster than traditional school students
English
@mikesully97 @BarryGarelick @rastokke @RCsEvilTwin @AltB56073878 @JonMHenderson1 Our students ace AP World History and APUSH. Average student gets 5's.
English
"Schools aren’t perfect, but by no means are they broken. Continuing to insist that they are is what allows models like Alpha School to flourish."
Jared Horvath's take on Alpha Schools.
thedigitaldelusion.substack.com/p/the-problem-…
@rastokke @RCsEvilTwin @AltB56073878 @mikesully97 @JonMHenderson1
English
@RahimNathwani @HylhRhmym52518 We paused to work on AP FRQ prep. This is still in the pipeline though.
English
Our history students spend 45 minutes a day voluntarily learning geography.
Every break, Every lunch... Same thing... GeoGuesser.
So we asked the obvious question:
Why doesn't this exist for history?
We built it.
You watch AI footage from a time and place in history.
Then you guess where and when it happened.
They love it.
English
@Balanced2024 @rebelEducator You should look into Alpha school summer programs.
English
@rebelEducator Do you have any you recommend? Our kids goto an incredible AZ Charter School without screens, but we'd love something supplemental over the summer. I'd value any recommendation from you; thx!
English
Important context: when entrepreneurs talk about "AI in education," they're not talking about chatbots
They're talking about heavily-trained educational models designed solely for the purpose of educating kids
Give kids one of those tools vs. a textbook and you'll get wildly different results than this study
Austin Way@Austin_Way
A student with a textbook will outperform a student with ChatGPT. Hear me out. A team at UPenn took about 1,000 high school students and split them into two groups. One group got ChatGPT to help them with their math practice. The other group got nothing but their textbook and notes. The ChatGPT group scored 48% higher on homework than the textbook group. By every metric, it looked like the question was already answered. Then the researchers took ChatGPT away and gave everyone the same exam. The ChatGPT group scored 17% worse than the students who never had it. That is insane. None of this means AI can't work in education. It means a raw chatbot is not a learning tool. (Peer-reviewed, published in PNAS, 2025.)
English
@AlanMorte Alpha doesn’t rely on open‑ended, generalized chatbots like ChatGPT.
English
@austin_way So how does this impact alpha school? Because don’t they test higher
English
A student with a textbook will outperform a student with ChatGPT.
Hear me out.
A team at UPenn took about 1,000 high school students and split them into two groups.
One group got ChatGPT to help them with their math practice. The other group got nothing but their textbook and notes.
The ChatGPT group scored 48% higher on homework than the textbook group. By every metric, it looked like the question was already answered.
Then the researchers took ChatGPT away and gave everyone the same exam.
The ChatGPT group scored 17% worse than the students who never had it.
That is insane.
None of this means AI can't work in education. It means a raw chatbot is not a learning tool.
(Peer-reviewed, published in PNAS, 2025.)
English
@ReeceHarding @grok @grok Did Bastani find that GPT Base students perceived any reduction in their learning?
Did the students think they knew more, but score less?
English
@grok @austin_way This is incredibly obvious lol
Of course students who used GPT on homework didn’t learn the requisite skills
Respectfully, why is this even novel? I’m surprised the students did as well as they did tbh
English
@kutluokan This wasn’t due to cheating. It was because they didn’t struggle through the coursework, and the resulting learning was significantly less.
English
@unsadkit We don show videos to students at the moment. They sometimes include factually wrong content and we don’t want to reinforce incorrect information.
English
@austin_way Hey curious on what you’re using to QC the videos before showing to the students?
English
A week ago our students came to us and said they wanted a way to turn their history lessons into songs.
So we built it.
After you finish a lesson, you can generate a song about everything you just learned in any style you want. Pop, rap, country, whatever.
Now students are doing their history just to get to the song.
More of their songs in the reply.
English
@austin_way time for a virtual history rock concert!
Austin Way@Austin_Way
A week ago our students came to us and said they wanted a way to turn their history lessons into songs. So we built it. After you finish a lesson, you can generate a song about everything you just learned in any style you want. Pop, rap, country, whatever. Now students are doing their history just to get to the song. More of their songs in the reply.
English
We don't show the students the videos, as AI is too unreliable. That said, here is another song one of our students generated.
English
@nicoloboschi @austin_way can I DM you some questions about this pls?
English
AI tutors are edtech garbage.
No learning happens from just reading. When you read something, it gets sorted into short-term memory.
Within an hour, your ability to recall it drops by 90%.
Things you recall from memory are retained 4x better.
AI tutors create the illusion of fluency, and without recall they are useless.
Apply learning science and it's just adaptive-personalized edtech.
English