Gavin Miller: Adobe Research
Detailed Insights
How the conversation moved
The episode begins with Gavin Miller discussing the intersection of creativity and technology at Adobe Research, setting the stage for exploring how AI can transform creative tools. Miller's initial framing suggests that Adobe's mission is to automate tedious tasks, allowing creatives to focus on generating ideas rather than manipulating pixels. This is rooted in his early work in voice synthesis and smart home technology, where he blended literary creativity with technological innovation, providing a foundation for his current pursuits at Adobe.
Miller argues that AI can enhance creative processes by offering smart defaults and improving tools like Photoshop's background removal, making workflows faster and more efficient. He provides concrete examples, such as the use of Generative Adversarial Networks (GANs) for object removal and background filling, though he acknowledges challenges in achieving high-resolution outputs and integrating common sense knowledge. The conversation highlights Adobe's approach to using AI for real-time user assistance and enhancing the learning journey with tools like Photoshop and Premiere.
Lex doesn't challenge Miller's framing here, though the obvious counter-position would be concerns about AI reducing the need for human creativity or leading to job displacement in creative industries. Instead, the discussion remains focused on the benefits of AI in creative workflows, with Miller emphasizing the potential for AI to not only assist in tasks but also to foster a deeper understanding and mastery of creative tools. This lack of pushback leaves the conversation somewhat one-sided, with little exploration of potential downsides.
The conversation pivots to exploring Adobe's approach to user data privacy and the integration of AR and VR in creative workflows. Miller discusses projects like Puppetron, which applies artistic styles to videos, requiring an understanding of 3D structures. The episode concludes with a discussion on the evolution of Miller's work in robotics, particularly snake robots, illustrating how his projects have advanced from basic designs to sophisticated models with Raspberry Pi, showcasing the potential for biomimetic robots in practical applications.
Surprising moments
Topics Covered
Memorable Quotes
Still open
Unresolved by the end of the conversation
- Gavin Miller pondered how AI tools could further evolve to include an ensemble of experts, each handling different tasks based on specialized training.
Jargon glossary
References & Resources
For the specialist
What a senior practitioner would find new
- Adobe's 'concept canvas' allows users to assign spatial regions to keywords, enhancing image search intuitiveness.
- Project Puppetron's style transfer requires understanding 3D structures, highlighting the complexity of realistic video editing.
- Gavin's snake robots use Raspberry Pi for onboard computing, demonstrating significant autonomy advancements in robotics.
Ask this episode Deep
A preview of how Deep chat answers, grounded in this episode with citations and timestamps:
Cite this episode
For papers, blog posts, anywhere.
Related episodes
Where to go next from this conversation.
AI-generated summary · last refreshed 2026-06-08 20:00:50 · how we make these
Quotes are matched verbatim against the source transcript; references are checked to resolve to real URLs. Even so, AI can misread structure or attribute claims imperfectly. If you spot an error, please let us know.