“I’m just a simple Unfrozen Caveman Web designer…”

AI/ML, Idle Philosophizing

I had a great time chatting with my fellow former Adobe PM Demian Borba about all things AI (creativity, ethics, ownership, value, and more). You can check out the conversation below, and in case it’s of interest, I used Gemini inside YouTube to create a summary of topics we discussed.

Here is a full list of topics discussed in the video, complete with timestamps:

  • Introduction to the Podcast and John Nack (0:00-0:54)
  • John Nack’s Early Life and Introduction to Technology (0:54-2:06)
    • Learning to draw at age three (1:11-1:22)
    • The Apple Macintosh as a “road to Damascus” moment (1:22-2:06)
  • The “Yes And” Philosophy in Creative Work (2:06-2:36)
  • Accessing Technology in College and Early Web Work (2:36-3:41)
    • Working with professors to put coursework online (3:11-3:15)
    • Observations on younger generations’ interaction with technology (3:16-3:41)
  • John Nack’s Career Trajectory and Joining a Dot-Com Agency (4:06-5:14)
  • Frustrations with Existing Tools and Engaging with Adobe/Macromedia (5:14-6:19)
    • Spending time on “dumb crap” and seeking improvements (5:37-5:55)
    • Being a “pissed-off Adobe customer” (6:08-6:19)
  • Becoming a Product Manager at Adobe (Flash and Web Standards) (6:19-8:40)
    • Joining around 2000 to make a web standards-based animation tool (7:39-8:28)
    • The dot-com crash impacting projects (8:31-8:38)
  • Transition to Photoshop Product Management (8:40-9:46)
    • Receiving a call from Brian Lamkin (8:52-9:20)
    • Spending 10 years on Photoshop (9:42-9:46)
  • The Nonlinear Career Path and Social Media Perception (9:52-10:14)
  • Inside Adobe: Challenges, Politics, and Legacy Codebase (10:49-11:47)
    • Photoshop codebase as a “museum” (11:08-11:13)
    • Photoshop 1.0 going to the Smithsonian (11:21-11:44)
  • The “Horseless Carriage” Moment of AI (11:48-13:06)
    • Applying old paradigms to new technology (12:00-12:40)
    • Questioning the need for traditional tools like layers and brushes (13:06-13:33)
  • New Paradigms in AI Creative Tools (Chat-centric, Node-based) (14:00-15:00)
    • Reve.art and the shift to art direction/choreography (14:00-14:37)
    • Node-based approaches (14:42-14:57)
  • The Shift in User Interfaces (Complex Photoshop to Prompt-based) (15:19-16:08)
    • Photoshop’s non-intuitive nature versus new prompt-based UIs (15:36-16:08)
  • The Modality of Tools: From Broad Possibility to Specificity (16:09-17:46)
    • Inverse pyramids analogy (16:59-17:10)
    • Prompt-based vs. direct input for precise manipulations (17:47-18:12)
  • Limitations of Language for Direct Manipulation (18:12-18:49)
    • Blade Runner and Minority Report examples (18:21-18:49)
  • Advanced AI Model Capabilities (WAN, World Models) (18:50-24:00)
    • Demonstration of WAN video and its impressive capabilities (20:33-21:57)
    • AI understanding depth, lighting, direction, and physicality (22:05-23:56)
  • Photoshop’s Vanishing Point Feature and Future UI (24:00-25:05)
    • Limitations of manual mesh placement (24:22-24:44)
  • Guiding Generative Models with Visual Structure (Flux Hackathon) (25:26-26:27)
    • Using perspective grids to cue AI (25:40-26:27)
  • Project Turntable: Generating Novel Views from 2D Drawings (26:45-27:57)
    • Vectorizing generated views (27:09-27:12)
    • Impact on artists (27:18-27:30)
  • AI for Compositing and Harmonization (28:05-28:57)
    • Removing backgrounds and harmonizing elements (28:11-28:20)
    • Past limitations of “state-of-the-art” technology (28:40-28:57)
  • The “Black Box” Nature of AI Models (28:58-29:33)
    • Engineers not fully understanding what happens inside the model (29:08-29:33)
  • Gaussian Splatting, Nerfs, and Uncanny Valley in AI Generation (29:34-31:20)
    • Jon Baron’s work (29:38-30:00)
    • Will Smith eating spaghetti example (30:37-31:20)
  • Human Brain Simulations vs. AI Simulations (32:04-33:11)
    • AI trained on past data, running simulations (32:37-32:47)
    • “Effing magic” of AI (33:07-33:11)
  • Biases in AI Models from Training Data (33:17-35:20)
    • Obama’s image scaling up as a white guy (34:02-34:06)
    • Nvidia’s finding on women looking more male when aged (34:12-34:20)
    • Correlation of data in advertising (34:49-35:20)
  • Cultural Context and Perception of Beauty (35:32-37:16)
    • Fiverr challenge on making a woman more attractive (36:02-36:09)
    • The future for kids growing up with AI (36:29-36:46)
  • “The Status Game” Book and Human Mammalian Programming (37:26-38:20)
    • Humans seeking status for gene propagation (37:51-38:20)
  • Status and Body Image/Skin Tone (38:20-39:19)

Thin vs. curvy and pale vs. tanned skin as status indicators (38:23-39:19)