Jul 27, 2023 · 39m · a16z

Classroom 2050: Unleashing AI, XR, Gaming

Sal Khan · 11m spoken Steph Smith · 8m spoken Anarupa Ganguly · 7m spoken Allison Matthews · 4m spoken Romy Drucker · 4m spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this episode of the a16z Podcast recorded at the Aspen Ideas Festival, host Steph Smith moderates a panel with leaders Sal Khan, Anarupa Ganguly, Allison Matthews, and Romy Drucker to explore how artificial intelligence, virtual reality, and gaming can overcome post-pandemic learning loss and modernize education.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The host holds 22.4% of the talking time here. How this is scored →

The host as informed peer 1.9 Guest teaching 2.3 Guest disagreement 0.1 The host pushing back 0.1
05100:0010:0020:0030:002:32–5:11 · The host as informed peer 2/10 Legal Disclosures & Title Sequence Steph Smith introduces the panel and sets up the premise of the episode with an audience show-of-hands before asking Romy Drucker for baseline pandemic learning loss data. Romy provides sober stats on math and reading decline in a collaborative manner.5:11–7:32 · The host as informed peer 1/10 Prisms VR and Spatial Computing in STEM Steph asks Anarupa Ganguly about STEM talent development. Anarupa educates the audience and host on spatial reasoning and abstraction using Isaac Newton's gravity example.7:32–12:28 · The host as informed peer 1/10 Khanmigo: AI-Powered Socratic Tutoring Sal Khan delivers an extended explanation of Khanmigo and how AI can act as a Socratic tutor rather than an answer generator. The host listens as Sal details guardrails and partner interactions with OpenAI.12:28–16:23 · The host as informed peer 2/10 Minecraft Education: Gamification and Intrinsic Motivation Steph prompts Allison Matthews on gaming in education and then Anarupa on common misconceptions. Anarupa reframes tech adoption, noting that focus should be on pedagogy rather than hardware.16:23–19:37 · The host as informed peer 3/10 Rethinking AI: From Cheating Tool to Co-Creator Steph directly prompts Sal regarding cheating concerns with AI. After Sal outlines AI co-creation, Steph offers a helpful parallel from her own educational exchange in Sweden.19:49–23:16 · The host as informed peer 2/10 Allison Matthews on Gaming, Minecraft, and Evolving Teacher Roles Steph asks Allison about the evolving role of teachers when gaming is introduced. Allison explains scaffolded learning and how Minecraft fosters growth mindsets and digital citizenship.23:16–25:52 · The host as informed peer 2/10 Romy Drucker on Student Agency, Inclusive R&D, and Educational Access Steph asks Romy how outside stakeholders can support classroom innovation. Romy details student agency and inclusive R&D models with families and teachers.25:52–30:14 · The host as informed peer 2/10 Anarupa Ganguly on VR Hardware Deployment, Adoption, and Supporting Teachers Steph questions how VR hardware costs and school board adoption are handled. Anarupa strongly rejects low expectations placed on teachers by district leadership, citing high opt-in rates.30:14–35:23 · The host as informed peer 2/10 Sal Khan on Addressing Educational Deficits with AI and Immersive Learning Steph asks Sal about the broader timeline and future impact of AI. Sal outlines sobering math proficiency statistics and connects generative AI to sci-fi visions of personalized learning.35:23–39:07 · The host as informed peer 2/10 Panel Discussion: Final Calls to Action for Education Steph guides the panel through closing calls to action. The guests emphasize collaborative execution and non-partisan commitment to student success.2:32–5:11 · Guest teaching 1/10 Legal Disclosures & Title Sequence Steph Smith introduces the panel and sets up the premise of the episode with an audience show-of-hands before asking Romy Drucker for baseline pandemic learning loss data. Romy provides sober stats on math and reading decline in a collaborative manner.5:11–7:32 · Guest teaching 3/10 Prisms VR and Spatial Computing in STEM Steph asks Anarupa Ganguly about STEM talent development. Anarupa educates the audience and host on spatial reasoning and abstraction using Isaac Newton's gravity example.7:32–12:28 · Guest teaching 2/10 Khanmigo: AI-Powered Socratic Tutoring Sal Khan delivers an extended explanation of Khanmigo and how AI can act as a Socratic tutor rather than an answer generator. The host listens as Sal details guardrails and partner interactions with OpenAI.12:28–16:23 · Guest teaching 3/10 Minecraft Education: Gamification and Intrinsic Motivation Steph prompts Allison Matthews on gaming in education and then Anarupa on common misconceptions. Anarupa reframes tech adoption, noting that focus should be on pedagogy rather than hardware.16:23–19:37 · Guest teaching 2/10 Rethinking AI: From Cheating Tool to Co-Creator Steph directly prompts Sal regarding cheating concerns with AI. After Sal outlines AI co-creation, Steph offers a helpful parallel from her own educational exchange in Sweden.19:49–23:16 · Guest teaching 2/10 Allison Matthews on Gaming, Minecraft, and Evolving Teacher Roles Steph asks Allison about the evolving role of teachers when gaming is introduced. Allison explains scaffolded learning and how Minecraft fosters growth mindsets and digital citizenship.23:16–25:52 · Guest teaching 2/10 Romy Drucker on Student Agency, Inclusive R&D, and Educational Access Steph asks Romy how outside stakeholders can support classroom innovation. Romy details student agency and inclusive R&D models with families and teachers.25:52–30:14 · Guest teaching 3/10 Anarupa Ganguly on VR Hardware Deployment, Adoption, and Supporting Teachers Steph questions how VR hardware costs and school board adoption are handled. Anarupa strongly rejects low expectations placed on teachers by district leadership, citing high opt-in rates.30:14–35:23 · Guest teaching 4/10 Sal Khan on Addressing Educational Deficits with AI and Immersive Learning Steph asks Sal about the broader timeline and future impact of AI. Sal outlines sobering math proficiency statistics and connects generative AI to sci-fi visions of personalized learning.35:23–39:07 · Guest teaching 1/10 Panel Discussion: Final Calls to Action for Education Steph guides the panel through closing calls to action. The guests emphasize collaborative execution and non-partisan commitment to student success.2:32–5:11 · Guest disagreement 0/10 Legal Disclosures & Title Sequence Steph Smith introduces the panel and sets up the premise of the episode with an audience show-of-hands before asking Romy Drucker for baseline pandemic learning loss data. Romy provides sober stats on math and reading decline in a collaborative manner.5:11–7:32 · Guest disagreement 0/10 Prisms VR and Spatial Computing in STEM Steph asks Anarupa Ganguly about STEM talent development. Anarupa educates the audience and host on spatial reasoning and abstraction using Isaac Newton's gravity example.7:32–12:28 · Guest disagreement 0/10 Khanmigo: AI-Powered Socratic Tutoring Sal Khan delivers an extended explanation of Khanmigo and how AI can act as a Socratic tutor rather than an answer generator. The host listens as Sal details guardrails and partner interactions with OpenAI.12:28–16:23 · Guest disagreement 0/10 Minecraft Education: Gamification and Intrinsic Motivation Steph prompts Allison Matthews on gaming in education and then Anarupa on common misconceptions. Anarupa reframes tech adoption, noting that focus should be on pedagogy rather than hardware.16:23–19:37 · Guest disagreement 0/10 Rethinking AI: From Cheating Tool to Co-Creator Steph directly prompts Sal regarding cheating concerns with AI. After Sal outlines AI co-creation, Steph offers a helpful parallel from her own educational exchange in Sweden.19:49–23:16 · Guest disagreement 0/10 Allison Matthews on Gaming, Minecraft, and Evolving Teacher Roles Steph asks Allison about the evolving role of teachers when gaming is introduced. Allison explains scaffolded learning and how Minecraft fosters growth mindsets and digital citizenship.23:16–25:52 · Guest disagreement 0/10 Romy Drucker on Student Agency, Inclusive R&D, and Educational Access Steph asks Romy how outside stakeholders can support classroom innovation. Romy details student agency and inclusive R&D models with families and teachers.25:52–30:14 · Guest disagreement 1/10 Anarupa Ganguly on VR Hardware Deployment, Adoption, and Supporting Teachers Steph questions how VR hardware costs and school board adoption are handled. Anarupa strongly rejects low expectations placed on teachers by district leadership, citing high opt-in rates.30:14–35:23 · Guest disagreement 0/10 Sal Khan on Addressing Educational Deficits with AI and Immersive Learning Steph asks Sal about the broader timeline and future impact of AI. Sal outlines sobering math proficiency statistics and connects generative AI to sci-fi visions of personalized learning.35:23–39:07 · Guest disagreement 0/10 Panel Discussion: Final Calls to Action for Education Steph guides the panel through closing calls to action. The guests emphasize collaborative execution and non-partisan commitment to student success.2:32–5:11 · The host pushing back 0/10 Legal Disclosures & Title Sequence Steph Smith introduces the panel and sets up the premise of the episode with an audience show-of-hands before asking Romy Drucker for baseline pandemic learning loss data. Romy provides sober stats on math and reading decline in a collaborative manner.5:11–7:32 · The host pushing back 0/10 Prisms VR and Spatial Computing in STEM Steph asks Anarupa Ganguly about STEM talent development. Anarupa educates the audience and host on spatial reasoning and abstraction using Isaac Newton's gravity example.7:32–12:28 · The host pushing back 0/10 Khanmigo: AI-Powered Socratic Tutoring Sal Khan delivers an extended explanation of Khanmigo and how AI can act as a Socratic tutor rather than an answer generator. The host listens as Sal details guardrails and partner interactions with OpenAI.12:28–16:23 · The host pushing back 0/10 Minecraft Education: Gamification and Intrinsic Motivation Steph prompts Allison Matthews on gaming in education and then Anarupa on common misconceptions. Anarupa reframes tech adoption, noting that focus should be on pedagogy rather than hardware.16:23–19:37 · The host pushing back 0/10 Rethinking AI: From Cheating Tool to Co-Creator Steph directly prompts Sal regarding cheating concerns with AI. After Sal outlines AI co-creation, Steph offers a helpful parallel from her own educational exchange in Sweden.19:49–23:16 · The host pushing back 0/10 Allison Matthews on Gaming, Minecraft, and Evolving Teacher Roles Steph asks Allison about the evolving role of teachers when gaming is introduced. Allison explains scaffolded learning and how Minecraft fosters growth mindsets and digital citizenship.23:16–25:52 · The host pushing back 0/10 Romy Drucker on Student Agency, Inclusive R&D, and Educational Access Steph asks Romy how outside stakeholders can support classroom innovation. Romy details student agency and inclusive R&D models with families and teachers.25:52–30:14 · The host pushing back 1/10 Anarupa Ganguly on VR Hardware Deployment, Adoption, and Supporting Teachers Steph questions how VR hardware costs and school board adoption are handled. Anarupa strongly rejects low expectations placed on teachers by district leadership, citing high opt-in rates.30:14–35:23 · The host pushing back 0/10 Sal Khan on Addressing Educational Deficits with AI and Immersive Learning Steph asks Sal about the broader timeline and future impact of AI. Sal outlines sobering math proficiency statistics and connects generative AI to sci-fi visions of personalized learning.35:23–39:07 · The host pushing back 0/10 Panel Discussion: Final Calls to Action for Education Steph guides the panel through closing calls to action. The guests emphasize collaborative execution and non-partisan commitment to student success.

speaking balance: gold is the host, purple is the guest (3 minute bins)

0:00 · the host 84.3% · guest 15.7%0:00 · the host 84.3% · guest 15.7%3:00 · the host 47.3% · guest 52.7%3:00 · the host 47.3% · guest 52.7%6:00 · the host 17.1% · guest 82.9%6:00 · the host 17.1% · guest 82.9%9:00 · the host 0% · guest 100%9:00 · the host 0% · guest 100%12:00 · the host 28% · guest 72%12:00 · the host 28% · guest 72%15:00 · the host 11.9% · guest 88.1%15:00 · the host 11.9% · guest 88.1%18:00 · the host 21.1% · guest 78.9%18:00 · the host 21.1% · guest 78.9%21:00 · the host 9% · guest 91%21:00 · the host 9% · guest 91%24:00 · the host 18.4% · guest 81.6%24:00 · the host 18.4% · guest 81.6%27:00 · the host 14.3% · guest 85.7%27:00 · the host 14.3% · guest 85.7%30:00 · the host 10.9% · guest 89.1%30:00 · the host 10.9% · guest 89.1%33:00 · the host 23.1% · guest 76.9%33:00 · the host 23.1% · guest 76.9%36:00 · the host 1.6% · guest 98.4%36:00 · the host 1.6% · guest 98.4%39:00 · the host 71.8% · guest 28.2%39:00 · the host 71.8% · guest 28.2%
Sharpest disagreement ▶ 29:04 Rejecting low expectations for teachers

Anarupa Ganguly forcefully rejects the narrative from district superintendents that teachers cannot handle VR technology, insisting leaders underestimate teacher enthusiasm.

Hardest push from the host ▶ 16:22 Challenging the cheating narrative

Steph Smith pushes Sal Khan directly on the popular premise that AI tools in classrooms will primarily be used by students to cheat.

Biggest teaching moment ▶ 5:26 Explaining fundamental indicators of STEM success

Anarupa Ganguly educates the host on how spatial thinking and 3D abstraction, rather than 2D representations, are the actual core indicators of post-secondary STEM achievement.

The host holds their own ▶ 19:36 Citing Swedish educational model

Steph Smith demonstrates her own domain familiarity by drawing a direct comparison between Sal's AI feedback process and the iterative grading system used in Sweden.

the scores for every segment, with the reasoning behind each
ChapterTopicThe host as informed peerGuest teachingGuest disagreementThe host pushing backWhy
Legal Disclosures & Title Sequence 2100 Steph Smith introduces the panel and sets up the premise of the episode with an audience show-of-hands before asking Romy Drucker for baseline pandemic learning loss data. Romy provides sober stats on math and reading decline in a collaborative manner.
Prisms VR and Spatial Computing in STEM 1300 Steph asks Anarupa Ganguly about STEM talent development. Anarupa educates the audience and host on spatial reasoning and abstraction using Isaac Newton's gravity example.
Khanmigo: AI-Powered Socratic Tutoring 1200 Sal Khan delivers an extended explanation of Khanmigo and how AI can act as a Socratic tutor rather than an answer generator. The host listens as Sal details guardrails and partner interactions with OpenAI.
Minecraft Education: Gamification and Intrinsic Motivation 2300 Steph prompts Allison Matthews on gaming in education and then Anarupa on common misconceptions. Anarupa reframes tech adoption, noting that focus should be on pedagogy rather than hardware.
Rethinking AI: From Cheating Tool to Co-Creator 3200 Steph directly prompts Sal regarding cheating concerns with AI. After Sal outlines AI co-creation, Steph offers a helpful parallel from her own educational exchange in Sweden.
Allison Matthews on Gaming, Minecraft, and Evolving Teacher Roles 2200 Steph asks Allison about the evolving role of teachers when gaming is introduced. Allison explains scaffolded learning and how Minecraft fosters growth mindsets and digital citizenship.
Romy Drucker on Student Agency, Inclusive R&D, and Educational Access 2200 Steph asks Romy how outside stakeholders can support classroom innovation. Romy details student agency and inclusive R&D models with families and teachers.
Anarupa Ganguly on VR Hardware Deployment, Adoption, and Supporting Teachers 2311 Steph questions how VR hardware costs and school board adoption are handled. Anarupa strongly rejects low expectations placed on teachers by district leadership, citing high opt-in rates.
Sal Khan on Addressing Educational Deficits with AI and Immersive Learning 2400 Steph asks Sal about the broader timeline and future impact of AI. Sal outlines sobering math proficiency statistics and connects generative AI to sci-fi visions of personalized learning.
Panel Discussion: Final Calls to Action for Education 2100 Steph guides the panel through closing calls to action. The guests emphasize collaborative execution and non-partisan commitment to student success.

Statements from this episode (19)

Assertion Supported
Drucker: US math achievement levels drop to lowest point since 1990
“We know from data that came out last week that math achievement levels are the lowest they've been in this country since 1990.”
Romy Drucker Jul 27, 2023 ▶ 4:26
Assertion Supported
Drucker: US literacy scores hit lowest levels since 2004
“And literacy achievement levels are the lowest they've been since 2004.”
Romy Drucker Jul 27, 2023 ▶ 4:35
Assertion Supported
Drucker: COVID-19 pandemic wiped out 20 years of educational progress
“And the pandemic erased two decades of progress that we had made in education through a variety of different change initiatives.”
Romy Drucker Jul 27, 2023 ▶ 4:40
Assertion Not checkable as stated
Ganguly: Spatial reasoning is top predictor of STEM success
“I discovered that the top indicators of success in post-secondary STEM is number one, your ability to think spatially.”
Anarupa Ganguly Jul 27, 2023 ▶ 5:27
Assertion Partly supported
Sal Khan: 40% of teachers used ChatGPT within months of release
“Within a month or two, 40% of teachers were using this very imperfect tool for themselves for things like lesson planning, and students were already using it”
Sal Khan Jul 27, 2023 ▶ 10:46
Insight
Khan: Generative AI accelerated education technology timeline by 20 years
“Generative AI kind of jumped us forward 20 years faster.”
Sal Khan Jul 27, 2023 ▶ 11:23
Assertion Supported
Sal Khan: Khanmigo refuses to give direct answers to students
“If you ask Conmigo hey, tell me the answer, it will not tell you the answer, but say I'm your tutor, it'll act Socratic and say, well, what do you think is the next step?”
Sal Khan Jul 27, 2023 ▶ 11:55
Assertion Not checkable as stated
Ganguly: 100% of targeted teachers opt into PrismsVR program
“I have a hundred percent of teachers opting in to our program.”
Anarupa Ganguly Jul 27, 2023 ▶ 15:13
Insight
Ganguly: Edtech builders must train teachers on pedagogy, not hardware
“It is then therefore incumbent upon those who are building learning solutions to over-index on training teachers in the pedagogy and not the tech.”
Anarupa Ganguly Jul 27, 2023 ▶ 15:31
Prediction Held up
Sal Khan: Khan Academy launching AI college essay tool within months
“In a couple of months we're going to be launching something, we're going to initially do it for college essays, which is a whole other, this is the first college admission cycle in the chat GPT era, so it's going to be very interesting.”
Sal Khan Jul 27, 2023 ▶ 17:16
Insight
Sal Khan: AI interaction logs let teachers grade process over output
“When you work with the AI and all of the interactions are logged a teacher can actually not just introspect on the final outcome, they can introspect on the process.”
Sal Khan Jul 27, 2023 ▶ 18:08
Assertion Partly supported
Matthews: Minecraft Education reaches millions of monthly players across 119 countries
“Minecraft education is now in a 119 countries, and we've got Millions of players every month”
Allison Matthews Jul 27, 2023 ▶ 20:19
Insight
Ganguly: Edtech fails when teachers must act as technical conduits
“If we make teachers the conduit for the tech, we are setting ourselves up for failure.”
Anarupa Ganguly Jul 27, 2023 ▶ 30:04
Assertion Partly supported
Khan: Detroit 8th grade math proficiency fell from 6% to 3% post-pandemic
“You know, people have looked at the NAEP results, and I think it was Detroit had three percent of kids, and I think they were eighth graders who were proficient in math, three percent. It's like, wow, the pandemic was pretty bad. Well, before the pandemic was …”
Sal Khan Jul 27, 2023 ▶ 30:43
Assertion Not checkable as stated
Khan: Average US student learns only 0.6 to 0.7 grade levels yearly
“And that's because the average American student learns about .6 to .7 grade levels per year. Which, yeah, after 12 years, you're at a seventh grade level.”
Sal Khan Jul 27, 2023 ▶ 31:41
Assertion Partly supported
Khan: 30–60 mins weekly on Khan Academy accelerates learning by 40%–50%
“Students put in 30 to 60 minutes a week, so not like a huge amount. They're accelerating 40, 50%.”
Sal Khan Jul 27, 2023 ▶ 32:07
Prediction Not checkable as stated
Khan: Star Trek-style immersive VR learning will happen in next 5 years
“And then I think you start to merge these things with platforms like you're hearing, and it becomes completely immersive. You can go to ancient Rome, you could talk to Caesar, you could talk to Cleopatra It, it's like the holodeck on Star Trek, and I honestly …”
Sal Khan Jul 27, 2023 ▶ 34:50
Prediction Not checkable as stated
Khan: Hesitancy to adopt edtech will exacerbate educational inequality
“And don't let fear get in the way of trying something out, because if we do that, all that's going to happen is all of our kids are going to benefit from all of this. I'm not worried about them, but then the other kids are not, and then that's going to create …”
Sal Khan Jul 27, 2023 ▶ 38:14
Insight
Ganguly: School districts suffer from execution deficits, not tech deficits
“And if I were to have a last word, the thing that I've seen missing in school districts, I've worked across many, many of them. I had very large teams. We met, we were missing execution. And that's, that everything is now there. Pieces are there. We just now n…”
Anarupa Ganguly Jul 27, 2023 ▶ 38:51
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 1,000 episodes transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.