Moonshots that Move the Needle
Summary
This SXSW 2026 panel, "Moonshots that Move the Needle," brings together four leaders at the intersection of education, technology, philanthropy, and government R&D to explore what it takes to set bold, time-bound goals that can catalyze real change in learning outcomes. Moderated by Kamar, a former Obama White House science office staffer and Renaissance philanthropy leader, the conversation features Arti (former DARPA director), Eden Zanakis (Chief of Staff, Bezos Family Foundation), and Steve (Chief R&D Officer, Carnegie Learning).
The session opens with a foundational unpacking of the word "moonshot" — distinguishing it from vague aspirational language and grounding it in specific, outcome-oriented bets. Arti uses the mRNA vaccine story as a canonical example: in 2012, her DARPA program manager Dan identified that mRNA technology could become a rapid-response vaccine platform, secured funding to pivot Moderna from cancer to infectious disease research, and ran a Phase 1 clinical trial proving immune response — a result that arrived in 2017 and enabled Moderna to ship COVID-19 vaccine doses just 42 days after the spike protein sequence was identified. The key lesson: moonshots require picking a pivot-point milestone that converts skeptics, not just proving a concept in the lab. The parallel to education is explicit — we finally have the tools, the basic research, and the applied infrastructure to attempt similar leaps in learning outcomes.
A central thread is AI's role as a capability multiplier rather than a monolithic solution. Arti argues the field is making the same mistake it made with "e-commerce" and "dot-com" — treating AI as "the thing" rather than as a substrate enabling many different tools. Steve illustrates how falling AI costs (token costs dropping ~100x every 18 months) have made previously impractical ideas feasible: Carnegie Learning's Mathia program delivers over 3 million personalized messages to students, and what was once text-only feedback can now be custom video, animation, or diagrams at scale. Steve also describes augmented reality glasses for teachers — showing real-time indicators over students' heads to identify who is productively versus unproductively struggling — a direct application of neuroscience and learning science insights. Carnegie Learning's Fast Forward product, based on the neuroscience of acoustic speech perception, trains students' brains to distinguish phonemes like B and P, addressing pre-phonemic awareness that standard reading curricula cannot reach.
The panel explores the social and institutional dimensions of moonshots — how you bring skeptics along rather than work around them. Arti recounts DARPA's Sea Hunter program: a fully autonomous ship with no sailors that the Navy initially tried to kill. By the time she left as director, the Navy had co-launched the program because skepticism from senior naval officers (who asked hard operational questions) had been converted into design requirements. Steve echoes this: Carnegie Learning's biggest skeptics become champions when co-designing field trials, because teachers are already experimenting — they just lack measurement. The panel emphasizes a "hotline" approach: giving teachers upfront transparency about experimental conditions and a real person to call, not a chatbot. Kamar draws a direct analogy to the vaccine crisis: even enormous technical wins can be undermined if the social contract with the public breaks down.
On funding and policy, Kamar notes that education R&D spending is below one-tenth of 1% of total education expenditure — so small it doesn't even appear on government R&D charts. Arti frames the current crisis in federally funded R&D as both a tragedy and a moment of clarification: when the crisis passes, the field must build a better system that not only replicates past glories but addresses the equity gaps that remain. Mississippi's rise from 49th to 9th in national reading scores is cited as proof that scaling research-backed reforms across an entire state produces transformational results. Eden highlights the Bezos Family Foundation's investment in an MEG machine at the University of Washington's Institute of Learning and Brain Sciences — allowing researchers to watch baby brains develop in real time and translate those findings into policy and family programs. The panel closes with rapid-fire personal moonshots: Arti urges clarity on the destination and relentless focus on "what does it take to get there?"; Eden points to civic participation and helping every person live to their full potential; Steve emphasizes that every student can learn if given authentic tasks, genuine motivation, and a teacher who takes their thinking seriously rather than dismissing non-standard approaches.
Key Takeaways
- •A moonshot requires a specific, measurable pivot-point milestone — not just ambition. DARPA's mRNA program defined success as proving immune response in a Phase 1 human clinical trial in 2017, which then converted skeptics and enabled Moderna to ship COVID-19 vaccine doses 42 days after the spike protein sequence was known.
- •AI token costs are dropping approximately 100x every 18 months, meaning something that cost $100,000 to do now costs $1,000 eighteen months later. This cost curve is making previously impractical personalized learning interventions economically viable.
- •Carnegie Learning's Mathia program generates over 3 million different personalized messages to students. What was text-only feedback can now be custom video, animation, or diagrams at scale — a qualitative shift in what personalized learning means.
- •Education R&D spending in the US is below one-tenth of 1% of total education expenditure — so small it does not appear on government R&D budget charts alongside defense, health, energy, and space R&D.
- •Mississippi rose from 49th to 9th in national reading scores by scaling research-backed reading science reforms across an entire state — a concrete proof point that moonshot-level outcomes in education are achievable when evidence is systematically applied at scale.
- •The biggest skeptics often become the biggest champions. DARPA's Sea Hunter autonomous ship program survived Navy attempts to kill it by treating naval officers' hard operational questions (like navigating the Strait of Malacca) as design requirements rather than obstacles.
- •Teachers are already experimenting constantly — what they lack is measurement. Carnegie Learning's approach co-designs field trials with teachers, framing the work as helping them understand what is actually working in their own school, not imposing conditions on them.
- •Carnegie Learning gave teachers augmented reality glasses that display real-time indicators over students' heads, showing who is productively versus unproductively struggling. This translates neuroscience and learning data into actionable classroom decisions as a teacher walks toward a student's desk.
Transcript
Thank you. >> Well, good morning everybody. >> There we go. Welcome, welcome, welcome to uh transition day where we get a mix of the edu audience and the Southby audience. It's great to have all of you here. Um, I think one of the things that I've been talking to the panel about before we started was just, hey, there's a lot of people here, but the topic is kind of broad, like moonshots. What does this audience actually want out of this conversation? So, one of the things that I like to do when ...
