tbd/con

2026

the future of ai is tbd.
so is everything else.

a conference for people who have more questions than answers.

AI isn't arriving cleanly. neither should the conversation about it.

Madison Karas, Patrick Boehler, Matthew Gore-Kormanik, and David Kuszmar are putting together tbd/con, a virtual conference for people who have more questions than answers about AI.

researchers are probing model behavior. security practitioners are breaking systems. journalists are trying to explain what's happening, or figure out how to survive it as competition. policymakers are being asked to act on incomplete understanding. developers are shipping anyway. the result is a fragmented conversation about a technology that doesn't respect the boundaries it crosses.

tbd/con is an attempt to look at that, be surprised and confused and inspired. no keynotes from on high. no illusion of consensus. just practitioners across disciplines comparing notes and figuring out what to ask next.

if that sounds like your kind of room, take a look at the site and let us know how you want to be involved.

01

what's tbd

>

who ai actually serves

>

what it does to information ecosystems

>

whether safety and openness can coexist

>

how technologists, journalists, researchers, and hackers work together

>

what comes next

the sessions reflect that: talks that end with open questions. workshops where you build something and fail and learn from that. office hours to consult with someone from another domain. a midway where you wander into something you didn't expect.

02

circles

thought

systems architecture, ethics, epistemology, and philosophy centered on the emergent realities of synthetic and synthetic-human cognition.

information

utility, influence and the infrastructure of what we know and believe.

security

adversarial AI, breaking models, the realities of responsible disclosure, and the arms race between offense and defense.

practice

the mechanics of building: architectures, data curation, and engineering at scale.

garage

wild prototypes, failed experiments, pet projects, and the weird stuff that happens on the edges and the margins.

03

how it works

all virtual.

all on gather.town.

talks

unsolved theories

demos

raw prototypes

workshops

collaborative failure

panels

civil disagreement

office hours

expert critique

the midway

random encounters

when

september 23-24, 2026

where

gather.town

virtual campus. accessible anywhere.

access

free

no admission cost.

got a question worth 30 minutes?

we are looking for talks that end with open questions. technical depth, social critique, and wild prototypes welcome.

submit a pitch
05

call for pitches

tbd/con is a virtual conference for people who have more questions than answers about AI. researchers, hackers, journalists, policy people, builders. people who rarely end up in the same room. we're trying to change that.

we're looking for talks, demos, workshops, and panels across the five circles. bring the session you wish existed. we'll help you shape it.

the circles

thought: systems architecture, ethics, epistemology, and philosophy centered on the emergent realities of synthetic and synthetic-human cognition
information: utility, influence and the infrastructure of what we know and believe
security: adversarial AI, breaking models, the realities of responsible disclosure, and the arms race between offense and defense
practice: the mechanics of building: architectures, data curation, and engineering at scale
garage: wild prototypes, failed experiments, pet projects, and the weird stuff that happens on the edges and the margins

what we're looking for

sessions that leave the room with a better question than it walked in with. technical depth, social critique, and wild prototypes all welcome.

  • a real problem you have not solved yet, not a finished success story
  • specifics: what you built, what broke, what you would do differently
  • something the room can use, not a sales pitch
  • pick the circle that fits best, we will help you land in the right one

who should pitch

you don't need to have all the answers. we especially want to hear from practitioners, independent researchers, developers, community journalists, hackers, and people who've watched AI develop up close.

timeline

submit by june 30, 2026
selections shared after the july-august review
conference september 23-24, 2026, on gather.town