Are you getting advice from AI on Reefing questions?

[KrisReef]

Every search or product review starts with an AI summary. That being said, it's not perfect. Their are online jobs out there right now for experts in field to check up on and teach AI how to improve upon its answers.

I cannot image a worse job for anyone to be employed at. I would rather argue with a real person than bang my head against a keyboard with Automated Insanity.

People have to work, it (if) you like talking to a machine I won't judge or admire,.

 

[KrisReef]

I'm getting po'd at *reef article writers* who appear out of nowhere posting nine articles a day, there's something about ai tone i can sniff

Comprehensive Article About Understanding Diatoms- Manta Systems Article

New Article Release! "Diatoms: Causes, Solutions, and Preventions" Diatoms in your saltwater aquarium can be both a common occurrence and a nuisance. These brown algae are often one of the first challenges for new and seasoned aquarists alike. Our latest guide dives into the causes behind...

www.reef2reef.com

Now I think I can identify some of that tone, at least for myself I often make errors in my spelings but AI generally had a better understanding of language and punctuation.

I think it has a place in improving communications between speakers of different languages but it isn't a mind, its just a mine.

 

[brandon429]

There's a dryness to it. That sedated, blunted affect seemingly created by changing the page breaks and bullet point headers + background art of whole articles returned by chatgpt

He he

Really though, if that's a real person in those articles I apologizeish. This is merely the gauntlet expected, total sarcastic rejection until I get shown that isn't ai. It seems like ai 1000000000000000000%

All writers need to be able to reef on command, or retire. By reef on command am meaning in addition to writing, and posting, I see you alongside us in work threads taking accountability for outcomes you predicted in your published articles.

If someone writes about how to fix gha and i can't find ten threads of them fixing it for others: they're getting ribbed mercilessly

 

[MrGisonni]

I cannot image a worse job for anyone to be employed at. I would rather argue with a real person than bang my head against a keyboard with Automated Insanity.

People have to work, it (if) you like talking to a machine I won't judge or admire,.

The pay is around $20 an hour. Not bad for sitting at home.

 

[I never finish anythi]

Nope , I just come to reef2reef instead. Screw ai

 

[rtparty]

Eh, I don’t use AI for reefing but I’d honestly trust it as much or maybe even a little more than the vast majority of other reefers.

The amount of poor or misinformation on the forums, groups, IG, FB, you name it…is mind blowing.

AI is only going to get better and hopefully it can pull more correct information than not.

I’ve asked a few very basic, new reefer questions and it was surprisingly good. Even if some of the information was slightly wrong, it wouldn’t wipe out a tank or even harm anything. The new reefer using it would 100% be on a better track than they were before asking the question. I’d say they would be on a better track than asking 80% or more of the current hobbyists

 

[drewzaun]

I hate AI, at least as it is now. It might be cool to have a computer or droid that can learn and talk and interact, but right now all I see is a computer program crawling the web stealing intellectual property and giving it away to anyone who asks for it, while serving ads and stealing your IP data.

Despite all the ways this kind of tech could help us experience our world more fully, it’s just making everything fake and cheap.

FWIW it appears to have a concise answer to questions and I’m sure a lot of people are enjoying it, but I’ve seen the movies I know where this winds up.

Maybe I’m just a grouch idk.

 

[Faurek]

Based on your picture you must like AI, but AI will never have years of experience and actual eyeballs to look at the animals reactions and just reefing instinct. Yes it will give general answers, but it's a no good imo.

 

[seths]

So I use it a good bit. But I’ve got projects and GPTs setup specifically for my uses.

Apex programming is a big one. A lot of schedule based stuff where I can say to adjust and it provides code. Had a lot of errors on this originally. Double defer statements and incorrect use of min-time.

I do use it for tank based parameters but the project for that is coached on a lot of specific setup items. And then I have DB tied in that’s got historical parameters, supplements on hand, livestock tracking, and parameters.

For me, I had to be smart enough to push the GPT in the right direction.

What has been very helpful is getting explanations on things like the correlation between elements. And letting it force me to think more critically.

My current lot of salt is lower on MG and when talking about why consumption was up it recommended double checking the water change parameters - which was the issue.

 

[seths]

… duplicate post in error.

 

[BeanAnimal]

All writers need to be able to reef on command, or retire. By reef on command am meaning in addition to writing, and posting, I see you alongside us in work threads taking accountability for outcomes you predicted in your published articles.

If someone writes about how to fix gha and i can't find ten threads of them fixing it for others: they're getting ribbed mercilessly

This gets old, and is another back handed attack against the bulk of the community, its participants and contributors. A person does not have to demonstrate to you or anybody else via “work threads” or “jobs” that their advice is valid.

The irony here is they all you do is reference “work” done by other people when you haven’t had your hands in anything larger than a 2 gallon tank in decades, if ever. To that end, almost every one of those threads involves contention between you and the rest of the community, regarding your rhetoric and rules about that “work”. Yet here you are bragging about “mercilessly ribbing” people that don’t meet your standards of reputability?

 

[seths]

This gets old, and is another back handed attack against the bulk of the community, its participants and contributors. A person does not have to demonstrate to you or anybody else via “work threads” or “jobs” that their advice is valid.

The irony here is they all you do is reference “work” done by other people when you haven’t had your hands in anything larger than a 2 gallon tank in decades, if ever. To that end, almost every one of those threads involves contention between you and the rest of the community, regarding your rhetoric and rules about that “work”. Yet here you bragging about “mercilessly ribbing” people that don’t meet your standards of reputability? Jaw dropping.

I remember one of my first interactions with Vic and Josh at WWC and they basically said to find someone successful and stick to that following. Not to take in input from too many people.

I guess that will be the game - what is success and who does it end up modeling. Or how do you coach it that way at least.

 

[BeanAnimal]

The reason I am asking you is two-fold:
First, I am curious if you use AI and how it worked out for you. I have included below two examples of

I use it a good bit for work, and as Randy said, an enhancement to search. It can’t be trusted (at all) and you must verify everything they it says. It can’t only regurgitate what it reads, and does so mostly as fact, be it lore, misinformation, etc.

Make no mistake, it does not reason or use logic like you think it does. It matches patterns in language and fills in blanks based on context.

 

[seths]

Randy, can you provide a good example where ChatGPT 4o or similar makes a mistake to demonstrate? I found that this was the case a few months ago, but since 4o came out (presumably, other models are similar--I only use ChatGPT), the accuracy has dramatically increased.

It took a while to get it to correctly calculate supplementation. It still makes errors.

It told me to dose 300 mL of magnesium today due to error in the ratio of increase per gallon to tank volume.

I responded saying something is wrong with your math and the way you’re calculating the ratio and it caught it’s error.

Not sure how someone inexperienced would know that. Also no way to guarantee no errors and not kill everything…

 

[BeanAnimal]

I remember one of my first interactions with Vic and Josh at WWC and they basically said to find someone successful and stick to that following. Not to take in input from too many people.

I guess that will be the game - what is success and who does it end up modeling. Or how do you coach it that way at least.

By all means, that is not bad advice. Success is whatever you define it as. We are all in this for different reasons, with different expectations and skills. What is failure for one may be the pinnacle of success for another.

 

[BeanAnimal]

Not sure how someone inexperienced would know that. Also no way to guarantee no errors and not kill everything…

They wouldn’t. And the AI therefore not only breeds misinformation, but it also sucks it back in as it is repeated by those who learned from it. Scary huh….

 

[BeanAnimal]

Is coming…

 

[gbroadbridge]

It took a while to get it to correctly calculate supplementation. It still makes errors.

It told me to dose 300 mL of magnesium today due to error in the ratio of increase per gallon to tank volume.

I responded saying something is wrong with your math and the way you’re calculating the ratio and it caught it’s error.

Not sure how someone inexperienced would know that. Also no way to guarantee no errors and not kill everything…

If you have to do the maths yourself and correct it, then it's a useless time wasting tool.

You may as well do it yourself correctly to start with, and if that means picking up a text book and actually learning something for yourself, well that's a good thing.

 

[SliceGolfer]

I work with AI every day. I find it useful in calculating dosage amounts, wire framing documents, and rewriting summaries in a creative and succinct fashion. Image creation is fun for the Battlecorals postings.

 

[Jungle Blue]

GPT has helped me in Product vs Product breakdowns, Quick Tips/Reference, Stocking, etc.
Was pretty handy when I didn't know the proper terms of certain plumbing parts. All I had to do was input the system purpose description with the main components and asked what parts needed to connect them. Dimensions and all. Insightful at times and usually quicker than google or waiting for a forum answer (sometimes the same questions flood the feed).
However, for deep dives and the more experimental part of things, I like to dig into R2R. & to hang out

Use it as a tool, everyone's results vary within their own tanks.
Tbh it helped me brainstorm with myself; I was able to ask more questions and there were a few times I had those "hmm, I didn't even consider that. neat. yes, expand please".

A successful reef still requires some hands on to be hands off.