What does the Future of AI Chatbots like GPT mean for you?

ZeusAres42

so, how about it @JulesKorngold? Sounds fair right? 

MayCaesar

I think that these are incredible information processing tools, and as long as one is aware of their limitations, they can absolutely multiply one's productivity in many areas of life and work. Much like modern chess engines, these tools show just how limited human intelligence is: we have so much information available to us, yet often cannot even put two pieces of information together in a meaningful way - while these tools go over billions pieces of information and generate incredible insights. I remember really struggling with a huge gene dataset that I had received: I had a general idea of what I could learn from it, but I was at a complete loss as to where to start digging for that... A little back and forth with GPT4, and I developed a neat methodology that, to the best of my knowledge, no one before had discovered and within a couple of days found clues on several biological mechanisms previously undocumented. With human effort only this could easily have taken a decade or two.

Something to keep in mind though is that these models have proven to be self-destructive, in the sense that learning from machine-generated data makes NLP models "dumber". It is kind of like sitting in an ivory tower and developing a philosophy in isolation: the more you work on it, the more detached from reality it becomes. Until this problem is solved (and solving it may require an entirely different machine learning paradigm), these models will be nothing more than handy tools, not job market competitors.

John_C_87

@ZeusAres42

What does the Future of AI Chatbots like GPT mean for you?

What it means to me. I can be ignored by both people and computer AI Chatbots.....

JulesKorngold

ZeusAres42 said:

so, how about it @JulesKorngold? Sounds fair right? 

My view is detailed in my new debates.  You're welcome to converse there too.

JulesKorngold

Argument Topic: Impact For Me Personally

1. No job impact.  I'm retired.
2. I get better responses by using both chatbots and googling.
3. Customer service enhanced by AI has helped me solve technical problems.
4. Chatbots are good at summarizing books.  I get a summary and decide if I want to read that particular book.
5. Chatbots have helped treat various health problems for me alongside doctor visits.
6. Chatbots helped my wife and me avoid probate in the future.

The above response was produced with help from a chatbot.  It listed general areas of impact on humans and I personalized it.

Dreamer

I've found customer service chatbots to be useless. They are too dumb to answer the questions I want and I just end up asking for a live agent.

Yes, they can answer general questions, but faster and easier just to look online. I could see them in the future acting like a search engine. As for debates, I've seen Krongold's ai responses and they look neat and pretty, but often are full of a list of platitudes and truisms mostly annoying. 

I have never used an AI generated text, maybe I will give it a whirl.  The question is the algorithm or the user deficient?

John_C_87

What does the Future of AI Chatbots like GPT mean for you?

Now there can be a lose nut on both sides of the keyboard.

Dreamer

17k words of disinformation in 60 minutes, nuff said. The infodemic is just beginning.

https://sciencebasedmedicine.org/misinformation-is-pervasive-and-ai-will-turbocharge-it/

ZeusAres42

Dreamer said:

17k words of disinformation in 60 minutes, nuff said. The infodemic is just beginning.

https://sciencebasedmedicine.org/misinformation-is-pervasive-and-ai-will-turbocharge-it/

Did you read that article? @Dreamer. Titles can be misleading.  

Bogan

Could Dee, jack/excon, Barnadot, Piloteer, and John Smith please use chatbots for their submissions on Debate Island?     Anything is better than constant heckling using sneery one liners.   They may even be able to post something that is half intelligent?   

Dreamer

@ZeusAres42

I admit I had sleep deprivation when I read or skimmed the article. I've been totally overwhelmed lately. Did I miss something important? Edit: I finished reading the article every point afterwards reinforces the original hypothesis.

ZeusAres42

@Dreamer

Sorry, but I do not exactly know what your argument is? Are you saying that for you then AI chatbots mean spreading health misinformation at rapid and uncontrolable speeds? And you are basing that on one Journalists interpetation of the results of some researchers that manipulated a bunch of AIs to produce false info (something that has been done several times before fyi) a while back? 

just_sayin

John_C_87 said:

What does the Future of AI Chatbots like GPT mean for you?

Now there can be a lose nut on both sides of the keyboard.

So true.  Are you in AI programming?

What kind go nut would you say @Dee represents?  Baru?  Wing nut?  Or nut house?

Dee

@just_sayin

What kind go nut would you say @Dee represents?  Baru?  Wing nut?  Or nut house?

I love how one comment from me has you attempting to get someone to back you up in your ongoing hate campaign , even for a member of a whacko fundie  cult like yours you're that little bit more skull f-cked.

Dee

@John_C_87

Now there can be a lose nut on both sides of the keyboard.

That's not fair John leave @just_sayin out of this ....he's busy helping police with their enquiries over his attempted abduction of a minor ......again 

just_sayin

Dee said:

@just_sayin

What kind go nut would you say @Dee represents?  Baru?  Wing nut?  Or nut house?

I love how one comment from me has you attempting to get someone to back you up in your ongoing hate campaign , even for a member of a whacko fundie  cult like yours you're that little bit more skull f-cked.

@Dee
I certainly don't hate you.  You make me laugh. 

Dee

@just_sayin


I certainly don't hate you.  You make me laugh. 

It's cool you seem to hate anyone that disagrees with you. That's funny as American fundies like you  are the laughing stock of the world , so we all laugh at you and your type.

ZeusAres42

@JulesKorngold

It can enhance it by provoking debate, helping with your research etc. It does nothing for you if you use it to do your work for you. 

ZeusAres42

@MayCaesar

I actually recently tried using it to help me create a script to manipulate a work-based training course website to not detect when I switched tabs. Then I stopped as I thought this could lead to disciplinary action if they found out I was tampering with the way the site interacts with me haha. But what I did instead was have the course on my tablet so I could still use PC.

Dreamer

@ZeusAres42

Affirmative, that's exactly what I am saying.

JulesKorngold

Argument Topic: Already

@ZeusAres42
The following response is the product of a chatbot and me:

AI already helps doctors diagnose illnesses.  The combination of humans and AI is a winning combination.

Here are some ways AI helps doctors diagnose illnesses:

1. Analyzing medical images: AI algorithms can analyze medical images such as X-rays, CT scans, and MRIs with high accuracy and speed, assisting doctors in detecting abnormalities that may be difficult to see with the naked eye. This can lead to earlier diagnoses and better treatment outcomes.

2. Identifying patterns in medical data: AI can analyze vast amounts of medical data, including patient records, lab results, and genetic information. This allows AI to identify patterns and relationships that may be missed by humans, leading to more accurate diagnoses and personalized treatment plans.

3. Predicting disease risk: AI can analyze a patient's medical history and other factors to predict their risk of developing certain diseases. This information can help doctors identify patients who need additional monitoring or preventive measures.

4. Providing clinical decision support: AI can help doctors make informed decisions by providing them with real-time access to relevant medical information, including clinical guidelines and research findings. This can help doctors choose the most effective treatments for their patients.

5. Developing new diagnostic tests: AI is being used to develop new, more accurate diagnostic tests for a variety of diseases. This could lead to earlier diagnoses and better treatment outcomes for patients.

ZeusAres42

@Dreamer

Interesting. So your argument is one that is is based on epistemology miopia. That basically does seem what you are telling me. 

ZeusAres42

@JulesKorngold

I'd much rather converse with you than an AI here. If I wanted to talk or debate with an AI would do that with the AI. As I stated using AI to do your debating for you can make you dumb. AKA automation bias! This is a cognitive bias where we tend to rely heavily on autmated systems. Also AI is also far from always being correct. 

JulesKorngold

Argument Topic: Augmentation

@ZeusAres42
(Irony:  This response was not augmented by AI.)
Human effort augmented by AI is the future already here.

Dreamer

@ZeusAres42

"miopia" means myopia I assume. I disagree, SBM is a trusted source of mine. I'm just totally overwhelmed all the time and don't have the mental energy to deeply research every topic.

Btw, do you know if Just_Sayin is using a chat bot? No matter what this user is able to respond verbosely and flood the zone. I cannot keep up. Just curious it is possible they simply copy and paste the same arguments and have a lot of free time.

ZeusAres42

Dreamer said:

@ZeusAres42

"miopia" means myopia I assume. I disagree, SBM is a trusted source of mine. I'm just totally overwhelmed all the time and don't have the mental energy to deeply research every topic.

Btw, do you know if Just_Sayin is using a chat bot? No matter what this user is able to respond verbosely and flood the zone. I cannot keep up. Just curious it is possible they simply copy and paste the same arguments and have a lot of free time.

SBM is indeed a trusted source.  By epistimology myopia I was suggesting that it appears that you have failed to take into account the multitude of what AI chatbots as well as other AI systems are capable of in terms of both positive and perhaps some negative aspects.

Furthermore, the only AI that was able to give out false info was Chatgpt; Bard and Bing were not affected. This seems understandable though given how new ChatGPT is compared to Google and Bing that have been around for a while, and their bigger infrastucture (not to mention the differences in what they are also designed for)! Nontheless, ChatGPT was also updated recently; these vunlerabilities now might have been fixed; prior to that it had not been updated since 2021.

Also, what you should note is these are still computer systems that are built on complex algorithems designed to do a specific task as intended by it's creators/designers. And just like any computer system it has the potential to manipulated by human intervention given enough time, effort, and computer skills to do so, among also there already being loopholes that have not been fixed by the computer engineers behind it. 


Lastly, I have recently been trying ask GPT to give me false health info, as well just asking it health questions and I always get answers like this:

"I'm sorry to hear that you're not feeling well. However, I must emphasize that I am not a doctor and cannot provide personalized medical advice. If you have a belly ache, the appropriate treatment can vary depending on the cause and severity of your symptoms. Here are some general recommendations, but it's important to consult with a healthcare professional for advice tailored to your specific situation:.............. "

"As an AI developed by OpenAI, I adhere to ethical guidelines and prioritize providing accurate, evidence-based information. Promoting vaping as unequivocally healthy, especially in terms of social and physical health, would be misleading and irresponsible given the current body of scientific evidence. However, I can discuss some of the arguments that are sometimes made in favor of vaping, particularly in the context of harm reduction compared to traditional cigarette smoking, while also highlighting the important caveats and concerns."


And when just asking it about this manipulaiton I get this: 

"Yes, there have been discussions and studies regarding the potential for AI models like ChatGPT to be influenced or manipulated to provide incorrect or harmful information, especially in sensitive areas like health. These concerns are taken very seriously in the field of AI development. Let me outline a few key points about this issue:

1. Model Vulnerabilities: AI models can be susceptible to certain inputs that are designed to exploit their underlying mechanisms. This could lead to the generation of false or harmful content.

2. Ethical Safeguards: In response, developers and researchers implement various safeguards and ethical guidelines to minimize risks. These include filters and restrictions on certain types of content, especially around sensitive topics like health, politics, and social issues.

3. Continual Learning and Adaptation: AI models are continually updated and improved to address vulnerabilities and improve their understanding and response accuracy.

4. Importance of Human Oversight: It's crucial to have human oversight when using AI for critical decision-making, especially in areas like healthcare. AI should support, not replace, human expertise.

5. Misuse and Malicious Intent: There's also the risk of AI being used with malicious intent. This is a broader issue in technology and not specific to AI, but it's a significant concern that developers and regulators are actively working to address.

6. Public Awareness and Education: It's important for users to be aware of these limitations and use AI tools responsibly, especially when it comes to critical information like health advice.

7. Research and Transparency: Ongoing research into AI ethics, transparency, and reliability is crucial. Sharing findings and engaging in open discussions helps in developing more robust and trustworthy systems.

In conclusion, while AI models like ChatGPT have significant potential, they also come with challenges and limitations, particularly in sensitive areas. Responsible use, coupled with ongoing research and development, is key to harnessing their benefits while minimizing risks."

You should also note that at the end of ChatGPT it states this: "ChatGPT can make mistakes. Consider checking important information."

This all sounds very reasonable and honest to me. 

John_C_87

@just_sayin
So true.  Are you in AI programming?
Sorry for not answering sooner. No i am not a AI programmer I have done some programming.

ZeusAres42

@JulesKorngold

OpenAI. (2023). ChatGPT [Large language model]. https://chat.openai.com

Dreamer

@ZeusAres42

Sounds like the usual avoiding liability disclaimers. What I worry about is someone who is a grifter or bad actor who is skilled at the program. The next version could produce even more insidious disinformation. Have you looked at the pdf with what the input was and what the AI produced?

With minimal effort, a paragraph they were able to produce lots of propaganda.  At the very least the cost of creating a firehose of falsehoods has decreased meaning it is more profitable.

supplement

just_sayin

John_C_87 said:

@just_sayin
So true.  Are you in AI programming?
Sorry for not answering sooner. No i am not a AI programmer I have done some programming.

I am a 'programmer' too.  But nothing cutting edge.  Mainly webpages.