The world of technology is constantly evolving, and those who are willing to embrace change and invest in new ideas are the ones who often reap the most rewards. While the immediate payoff might not always be clear, investing in early-stage concepts and technologies can lead to game-changing outcomes, even when the initial ideas weren’t intended to achieve grand results. Lately, I've been thinking a lot about this concept, especially as I’m reading Creativity, Inc., listening to various tech / startup focused podcasts, and pondering some recent projects at work.
Sparks: The Unexpected Rise of Pixar Short Films
Pixar, a studio known for its critically acclaimed and commercially successful feature films, took an interesting approach to nurturing creative talent in 2019. They launched an internal initiative called SparkShorts, with a simple goal: empower employees to explore their creative ideas by producing short films. These films were subject to two constraints: they had to be viewable by the typical Pixar audience, and they had to be completed within a six-month timeframe. This initiative served as a platform for internal talent development, allowing individuals to experiment with new ideas and hone their craft.
Initially, the plan was for these films to only be viewed internally. Pixar wasn't focused on creating marketable content; the priority was to empower their employees to explore creative ideas. However, Pixar quickly recognized the unique quality and creative potential that the SparkShorts produced.
These short films ended up becoming valuable assets for Pixar's new streaming platform, Disney+. The fresh and innovative nature of the SparkShorts resonated with viewers, contributing to the success of the streaming service.
This unexpected payoff is a reminder that even seemingly small, internal initiatives can spark valuable innovation and contribute to future success in ways you might not anticipate.
NVIDIA: A Chip Designed for Gaming, Reinvented for AI
Another striking example comes from NVIDIA, a company synonymous with high-performance graphics processing units (GPUs). The initial goal of these GPUs was to power the gaming industry, which offered a clear market need. NVIDIA, however, didn't stop there. They understood that the video game market was a powerful "flywheel" for innovation. By investing heavily in gaming GPUs, they were able to push the boundaries of graphics processing, leading to a significant increase in performance and efficiency.
Their continued investment in research and development, coupled with their understanding of the evolving technological landscape, led them to realize that their GPUs had a massive potential in the emerging field of AI. The chips they designed for gamers, with their incredible processing power, turned out to be perfectly suited for training the massive AI models that were starting to take off.
NVIDIA's prior investments and willingness to adapt led them to launch a new product line, the Blackwell Platform, specifically designed for AI. They were able to react to changing market demands, which led to a massive surge in their company value. They had invested heavily in GPUs for gaming, a decision that ultimately paid off in a completely unexpected way.
By focusing on gaming, Nvidia inadvertently built the foundation for their dominance in the AI space. This is a testament to the power of long-term vision and the ability to adapt to unforeseen opportunities.
The Power of "Investing" in Ideas
The stories of Pixar's SparkShorts and NVIDIA's GPU evolution highlight the potential for unexpected rewards when investing in early-stage concepts and technologies. Both examples illustrate how the initial goal of a project, as powerful as it may be, can sometimes lead to producing an even greater outcome if a team is able to change and adapt.
I’m lucky enough to work on a team within Google Labs. Labs exists “to discover and deliver new products that advance Google’s mission.” We're dedicated to fostering a culture of experimentation and exploration, and we like to build to learn (you can check out some of the projects from the team at labs.google).
Recently, my team hosted an internal hackathon which culminated in a Demo Day where we encouraged folks to present their ideas and prototypes - regardless of the stage they had gotten them to. The goal wasn't necessarily to create working products, but to spark new conversations, explore new possibilities, and build a collection of concepts that we can draw upon as we continue to develop our AI products.
To kickstart the hackathon, we encouraged a spirit of open exploration while providing some light guidance. Instead of rigid constraints, we offered a few broad themes to spark inspiration:
Judgment & Evals: Evaluations are crucial in any field, but how do you measure something as subjective as humor or creativity? We challenged teams to explore new ways to approach evaluations, especially for outputs that exist on a spectrum.
Enabling New Workflows: Ideas are often fleeting. How might we design tools and technologies that help people capture those sparks of inspiration and transform them into something tangible?
Rethinking UX: Generative AI demands a reimagining of user experience. We pushed teams to think beyond the familiar chatbot and co-pilot paradigms—what does a truly AI-native UX look like?
Agents: The concept of AI agents is gaining traction, but what defines an agent? What role does memory play? We encouraged explorations into the fundamental building blocks of this emerging technology as they could apply to tangible problems and products.
Generative UI: Inspired by Jensen Huang's prediction that all pixels will soon be generated, not rendered, we challenged teams to dream up experiences and products that embrace this generative future.
Sidenote: there was a great podcast this week from Sequoia, featuring an interview with LangChain’s CEO Harrison Chase, where he talks about some of these themes - from new UX patterns, to the future of agents: TRAINING DATA: EP1
Just like Pixar's Sparks and NVIDIA's GPU journey, we believe that these early-stage ideas, even if they aren't fully realized today, will help us navigate the future of AI. By investing in ideas, we're creating a reservoir of possibilities that could lead to groundbreaking innovations that will continue to pay-off in the years to come.
Time to Experiment!
I encourage you all to continue playing around with Generative AI, even if it’s as simple as just prompting some different models to explore what is possible! While my “go-to” is Google AI Studio (and the Gemini Cookbook if you want to use the API), I’ve also been having fun this week playing around with the newly upgraded Runway ML, as well as Luma Dream Machine, and Suno.
Thanks for the recommendation of Training Data podcast; that looks very interesting, especially re possible new UX.