AI: Summary
The session explored how text, interfaces, and social structure shape meaning-making in XR-era knowledge work, moving fluidly between concrete interface design decisions, the limits of current citation systems, and the social mechanics of sustaining a focused yet inclusive research community. Participants reflected on AI and XR maturity, the realities of cost and access, and the mismatch between print-era assumptions and digital practice, grounding abstract ideas in lived design and editorial challenges.
AI: Highlights
Mark Anderson emphasized that current public narratives around AI obscure the real experience of use, which often involves multiple paid systems, significant iteration, and an emerging access divide that risks creating new knowledge “haves and have-nots.”
Brandel Zachernuk highlighted the value of locally run models, particularly Apple foundation models, as a stabilizing counterweight to hype-driven, subscription-dependent AI services.
Peter Wasilko clarified that his long-form paper intentionally preserves the full AI–human dialogue to demonstrate how expertise, judgment, and iterative prompting are inseparable, and that compressing this interaction would erase its core insight.
Frode Hegland proposed explicit spoken markers—verbal “sidebars” and named intentions—as a way to turn conversation itself into addressable, linkable structure for later analysis.
Participants collectively recognized that diversity in the community increases cognitive friction, but that this discomfort is productive and essential to better thinking.
AI: Insights
Citation remains fundamentally a print-era workaround: while intended to convey provenance and attribution, its digital implementations rely on brittle conventions, incompatible schemas, and visual validation through printing, revealing a deep conceptual gap between intent and mechanism.
AI interaction is best understood not as question–answer exchange but as a negotiated process that exposes the user’s own knowledge limits; without domain grounding, AI output becomes noise rather than leverage.
Stretch text and disclosure-based reading are not aesthetic choices but cognitive tools, allowing dense material—whether AI transcripts or scholarly argument—to remain legible without sacrificing completeness.
The effectiveness of themed discussion depends less on enforcement than on shared expectation-setting; clarity about what is in-scope and what is explicitly deferred reduces social friction without suppressing exploratory thought.
“Epi-markup,” where natural language doubles as machine-parsable structure, points toward a future in which spoken or written intent can be computationally recognized without formal syntax, enabling transcripts to function as semantic infrastructure.
AI: Resources Mentioned
Apple foundation models, discussed by Brandel Zachernuk as locally executable AI infrastructure.
BibTeX, referenced by Mark Anderson as an aging but still comparatively robust citation schema.
Marc Bernstein, mentioned by Mark Anderson in relation to reference manager integration questions.
Ken Perlin, cited by Frode Hegland as a growing presence whose involvement raises the importance of thematic clarity.
Bob Horn’s work on structured text and information design, referenced implicitly through discussion of readability and white space.
This track started as a meeting on The Future Text Lab website futuretextlab.info which was then transcribed (sonix.ai) and then summarized using AI for the website, including an AI summary in song lyrics which was then orchestrated by AI (suno.com). This is an experiment around a different aspect of text, to see if a more poetic, though machine made, presentation of what was discussed can spur thought and dialog.
This track is an AI orchestrated piece inspired by the transcript of this meeting, meant as a fun provocation to further thought. (suno.com)