Takashi Sakamoto (@bookslope) header
Takashi Sakamoto (@bookslope) profile picture

Experience Design for Autonomous Mobility | Photography & Visual Observations | Author of “IA Thinking” | Japanese History Explorer

Tokyo, Japan
Joined April 2007
Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)Photo by Takashi Sakamoto (@bookslope)
flying car?! #tesla
Turns out the future does have flying cars
1
457
長らく放置していた個人ブログを、ようやく再構築しました。 写真関連もこのブログに統合したので、デザインはもとより考えていることや取り組んだことを少しずつ残していこうと思います。 bookslope.jp/?view=home&v=2
1
10
719
America.gov is fascinating. Built with the National Design Studio, it connects 29,000+ government websites through a single AI front door. Not just websites: User Intent → AI → Service This feels like one direction UX and service design are heading in the AI era.
Today we’re launching America.gov, a single front door to the federal government. Instead of opening a bunch of tabs across different agency sites and trying to figure out which information actually matters to you, you can ask America a question and get the relevant answer from across 29,000+ official government websites, with the source and what to do next.
1
3
493
UX design in the AI era is shifting from “making” to designing how far AI should make decisions. Alison Rand explored the “Judgment Layer” at #ShiftUX. Jonathan Walter frames AI systems in four layers: Generation / Recommendation / Autonomy / Judgment #UX #AI #IA
1
1
300
The key is the Judgment Layer: What should AI decide? When should humans step in? How confident should AI be before acting? With AI, we may need to preserve friction for confirmation and oversight. This feels close to the shift from Designer to Architect. #UX #AI #IA
1
190
Meta unveils its own Tamagotchi? #meta #ai
Replying to @finkd
And we've been building a joyful little device for talking to your Muse that fits on your keychain. Muse Charm -- shipping in December.
1
423
Dario Amodei’s “Pace the Frontier” proposes three steps: Embedded Evaluators Democratic Coordination Global Coordination It feels like a major turning point in the history of humans and machines, and an important step toward global safety and order. darioamodei.com/post/we-must…
221
iPhone Duo #Apple2026
Can't stop playing with the iPhone Duo folding slider
1
570
Paz Perez「Context Architecture: Applying IA to AI Systems」 - プロンプトより「文脈」が重要 - AI が使う情報を構造化する - AI Skill にも Findability と Mental Model が必要 - AI Memory も構造化が必要 - AI の振る舞いを設計する 情報設計が AI 時代にまた中心に戻ってきた感。 #IA #AI
1
2
3
277
Paz Perez は、Google / Google DeepMind で AI × UX に取り組んできた Designer。 動画では、AI における「文脈」を IA の視点からどう設計するかを解説。Prompt Engineering から Context Architecture への広がり、AI Skill や Memory の設計まで扱っています。 #IA #AI maven.com/p/70e491/context-a…
1
184
Since I started using AI more deeply in my work, I’ve been thinking a lot about one question: What does it really mean to “work” or to “make” something? #IA #AI
1
4
420
Paz Perez’s idea of “Context Architecture for AI systems” points in a similar direction. Information Architecture is evolving from organizing content into shaping the context that influences how AI systems behave. #IA #AI rosenverse.rosenfeldmedia.co…
1
100
Information Architecture -> Context Architecture -> Intelligence Architecture Maybe AI is not simply changing my work. Maybe it is revealing more clearly what I have actually been doing all along. I’m beginning to think I’m less a Designer, and more an Architect… #IA #AI
99
A Waymo x TU Delft paper stood out for AV UX. It models human collision avoidance using active inference. Key insight: re-planning triggered by accumulated surprise. Surprise builds until crossing a threshold, then drivers rewrite their action plan. nature.com/articles/s41467-0…
1
357
Waymo uses this as a behavioral benchmark: how would a careful, competent human driver have responded? Previous models only captured last-second reflexes. This one reconstructs the full lead-up, including why the driver acted at that precise moment. techcrunch.com/2026/06/10/wa…
1
271
Safety isn't only about preventing accidents. It's about matching the timing humans expect. When a robotaxi drifts from human cognitive models, people can't read what comes next. Unpredictability is a trust problem.
173