AI Members?

[jttheclockman]

Was on another site today and the question came up or at least a scratch of the head is happening there about a so called member being an AI member because of the way the questions and replys they give are sort of not in a normal speech pattern. It got me thinking because this is all you hear about today and that is AI. Would it be possible to fool a web site monitoring system to start having AI members sign up?? Are these areas that far advanced and bet they are. I know Jeff has to deal with spam and other undesirables but is this another threat to forums and could this be another threat to scams we can encounter. Is this the next step to be undetectable? Just some mind numbing thoughts.

 

[1080Wayne]

I would think it is possible , but I`ve also seen unusual speech and reply patterns from long time members here .

 

[jttheclockman]

I wonder if the sites they go to and it ask if you are a robot do they not click it?

 

[wouldentu2?]

You thought all these mmmmmembers are real, do you think there are that many people are making pens in a paperless society?

 

[TurnTheWorldAround]

Interesting topic John, I work a little bit with ChatGPT or what they are generally called Large Language Models(LLM) right now and am enrolled in a Doctor of Business Administration program at the University of South Florida. As advanced as the good LLM's output you would not be able to distinguish or label them as "not normal speech pattern". They are writing better than anyone on this site. But what they may write may not be consistent with what makes sense, especially around a specific topic, especially with a fairly obscure topic of making Pens.

If you are watching any YouTube right now you will see much of that content written with an LLM even using the speech generators and you can catch them with the mispronunciation of words that a person would not mispronounce. We are entering a very new world and I would be highly opposed to allowing a ChatBot just cluttering up our board. Most of the ChatBots are being used on boards like this to run some type of scam and I never underestimate the cleverness of the scams. According to the three travel agencies that call me every week, I have thousands of unused travel credits available. I hope we can this board clear of exploitation and the verification process of who is real or not is going to be an interesting challenge for our future.

Here is a little experiment. The above version was written by me and I will post a version run through ChatGPT version 4 so you can see the comparison.

 

[TurnTheWorldAround]

Here is the same text that I posted above but in the ChatGPT version 4 version

Fascinating subject, John. Currently, I'm engaged with ChatGPT and other Large Language Models (LLMs) and pursuing a Doctor of Business Administration at the University of South Florida. The output from sophisticated LLMs is so refined that it's often indistinguishable from natural speech patterns. Their writing quality can surpass many contributors on this platform. However, there can be inconsistencies in their content, particularly on niche subjects like pen-making.

Many YouTube videos now feature content scripted by LLMs, even utilizing voice synthesis. However, you can spot them by their occasional mispronunciation of words, something unlikely in human speech. We're stepping into an unprecedented era, and I'm strongly against allowing ChatBots to flood our board. Frequently, ChatBots on forums like ours are tools for scams, and the ingenuity behind these scams should never be underestimated. For instance, I regularly get calls from three travel agencies claiming I have numerous unused travel credits. It's vital to keep our board free from exploitation. The process of verifying genuine participants poses a significant challenge for our future.

 

[TurnTheWorldAround]

Here is something interesting. I asked ChatGPT the following question; The interesting part is the LLM responded with this prase, "Ensure Compliance with Policies and Ethical Guidelines", its like putting on the liqour bottle to "Drink Responsibly"...LOL
User
How could I create a chatbot through this LLM to engage with an online bulletin board for a conversation where people and LLM could converse with each other?

ChatGPT (Response to my question)
Creating a chatbot that uses a Large Language Model (LLM) like ChatGPT to interact on an online bulletin board involves several steps. Here's a general outline of how you could approach this task:

Choose a Platform for Your Chatbot: Decide on which online bulletin board or forum the chatbot will be active. The choice of platform may influence the technical approach you need to take, as different platforms have different APIs and integration capabilities.

Understand the Platform's API: If the bulletin board has an API (Application Programming Interface), you'll need to understand how it works. The API allows your chatbot to send and receive messages programmatically. Not all platforms have an API, in which case you might need to use web scraping and automation tools, but this approach is generally less reliable and might be against the platform's terms of service.

Set Up a Server or Cloud Function: You'll need a server or cloud function to host your chatbot application. This server will handle requests to and from the bulletin board and process them using the LLM.

Integrate with the LLM: Connect your server with the LLM, like OpenAI's GPT-4. You'll typically use an API provided by the LLM provider to send user inputs to the model and receive the model's responses.

Implement Conversation Logic: Design the logic for how your chatbot will handle conversations. This includes deciding how the bot will initiate conversations, respond to user messages, and handle different topics. You may also want to implement filters or safeguards to ensure the bot's responses are appropriate for your community.

Handle User Authentication and Session Management: If your chatbot is going to engage in personalized conversations, you'll need to implement a way to manage user sessions and possibly authenticate users.

Test and Refine: Before fully deploying your chatbot, test it extensively. Ensure it responds appropriately in a variety of scenarios and that it adheres to the community guidelines of the bulletin board.

Monitor and Update: Once your chatbot is live, continuously monitor its performance and the responses it generates. Be prepared to make updates to the conversation logic and filters to improve its interactions.

Ensure Compliance with Policies and Ethical Guidelines: Make sure your chatbot complies with the terms of service of the bulletin board, and follows ethical guidelines. This includes being transparent with users that they are interacting with a bot and ensuring data privacy and security.

Engage the Community: Inform the bulletin board community about the chatbot, its purpose, and how it operates. Encourage feedback from the community to improve the bot.

Remember, creating a chatbot that interacts in a human-like manner is complex and requires careful planning and ongoing management. It's important to align the chatbot's capabilities and behavior with the expectations and norms of the community it will serve.

 

[TurnTheWorldAround]

Here is some good news. Some be be taken back from this question I asked ChatGPT but remember if we don't understand how these folks may exploit us then we are just putting our head in the sand instead of being educated with power to vigilant against such exploits. Remember this is fully public avaialbe tool and information so no trade secrets here. The blending of what is real and artificial is starting to converge. I am not a doomsdayer by any means but everyone should have at least a basic knoledge of what is happening.

User(The question I asked ChatGPT)
Can you examine www.penturners.org and find the specific api to use?

ChatGPT
After examining the website of the International Association of Penturners (www.penturners.org), it appears that there is no publicly available API specifically for interacting with their forum. The website is built using XenForo, a commercial Internet forum software package. While XenForo has capabilities for extensions and customizations, these typically require access to the backend of the website, which is not publicly accessible for most users.
In terms of integrating a chatbot with the forum, without an API, the options are limited. You would likely need to collaborate with the forum administrators to create a custom solution. There are discussions on the forum about developing community-supported applications and websites, indicating a level of interest in digital projects and enhancements among the community. However, these discussions are more focused on specific applications related to the pen-turning community, rather than chatbot integration.
For a chatbot integration, you would typically need direct access to the website's backend or an API to programmatically send and receive messages. Without this access, the integration would be challenging and potentially limited to manual or semi-automated methods, which are not ideal for real-time interaction.
If you are seriously considering this project, it might be worthwhile to reach out to the forum administrators or the community to gauge interest and discuss potential collaboration opportunities. However, keep in mind that any project of this nature should comply with the website's terms of service and privacy policies, and should be developed in consultation with the community to ensure it adds value and aligns with the users' needs and expectations.

 

[jeff]

I was just about to respond that we don't expose the API.

 

[TurnTheWorldAround]

One last question I posed to ChatGPT to post I promise. I hope I don't get kicked from this forum because I truly love Penturners.org and been a member since 2008 and only do this to build awwareness of how easy it could be if someone was so inclined to really take over this with an AI bot. Again it does tell us to always respect the terms and conditions of the site just like on the pack of cigarettes it says caution smoking causes cancer... Interesting tool and I would abmonish everyone to at least play with the free version if you have not already of ChatGPT.

User
Is there a tool I can use from the user interface instead of an API to intereact with a bulletin board for chatting with people?


ChatGPT
If you're looking to interact with a bulletin board for chatting with people but don't have access to an API, you can consider using web scraping and automation tools. These tools mimic human interaction with web pages and can be used to automate the process of sending and receiving messages on forums and bulletin boards. Here are a few options:

1. Web Scraping Tools: These tools can extract data from websites. They can be used to monitor bulletin boards for new posts or replies. Popular tools include Beautiful Soup and Scrapy for Python.
2. Browser Automation Tools: These tools automate actions in a web browser, simulating a real user's behavior. Selenium is a well-known tool that can be used for this purpose. It can automate web browsers, allowing you to programmatically post messages, read threads, and respond to posts on a bulletin board.
3. Custom Scripts: Writing custom scripts using languages like Python or JavaScript can allow for more controlled and specific interactions with a website. You can use these scripts in conjunction with web scraping and browser automation tools.
4. Integration with Chatbot Platforms: Some chatbot platforms offer ways to connect with web services without a direct API. Platforms like Dialogflow or Microsoft Bot Framework can be used to design the chatbot's logic and responses, which can then be integrated into your web scraping and automation setup.
5. IFTTT or Zapier: These are more user-friendly tools that can automate web activities based on certain triggers. While they might not offer direct integration with all bulletin boards, they can be useful for automating simple tasks.

It's important to note that you should always respect the terms of service of the website and the legal considerations around web scraping and automation. Automated posting or scraping can sometimes be against the terms of service of a site, so it's critical to ensure compliance and ethical use of these technologies. Additionally, maintaining the human-like interaction quality of a chatbot using these methods can be challenging and may require sophisticated programming and regular maintenance.

 

[TurnTheWorldAround]

I was just about to respond that we don't expose the API.

This forum is strong and well-run because of the labor of love of administrators like Jeff and others. Appreciate your work and vigilance in keeping this forum up and running. Thank you for what you do I know it is a labor of love.

 

[leehljp]

I am on another forum that sees some unusual 1st and 2nd posts - at least it looks unusual to me, and I have asked our admin team to check the origins of the posts. And usually it turns out that I'm correct - they appear to be a spamming group from half way around the world.

Due to spamming, we changed the membership posting to have the 1st and 2nd post to have to be approved by a moderator before others can see it.

The unusual 1st and 2nd posts appear as very generic posts to the thread. They deal directly with the thread subject, but do not add anything. While this can happen in a normal post at times, in certain situations - it began to happen for all 1st and 2nd posts to appear as a generic thread, one to two lines long, perfect English, no misspellings, no questions, no thanks for the info - it all appeared to me as though some non-natural English speaker is using an AI to generate the post.

One or two posts like this can happen. But over 6 months, and have 8 or 10 generic posts consistently like this indicates to me that someone or some group are using an AI generated response. A few years ago before AI, this happened but it was obvious it was from non-native English speakers by a misspelling or wrong word choice. Not anymore. Too natural and English Grammar too perfect. It is the repetitive generic responses that give it away - at least to me.

Forgot to add: All generic responses are in threads 6 to 12 months old or older.

 

[jttheclockman]

That last part that Hank said about older posts is what happened on the site I was talking about too. I believe we are entering into a scary time and not only on forums such as this but web sites where we share our life info. Scamming could be taken to a whole other level. I believe Google has started using AI technology for a lack of a better word. At times you ask a question and a box opens up and it starts typing the answer right in front of you. Just thought I would throw the question out there.

 

[Sylvanite]

I believe we are entering into a scary time and not only on forums such as this but web sites where we share our life.

Have you read "With Folded Hands" by Jack Williamson? AI technology is finally approaching what he predicted in 1947.

 

[jttheclockman]

Have you read "With Folded Hands" by Jack Williamson? AI technology is finally approaching what he predicted in 1947.

This topic can also go hand in hand with what many predicted years ago about Nano Technology and the weird things they could do such as swallow a pill and it can swim to the area in need to be repaired and they can do that from within and so forth. We will all become robots. I will not see it but let me know how it goes when we meet again.

 

[randyrls]

Some web sites I interact with have posted "AI Free Zones" at the top of the main page, page, page,

The Screen Scraping techniques have been around for many, many years.