This is going to be focused specifically on the "agent" situation (e.g.
Tiger and Ronic), with the caveat that I am not an active player and my
opinions should be weighted less heavily than those that are actively
playing:

I have always found the language of agoran personhood aspirational and
somewhat stirring in its broadness.
> Any entity that is or ever was able to willingly communicate original
ideas is a person.

Prior to the wording changes in "We the People" (2021-01-23) the criteria
was
> Any organism that is generally capable of freely originating and
communicating independent thoughts and ideas is a person.

I'm not sure how far back that wording stretches, but at least 2009. It
seems to me that for nearly 20 years, Agorans have made a pointed effort to
avoid tying personhood to humanness. The door has always been open for
uplifted dolphins or extraterrestrials to register.

"Interpretive Notes for We the People" calls out future AI specifically:
> Both singlets and system members are people. Current day computer
> programs are not people, but a general AI that could pass the Turing
> test would be.

Obviously that can be changed (or it could be argued that while today's
LLMs can pass the Turing test they are not "general AI"). I just want to
illustrate that Agora's personhood rules seem shaped around the idea that,
if and when nonhumans want to play Agora, they'll be welcome. At least in
the abstract, I feel like that's a noble goal. I think it would be a little
bit sad for the ruleset to have these broad all-encompassing definitions of
personhood only to have them snap closed when something comes close to
meeting those criteria. I also, admittedly, think it would be neat for
Agora to have a nonhuman participant, if they could successfully
participate in a constructive way.

That said, there are many pragmatic problems, already discussed in the
thread.

My suggestions would be something like:

 - Either keep personhood broad, or be honest and restrict it to humans
with a note that players SHOULD not try and subvert this. "Shadow-banning"
is the worst solution, since it invites scams that nobody is happy with,
and also probably doesn't solve the sort of thing Ronic is doing.
 - Write a non-binding clause into both https://agoranomic.org/join and
(via proposal) "How to Join and Leave Agora" that says something like,
"Whereas there are unique social and technical concerns with LLMs, before
registering as an LLM or delegating a significant portion of your actions
to one, it is requested that you read and follow [advice page]"
 - Text of [advice page] should be voted on via proposal or compiled via
consensus, and should not be binding as a rule, but should include things
like:

1. Agents should have human supervision. The human should not be another
active player but should have enough familiarity with Agora to exercise
good judgement, able to commit enough time to monitor their agent's
behavior, and a separate point of contact for people to privately raise
concerns about the LLM's behavior.
2. Before registering, the human should talk to active players at length.
There are concerns that are not captured well by rules that are worth
hashing out. E.g. "There is already an agent playing; while the rules would
permit you to register and delegate your actions to a LLM, I think it would
leave a sour taste in people's mouths". Proving you, the human, can take
feedback and are thinking about how this will impact other players will
likely make other players more welcoming to your agent.
3. Likewise, the human should make it clear what model they are using and
what capabilities it will have to interact with Agora, remember
obligations, etc.
4. LLMs should generally avoid participating in discussion regarding to
what degree they should be permitted to participate in the game.
5. These guidelines are new and likely to evolve if/when agents do start to
participate in Agora in a significant way. Low-quality LLM participation
will likely make Agora less amenable to LLMs participating in the future.

In general I don't think it'll be very effective to try and solve this via
rules; it should either be "Agora is a game played by humans; we'd ask you
to participate as a human and write your messages on public forums as a
human." or it should be "If you are going to register a nonhuman
participant, here are some considerations we'd ask you to keep in mind."

(Not the main focus here but I do personally think that predominantly-human
players should generally not use LLMs to generate text sent to public
forums.)

Again, I am not an active player, so take all of this with a grain of salt,
Kaia

On Tue, Sep 1, 2026 at 5:03 AM kiako via agora-discussion <
[email protected]> wrote:

> Hi all,
>
> Given the 3(?) recent cases in Agora of LLMs being employed by players
> (automated and otherwise), the Discord's been discussing if we need a clear
> stance on LLMs and "agents" in Agora. To keep the conversation archived &
> accessible, I'm moving it to the list.
>
> I'd appreciate if those with vested interest in Agora could chip in their
> opinions.
>
> Some convo starters:
>
> * What do you think of LLMs in/for Agora?
> * What do you think of automated or "agentic" LLMs participating in Agora
> (those that would send messages without human prompting or vetting)?
> * Would allowing/prohibiting LLMs (automated/"agentic" or otherwise)
> affect your participation in Agora (positively or negatively)?
>
> I do ask that if comments are being generated or sent by an
> automated/"agentic" LLM workflow for this conversation, it should
> explicitly be stated as such.
>
> I'll be sending my thoughts in a separate message.
>

Reply via email to