Building narrative-driven games & learning experiences using AI, grounded in long-form fiction & interactive design. Former lecturer of data science & librarian
My motivation for my work is that my kid remembers, "My father redesigned his life so that telling multimedia stories was the main thing."
By showing up at my work everyday, my kid grows up with a living example that it's valid to treat stories as serious work.
I'm demonstrating the willingness to take risk & rebuild a life around creativity. My kid grows up with the sense that this is somethign people in my family do.
Working with Opus 5 on re-creating an audio voice that I had created in Bark TTS in 2023. AI is handling the hard stuff & giving me a nice custom tool for analyzing the results. My time is on thinking about audio quality adjustments & not coding.
My students were always more interested in sports than me. This basketball dataset would have served as a good example when I was teaching data for deep learning. The dataset contains 11,000 valid offensive possession clips and 31,567 annotated events with player names, captions, event types and timestamps.
https://t.co/kwSkp5e4Sj
With my Amazon orders, I started to take the Amazon Day delivery with the $ digital reward since my Amazon Prime deliveries are almost ALWAYS delayed by a day or two anyway.
Flux 3 from a prompt with 8 shots, 20 seconds. Overall really good.
Main problems: the video projecting against the wrong wall. And writing the え character is off but all models are having problems writing Japanese kana.
Prompt: S04-001 — Dust and light
A wide interior view of a small screening room inside an abandoned factory in Kanda. Afternoon light leaks through high windows. Dust floats thickly in the projector beam, sparkling like static. The walls are rough and stained, barely containing the light.
S04-002 — Images on the wall
A medium-wide shot of the projected film on a cracked wall. Flickering images—trains passing, blurred faces, puddles, street signs—bleed into one another. The frame trembles slightly as the projector hums.
S04-003 — Watching
A medium shot of Alex seated near the wall, turned three-quarters toward the projection. The moving images wash across his face, fragmenting his features. He watches intently, leaning forward.
S04-004 — The frame
A tight close-up of Emi’s fingers holding a single frame of film up to the light. The tiny image inside the frame is barely legible. Her fingertips glow where the light passes through.
S04-005 — Writing on film
A close-up of Emi’s hand writing え on the edge of a discarded reel with a wax pencil. The white mark contrasts sharply against the dark film stock. The background is soft projector flicker.
S04-006 — Horizon memory
A close-up of Alex’s eyes as he studies the character. Reflected in his pupils is a thin horizontal line of light, reminiscent of a distant horizon dividing sky from city.
S04-007 — Shared glow
A medium two-shot of Emi and Alex standing close beside the projector. Light spills between them, dust drifting through the beam. Emi glances toward Alex as the images continue to flicker behind them.
S04-008 — Light as memory
A final close shot of the projected wall as the image momentarily flares brighter, nearly washing out the picture before settling back into grain and shadow.
As AI progresses: new types of jobs will be created but people in those eliminated jobs will not be the ones who will get the new jobs due to lack of skills and training, and needing the time to reskill even if they had the aptitude.
RunwayML Gen-4.5. Started with an image from Midjourney. Gen-4.5 prompt: The locked-off camera remains perfectly still. Medium shot of a woman's silhouette standing motionless in front of a large glowing digital display wall in a darkened room. The woman is backlit by the bright screen, her figure rendered as a sharp black silhouette against the luminous blue-white glow. On the display, multiple human faces appear in a grid pattern, each face animated and speaking — mouths moving, subtle head tilts, natural facial expressions. The faces are rendered in cool blue tones with soft digital scan lines. Minimal subject motion only. The woman's silhouette holds completely still as the faces on the screen continue their silent animated conversation. High contrast lighting — deep blacks in the foreground, bright digital glow from the screen creating a halo effect around the woman's edges.
I needed a baseline for my own thoughts as I research creativity, imagination, & AI.
There's also quite a bit here about my writing process, from handwritten 1st draft to revising & revising.
https://t.co/y8z4CnCMUA
In walking me through Cloudflare configuration, Opus 4.8 was much better than Opus 5.
Opus 5 often tells me "You're right, and my last diagnosis was wrong."
Next time, I'll try Cloudflare's AI agent (or just refer to my own notes).
One of my most memorable courses from college was a close reading of the Iliad & the Odyssey. 40 yrs later, I still think about that course. While I will reread those works & listen to audiobooks, I have no interest in seeing a movie on the Odyssey.
Most people will not use AI to create their own personal apps.
In 1993, HTML was so easy that I thought nobody would pay anybody anything to create web pages.
There will be plenty of opportunities for those who understand AI, software, etc.
I'm continually astonished at the OCR capabilities of AI to recognize my messy handwriting. I still write a lot by hand; it helps slow down my thought process. Gemini handles the OCR perfectly.
1996: the tech we use now seemed impossible. Yet, many people assume that 30 yrs from now (2056) will be like today.
Helping students think in decades is a challenge. When I was 20, just 2 or 3 yrs was a big deal. I couldn't grasp thinking about 10, 20, 30 yrs ahead.
Fable 5 can still generate slop in game dev despite what seems like a well constructed prompt. The good news is that there's work for people who really understands these systems, which is a small portion of the world now.
This system map was by Opus 4.8. I'll use it to keep iterating with Fable 5.
Friends will come and go, jobs will come and go, politicians will come and go, but we are still here. Every day offers the potential for hundreds of beautiful moments, one after another.
As artificial intelligence decimates what we have known as knowledge work, we must recover our individuality. Our contributions to society will be augmented by AI, our capacity to accomplish more will be expanded by AI. Do not race to keep track of every new development in AI. Do not fear missing out. What you are seeking is within yourself. Let the questions of your life emerge. Build your work around your unstoppable ideas.