Revisions to the policy on scripted agents
Bumping Pixels
In the interest of positive change and discouraging misuse and abuse of the Second Life platform, I want to propose the following changes to Second Life's policy on scripted agents.
- Scripted agents must be linked to a botmaster "main" account, this would work like a partnership request such that the scripted agent account dashboard would let you pick the botmaster account, and that association must be accepted by way of an email confirmation or dashboard interaction on the human operated account. Botmaster accounts may have multiple scripted agents associated, and they can view this list on their account dashboard at any time. Failure to have a linked botmaster prevents the scripted agent from logging in.
- To be eligible to be a botmaster, a human operated account must complete a short terms of service quiz similar to the mesh IP upload exam.
- Payment information must be on file for an account before it can be botmaster for a scripted agent account.
- Profiles of scripted agents contain the "scripted agent" account type rather than "resident", and have a clickable SLURL to the profile of their linked botmaster that any observer of the profile can view and navigate through.
- Before completion of the scripted agent setup process, a text field must be populated with the justification and/or purpose of the scripted agent, e.g. "Region moderation", or "Group chat monitoring".
These changes would ensure several things, most notably instilling a sense of accountability and transparency around the concept of scripted agents.
Benefits of these changes would include but aren't limited to:
- Residents encountering a scripted agent may without any difficulty identify the resident who is responsible for the scripted agent, and contact them if necessary which may due to a malfunction, or misunderstanding of the scripted agent's purpose.
- Scripted agents would become detectable by the viewer, so that they can be identified differently and with notice banners on profiles, agent interactions, and messages.
- The provided justification and/or purpose for the scripted agent can be included on messaging in the viewer which reduces overall confusion.
Log In
Madi Melodious
Works for me.
Streex Foxtrot
I agree with all of this! PLEASE lets get some common sense Scripted Bot rules in place.
Dana Enyo
I'm not sure of all the details (although I think I probably agree with them all), but overall I just want all scripted agents/bots to be clearly identified up front so I can block them. I'm in SL to interact with humans, not robots.
Bumping Pixels
Dana Enyo Scripted agents often serve important jobs, like greeting you to a new region or store, sending invitations to invite-only groups, or issuing moderation warnings. Second Life residents do not typically "interact" with most scripted agents as they lack any coding to do beyond their purpose.
The Discord group chat relay bot I developed does not answer to IMs, nor does it message anybody, it just copies messages from a group chat into Discord, and copies responses back from Discord into the group chat. This facilitates customer support while not inworld, and a more connected community.
CasperTech employs scripted agents to manage support tickets, and moderation in their group, without them the lives of the support representatives there would be much harder as the bots automate a lot of manual repetitive work.
I think you misunderstood this as relating to the character AI program, or conversational chat bots. While this technically applies to them, they're only a tiny subset of what this policy governs, and what these changes would influence.
Dana Enyo
Bumping Pixels Thanks for your response. These edges get very fuzzy sometimes, and it can be really hard to know where a particular use of these things falls. (You can make a pretty good analogy to the "Press 1 for reservations, press 2 for..." telephone bots that plague customer service calls...). I'll hang pretty tight to the idea that ANY of these things that could be mistaken for a human be clearly identified.