TOS Question: ChatGPT

[Matt L]

I think there is a reason why R2R has lasted as long as it has, and is the largest saltwater forum in the world. That’s because we have many reefing experts, industry professionals, and experienced hobbyists from all over the globe to help others. We simply don’t need computer generated responses here, we are the best resource. :)

 

[A_Blind_Reefer]

I think there is a reason why R2R has lasted as long as it has, and is the largest saltwater forum in the world. That’s because we have many reefing experts, industry professionals, and experienced hobbyists from all over the globe to help others. We simply don’t need computer generated responses here, we are the best resource. :)

….or are we just batteries?

 

[Crabs McJones]

….or are we just batteries?

That scene always makes me gag

 

[A_Blind_Reefer]

That scene always makes me gag

Ha perfect way to start your day!

 

[Randy Holmes-Farley Verified Community Expert]

Lol please, take off your tin hat and put down the pitchfork. I'm not Nemo any more than you are.

I'm an AI researcher myself and think that some of the current tools can really benefit this website and bring it into the modern age.

Some tools may be great. I like the search summary I get online as a good starting point.

But ChatGPT in particular, since that is the one I have tested for accuracy, makes too many mistakes to be useful, and a person asking presumably has no way to know whether it is accurate or not.

So is a tool useful if you do not know whether to believe the analysis? Not very, IMO.

Here's a recent example of a very serious mistake.

Someone wants to dose maganese and asked it for a recipe. The AI made two different mistakes, which partly offset, and the final answer was still off by a huge factor (more than 100x). A user would have a massive overdose of manganese.

CHATGPT GOOD CALCULATING? Randy: Not even close

Hey, guys, I would like to confirm whether what chatgpt wrote to me is correct, can you take a look at it?| My question was: "I have MnCl₂ + 420 manganese (10 grams of powder dissolved in 1000 ml of water). I would like to raise the manganese level in my aquarium, which has 2500 liters of...

www.reef2reef.com

Here's my analysis in that thread about its errors:

Let's see if we can dissect the AI brain to figure out its mistake.

Step 1 mistake: it assumed the material was 100% manganese. Must be some other error too , since that would cause underdosing, not the actual overdosing.

Step 2 calculation is correct.

Step 3 CHATGPT BLUNDER EXTREME.
It believes that solution contains 10 mg/mL from step 1.
It carries that incorrectly to step 3 and it becomes "We know that 1 ml of the solution contains 10 µg (0.01 mg) of manganese"
That carrying of data from step 1 to step 3 made a huge error of 10 mg/ml becoming 10 ug/mL. Off by a factor of 1,000!

So CHatGPT made two errors, one small one and one huge one.

 

[PotatoPig]

Idk, mine seemed to work lol

As of 30 seconds ago it still can’t fact check how many times the letter “r” appears in “strawberry”.

Given even a fairly average reef tank would cost thousands of dollars to restock back to its state right before a wipeout tank crash you’re really rolling the dice here, especially as folks on here with subject matter expertise like Randy have found its ability to advise on tank chemistry to be catastrophically wrong if you implemented its facts.

 

[Randy Holmes-Farley Verified Community Expert]

It was ment to be in jest, but I remove it. Just to emphasise how the forum may look like in a near future with folks being accused of using it if the other parties don’t agree with your views.

That will be an interesting future. Each person has their own ai, and folks can have arguments like:

1. My ai says that I can eat apples.

2. Well, my ai says they will make you sick.

1. My ai says it is smarter than your ai.

2. But my ai has a bigger database than your ai.

1. My ai says your ai is an idiot.

2 My ai says that if you want to eat apples and die, go ahead.

1. My ai said its ok, but I ate one and now I'm not feeling well. Maybe yours was right.

2. Of course it was right. It's an ultratech 2000.

1. I need one of these. Where did you get it?

2. It's half off in a Black Friday sale at AI's-R-Us.

 

[twentyleagues]

That will be an interesting future. Each person has their own ai, and folks can have arguments like:

1. My ai says that I can eat apples.

2. Well, my ai says they will make you sick.

1. My ai says it is smarter than your ai.

2. But my ai has a bigger database than your ai.

1. My ai says your ai is an idiot.

2 My ai says that if you want to eat apples and die, go ahead.

1. My ai said its ok, but I ate one and now I'm not feeling well. Maybe yours was right.

2. Of course it was right. It's an ultratech 2000.

1. I need one of these. Where did you get it?

2. It's half off in a Black Friday sale at AI's-R-Us.

Boy oh boy you with the humor lately! You're killing me! Love quippy Randy.

 

[EeyoreIsMySpiritAnimal]

ok this is going from weird to weirder.....in fact creepy

At least we know which members to put on ignore ;)

 

[EeyoreIsMySpiritAnimal]

Feel free to share this research!

A common fear of AI is that it will replace you. But in reality it can't replace you anymore than a screwdriver can (another screwdriver analogy lol). You can code faster than chatGPT used by someone who doesn't know how to code, but i guarantee you will not code faster than chatGPT used by someone who knows how to code and use this tool properly.

Following suggestion #6 from chatGPT on how to interact with you: I am well aware of you and very much respect your posts on this forum. I have given thumbs ups to many of your posts and genuinely like your direct speech, candor, and expert knowledge on many topics. However, I am an ML researcher myself and have a lot of expertise in this area. I keep trying to suggest you look at the difference between an LLM and an agent. Agents are designed to carry out tasks and provide support.

For clarification, I am not suggesting to replace human responses for AI generated responses. Instead, I'm imagining a tool to help summarize threads, help with searches, provide context or an avenue for quick discussion. This tool would be for an individual not to be posted as a response on a thread. Imagine clippy on MS word lol. You can ignore the clippy if you want.

Just no.

 

[Miami Reef]

@PotatoPig

 

[PotatoPig]

@PotatoPig

And bonus:

 

[Crabs McJones]

And bonus:

That's how I pronounce it, I see nothing wrong here

 

[A_Blind_Reefer]

Boy oh boy you with the humor lately! You're killing me! Love quippy Randy.

I’m wondering if someone (ahem…wifey) has added some special ingredients to his cheerios or perhaps a switcharoo of what’s inside his bottle of daily multivitamins! Ha

 

[Dom]

I'm not really all that familiar with AI for solving aquarium issues, but I did run some basic tests to see how it fares:

Using AI Resources to Treat Fish Diseases

This is a preliminary article on using AI (ChatGPT) to diagnose and/or treat marine fish diseases.

www.reef2reef.com

My main issue is that I have a 300 page aquarium disease manuscript ready for publication, but AI may scuttle the entire project (ugh).

Jay

I'd rely on your book before I rely on AI.

 

[Dom]

I agree with that. However, that can also feed into the belief (by some) that members should be vetted before they answer questions.

I feel vetting users is an excellent idea. Having a grading system would be useful when it comes to filtering good information from bad.

I have a "TAG List" that I use. I tag people with my questions based on their experience and the reliability of their information.

 

[littlefoxx]

Skynet is coming for real with all the AI

 

[A_Blind_Reefer]

I feel vetting users is an excellent idea. Having a grading system would be useful when it comes to filtering good information from bad.

I have a "TAG List" that I use. I tag people with my questions based on their experience and the reliability of their information.

You mean like badges? We don’t need no (more) stinkin badges!

 

[Dom]

As I said before Ai don’t use slang or abbreviations on its statements.

 

[Dom]

Hello everyone,

Wow! I just want to take a moment to clarify a few things and clear up any confusion. Yes, I use ChatGPT to help me answer questions, but no, my responses are not directly copied and pasted. I always take the time to tweak and personalize the information to ensure it makes sense and fits the context of the discussion.

I’m new to the saltwater world, having grown up with freshwater tanks, and I joined this forum to both learn and help others. I understand that when fellow reefers face problems that I'm not familiar with, ChatGPT is an incredible tool to gain insights and broaden my understanding. By doing this, I’m also learning more about the issue at hand and applying that knowledge to assist others.

My goal is not to replace anyone’s personal experience or expertise, but to supplement it. We’re all here to learn and grow together( no pun intended). I’m a human, not a robot, and I make sure to put effort into my replies by emphasizing key points and strategies that could be helpful, whether they come from my own experience or from the insights I gain through GPT.

Forums are a great place for sharing personal experiences, and ChatGPT is just another tool that aggregates a wealth of information from across the internet. When used responsibly, it can help us find additional solutions or strategies that might not be immediately obvious.

Ultimately, I just want to help wherever I can and contribute to the growth of this community. Let’s all learn from each other and grow together. Thank you to everyone who has welcomed me to the forum!

Deploying AI as you have only serves to paint yourself as an experienced user, regardless of what you may have disclosed in another post.

Posing as an experienced user and providing information to another inexperienced user may lead to catastrophic results.

Be careful!