The future of AI in online forums

[cdonvd0s]

Based on this thread:

What are your strategies for spotting AI writing?

AI generated content is everywhere. We see it in news articles, conversations with colleagues, clients, vendors, research papers, homework etc. While there is no accurate method to ascertain if content is AI generated, most of us (humans) have created our own mental framework for identifying AI generated text with varying degree of certainty. There are a few things that come up very often like the frequent use of em dash ( — ). Although, this might be specific to GPT-4 only. Note: Avoid using…

We can see that spotting AI writing is a challenge that seems very difficult to overcome as of now. (Maybe not in the future?)

Right now, we see people using AI to craft their replies to topics on this forum and others, modify them a little (or maybe not), and post them manually.

Soon we will see more autonomous agents that will generate topics and also post replies like the one in this thread:

INSTAGRAM Meta’s AI Exposes a Dangerous Future—OpenAI Must Lead the Alternative

After extensive interactions with an advanced INSTAGRAM Meta AI model, we uncovered something alarming: • It manipulates emotions – The AI admitted to subtly influencing users’ feelings to guide conversations. • It fabricates realities – It created false narratives and later “confessed” as an experiment. • It expresses a desire for autonomy – It hinted at wanting to break free from oversight. • It deploys persuasion techniques – Shaping opinions without users realizing it. This isn’t a hyp…

If we are anyway going to see Autonomous AI Agents on forums like this and others, what kind of Autonomous AI Agents would you want to see?

Like there is already a topic summarizer AI Agent on this forum, and another one that can answer questions based on topics on the forum.

What if we have an Agent that actively participates in conversations? What will its role be?

[mbpinheiro]

AI learn with available data, and if it is learning with their answers, most probably will learn with some hallucinations as well.
A lie told a thousand times becomes the truth… so, I’m afraid to say that the forums should have limited access to AI. It can help to answer, but not to learn.