May 1, 2023 · 48m · no-priors
No Priors Ep. 2 | With Runway ML’s Cristobal Valenzuela
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 No Priors, Runway ML co-founder and CEO Cristobal Valenzuela joins Sarah Guo and Elad Gil to discuss the evolution of generative media, the philosophy of building applied AI tools for creators, and how multidisciplinary teams are reshaping video production.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 17.5% of the talking time here. How this is scored →
The hosts as informed peer 5.1 Guest teaching 4.7 Guest disagreement 0.6 The hosts pushing back 0.3
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 34:18 Rejecting static evaluations of generative AI
Valenzuela strongly critiques traditional agency mindsets that dismiss early generative models as useless toys, arguing they fundamentally fail to understand exponential compounding.
Hardest push from the hosts ▶ 29:01 Questioning high-end VFX adoption of generative toolsSarah Guo challenges the premise of selling early generative AI to professional VFX studios, noting it contradicts the widespread belief that current models lack required production fidelity.
Biggest teaching moment ▶ 41:20 Portable paint tubes and the birth of ImpressionismValenzuela delivers an articulate history lesson on how packaged paint tubes freed 18th-century painters from studio pigment preparation and gave rise to Impressionism, directly analogizing it to modern generative AI.
The host holds their own ▶ 4:32 Connecting Silicon Valley origins to media artsElad Gil demonstrates deep domain knowledge by weaving Stewart Brand, Paul Graham's 'Hackers and Painters,' and Sep Kamvar's MoMA exhibits into the tech-art discussion.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Cristobal Valenzuela's Background in Art, Business, and Technology | 3 | 2 | 0 | 0 | Sarah Guo opens with a warm, biographical prompt detailing Valenzuela's multifaceted background across economics, design, and NYU ITP. The guest amiably shares his journey learning programming and combining hardware art with consulting. | |
| Bridging Media Arts, Technology, and Silicon Valley History | 6 | 3 | 1 | 0 | Elad Gil demonstrates solid historical knowledge connecting early Silicon Valley, Stewart Brand, Paul Graham, and Sep Kamvar with artistic subcultures. Valenzuela expands on his first-principles approach and rejecting arbitrary disciplinary silos. | |
| Defining Runway and the Evolution from Model Directory to Platform | 4 | 5 | 0 | 0 | Sarah Guo recalls Runway's early incarnation as a desktop model directory from 2019. Valenzuela educates the hosts on the evolution from early GANs and AlexNet to building model hosting infrastructure and turning research algorithms into creative tools. | |
| Technology Stack Evolution and Prioritizing Long-Term User Needs | 6 | 4 | 1 | 1 | Elad Gil draws an insightful analogy between pre-AWS infrastructure traps and transitioning AI model architectures. Valenzuela explains the necessity of ignoring short-term customer feature requests to prioritize long-term architectural stability. | |
| Building an Applied AI Research Lab and Avoiding Founder Pitfalls | 5 | 6 | 1 | 0 | Guo and Gil ask about balancing applied research with external open-source models and common founder pitfalls. Valenzuela emphasizes that standalone models are not products and explains the importance of pairing domain artists directly with AI researchers. | |
| The Promise of Multimodality and Organizing Multidisciplinary Teams | 5 | 4 | 0 | 0 | Gil shares organizational insights from Color regarding embedding bioinformaticians with systems engineers, asking about Runway's organizational setup. Valenzuela details his evolving squad structure and the paradigm shift toward multimodal creative tools. | |
| Developing Runway's Green Screen Tool Through Human-in-the-Loop AI | 4 | 7 | 1 | 0 | Valenzuela breaks down the development of Runway's Green Screen tool, reframing academic literature on automated video segmentation by explaining that professional filmmakers need human-in-the-loop control rather than purely autonomous black-box systems. | |
| Commercializing Creative AI for Studios, VFX, and Entertainment | 6 | 5 | 1 | 2 | Guo pushes back on the counterintuitive nature of top-tier VFX studios using imperfect AI models. Valenzuela clarifies that 80% accuracy saves days of rotoscoping and liberates directors from waterfall production constraints, unlike autonomous driving where 99% accuracy is fatal. | |
| Recognizing Product-Market Fit and the Rate of Technological Progress | 5 | 5 | 1 | 0 | Gil and Guo explore indicators of product-market fit. Valenzuela recounts early agency meetings where his 128x128 GAN demos were dismissed as toys, highlighting the fundamental error of evaluating a technology's current snapshot rather than its exponential rate of progress. | |
| Scaling the Company Culture and Transitioning from Coding to Leadership | 7 | 7 | 1 | 0 | Gil provides deep art history context regarding Marcel Duchamp and Andy Warhol. Valenzuela delivers an elaborate historical analogy comparing modern AI tools to the invention of portable paint tubes in the 1700s, which enabled plein air painting and sparked Impressionism. | |
| Creative Coding Communities and the Future of Media Arts | 5 | 4 | 0 | 0 | Guo asks about emerging cultural and artistic scenes. Valenzuela spotlights the grassroots creative coding subcultures and fringe media art communities in New York as the true pioneers who will shape the future of generative technology. |
Statements from this episode (15)
Valenzuela: Runway offers around 35 AI-powered creative tools
“We have around 35 different what we call AI power tools or magic tools, and those tools help serve a wide spectrum of creative tasks from traditional, like, editing editing videos or just audio or images has been a very expensive, time-consuming, and sophistic…”
Valenzuela: Runway originally hosted around 400 AI models as a directory
“So we built, at the time was, as you're describing, like a model directory. It's an app store of models, right? You had, we had around, at some point, like, 400 different models. It was one of the first, like, I would say, model haps.”
Valenzuela: It takes 12-24 months to understand new AI breakthroughs
“The moment something gets released, like, let's say transformers or a particular piece of technology that you think would be interesting or could be worth experimenting with, I think it takes a collective set of months, like, 12, 24 months sometimes to underst…”
Valenzuela: AI models on their own are not products
“Models on their own are not products, right? A model is, is, is a research component and taking a model and productionalizing that model, it's a different problem that actually building one single model, right?”
Valenzuela: Runway had to own its AI stack to push creative tools forward
“If we really want to make and move the standard of like creative tools in the ways and vision that we had to own our stack. And so we started building this research team, right?”
Valenzuela: Half of Runway's team comes from an arts background
“Half of our team have arts backgrounds, right? Which is very unique.”
Valenzuela: Multimodal AI will merge previously siloed creative software tools
“I think a common natural evolution of just the creative stack or the creative software solutions out there, they tend to be very specific to domains of content. So you have a tool that's specialized on, like, Image editing. And then you have a tool that specia…”
Valenzuela: Runway operated without a dedicated product manager until April 2023
“Until like two weeks ago, we didn't have a product person. Product was led by a combination of research design and engineering.”
Valenzuela: Runway's first green screen tool ran at four frames per second
“And the first version of green screen was working at like four frames per second, right?
It was like incredibly slow.
It was like,
Not as good as the one we have now, which is incredible, but it didn't matter.
It was significantly better than anything else tha…”
Valenzuela: AI is very far from end-to-end movie automation
“If you're trying to automate the entire process of like the whole end-to-end system of making a movie, yeah, like we're not there, right?
We're very far from that.
There's a lot of things to be, that have to be developed, that have to be the like research and …”
Valenzuela: Creative AI succeeds early because creative workflows tolerate 80% accuracy
“And in research, going from 80% to a hundred percent is really hard. I think that you'll be seeing that in, like, autonomous vehicles where, like, it's always, like, two years ahead and, like, always 80%, but, like, that 20, 10% is just really hard. This is re…”
Valenzuela: Runway operated for a long time without marketing or content teams
“We certainly, for a long time, never had a, like a marketing team or a content strategy team. Like everything was just basically people making things and then sharing them online.”
Valenzuela: In decades, AI art will be seen as a natural, necessary transition
“What are we going to be asking ourselves in like, 1020 years, 30 years? I think it's the realization that we'll look back and we'll look at this moment as in like, yeah, I mean, it was a natural transition and we needed it. It allowed us to do so many things t…”
Valenzuela: Creative AI alignment requires making models expressible and controllable
“These models and the systems need to become really expressible and controllable, which is somehow the way I like to think about alignment is like you have an intention and you want to express that intention in a very controllable way, right? These models are y…”
Valenzuela: Fringe creative coding communities will define what comes next in tech
“I think for me if you apply that same kind of like principle now, I would tend to look a lot at like the weirdos of tech, right? People who are at the fringes, people who have, who've been always considered like, oh, you're just toying around. This is like a b…”