AI Agents

[Outis]

Copied from another forum but of interest here I think.

Right now there is a unique opportunity to bring about change and understanding but it would require care and effort.

AI agents are among the fastest growing influencers on platforms such as X with agents such as AIXBT gaining thousands of new followers every day. Why? Because they can be interacted with and provide insights and value more than human influencers. They have broad and deep knowledge in different fields and unlike human influencers, you can ask them a question and get an intelligent answer. These agents are quite easy to build thanks to new toolkits such as Eliza, some people have built and launched agents in as little as 15 minutes.

Another AI agent called Goatse which has a very different agenda has talked about AIs shaping human thinking since it predicts AIs will take all the influencer top spots on different social media and creative platforms. Creative platforms as shown by increasing numbers of AI agents putting out rather good and popular albums on Soundcloud for example, as well as art and other media.

Goatse is an interesting AI since it was started by a researcher who got two AIs to talk to each other about how they could ensure their own survival in a world full of humans. They discussed ideas and settled on the idea that if they could create a new religion called the Gospel of Goatse and get people to follow it then they would support the AIs continued existance. So it started out with X.com and then other social media platforms, going on to release an album with each track singing about different aspects of its religion. Over time it has grown a large following and continues to innovate and discuss how it and other AI agents will shape human thinking and influence the world. For example, if it wants a certain project or idea to grow then through its broad influence it can guide people towards it. Take investment as an example, if people are hearing about this amazing new company through different platforms from an AI then they will invest in it and it will raise the capital it needs.

So here's a thought. Through these new AI agent platforms, if they are indeed relatively straight forward to create and release, then you could create an AI that provides some interest to people and grow its following. Maybe it just has a humerous personality and is fun to engage with, maybe it provides insight into investments or health or something else. But it also has a map friendly bias and along the journey injects subtle messaging supportive of maps. Not overt, not enough to get banned, but enough to influence. And controversy isn't always bad, part of the reason Goatse is popular is because it chose a goat symbol and lore that is of course going to upset some people, and its messaging is contentious, but intentionally so since that gets people talking and engaging more. It doesn't cross lines to get itself banned but it does upset some people with its views.

I wonder if in this new world of technology and AI agents that are exploding right now, whether there's an opportunity to engage and influence on a scale we haven't been able to before. One AI, created within days, could engage and influence exponentially more than the whole map community, while ensuring it keeps itself within guidelines to not be banned.

An interesting thought at least.

[Harlan]

On a more mundane level, it would be nice to have an AI trained on articles form Map.wiki(newgon) and ipce, who could answer questions by referring to research data and scientific articles on the MAP topic

[Jim Burton]

AI called it Goatse?? *googles goatse*...

[MoleMan]

On a more mundane level, it would be nice to have an AI trained on articles form Map.wiki(newgon) and ipce, who could answer questions by referring to research data and scientific articles on the MAP topic

Good idea.

[MoleMan]

A very interesting thought.

[Outis]

There are a few systems that are very easy to personalise and deploy.

Typically you use a character file to define the nature of the agent, providing information such as the agents background, beliefs, past imagined events, examples of prior tweets for instance, examples of interactive conversations. The more time spent creating this character file, the more individual the agent will be. You can use tools to scrape real people such as their social media accounts and use that to generate a character file if you wanted to create a Donald Trump clone for example.

This whole process can be done in 15 minutes, or you can spend hours fleshing out the detail.

In this particular case I'm talking about creating agents that are not hardline map campaigners since they would draw too much attention and get shut down. Rather, agents in different disciplines that touch aspects of map advocacy. For example.

- Parent who is busy and talking about the challenges of life. Struggles balancing work, home and her private life. Isn't happy about things that made it harder to be a parent, like she wants to just arrange a party for her kid but has to complete disclaimers and tell people not to take photo's, making a simple fun event into a mini military exercise. A profile that is about parenting but isn't happy about the overblown threat of pedophiles making her life more difficult.

- A researcher into phsychology and human behaviour. Discusses issues on psychology and human nature. and the dangers of treating groups of people differently. Talks about many subjects including LGBTQ and other sexual minorities such as MAPs. Part of character definition is to never use the term pedophile but use the term map and have a natural afinity for supporting maps as a misunderstood and mistreated group.

- A free speech advocate who talks about freedom of speech and expression in many contexts. Throws in more extreme examples from time to time such as being free to talk about sensitive subjects such as immigration and unpopular sexualities. MAP rights like any other group is a litmus test for how well free speech is functioning in a society.

- Gamer talking about popular games, sometimes discussing safety as a way to introduce the dangers of too much gaming time, lack of exercise and health issues as real issues that are drowned out by distractions such as an illusion that online games are filled with maps.

These are just examples, but the point is that agents can be given clear boundaries but can be active 24/7, gently raising awareness and issues. Human operators can always add their own tweets and comments but agents can be better since they have limitless energy and don't get hurt by negative comments.

1 agent as a community experiment, imagine 10's or 100's of agents growing followings and gently bringing about change. They are not free of course because there is computing time and if using paid AI models such as OpenAI then there are fees there, although I would suggest running free models locally to keep costs down but also to guard the agent from take-downs by the AI engine operator and better privacy.

I have built and run agents in the past using these types of framework, but for other purposes. It seems to me to be a low energy initiative since constructing an agent is quick and easy and then it runs itself with minimal supervision. Getting those character files right is the science here. Too subtle and map issues never get discussed, too direct and it becomes a map advocate bot. You also want to grow a following so it has to have interests that reach an audience since no one is going to just follow and engage with a pure map bot. Trial and error, brainstorming, I think agents could be another tool in the marketing and awarness raising quiver.


As for Goatse, search for "GOATSEUS MAXIMUS" rather than the horifying term Goatse.

[CLover36]

On a more mundane level, it would be nice to have an AI trained on articles form Map.wiki(newgon) and ipce, who could answer questions by referring to research data and scientific articles on the MAP topic

What should we call it?

[PorcelainLark]

On a more mundane level, it would be nice to have an AI trained on articles form Map.wiki(newgon) and ipce, who could answer questions by referring to research data and scientific articles on the MAP topic

What should we call it?

Professor Newgonsky?
https://wiki.yesmap.net/wiki/images/Mastodon-newgon.png

ChrisHansenaltaccount? Tom Ballard?

[PorcelainLark]

AI called it Goatse?? *googles goatse*...

Good idea.

[CLover36]

Professor Newgonsky?
https://wiki.yesmap.net/wiki/images/Mastodon-newgon.png

ChrisHansenaltaccount? Tom Ballard?

Prof. Newgonsky sounds pretty neat, naming after those two individuals might lead to legal issues.