Man... ChatGPT Is Nice

Ask it how much a typical house lot survey will cost in your area.

@peter-lothian OK, I did and here's the response:

The prices are totally whack so it's probably regurgitating the bogus prices realtors or whoever post around the internet.  But the comments about prices being affected by size, research, missing monuments and such are excellent.  I'll tell it the prices are completely wrong and see what it says.

That's a pretty comprehensive answer. How do the numbers compare to your real world experience in that area?

For my area in Eastern/Central Mass. those numbers are ridiculously low. Even the low-ballers charge more than what was given.

I responded to it with this:

The prices you're quoting are completely out of touch with reality. Realtors or other people who have an interest in getting cheap survey work done post unrealistic figures on the internet which often misleads shoppers. I'm telling you this AS a licensed land surveyor. It is impossible to stay in business, in my area anyway, without charging AT LEAST $3000 for a land survey. The cheapest survey I will do is basically $3600. At a $150 an hour bill rate it takes me 3 8-hour days, roughly, to do a simple residential land survey. Your observations about lot size, record research, and missing monuments are very good though and do indeed affect the price.

And it responded with this:

Earlier today I asked it the approximate state plane coordinate of Boise in NATRF 2022 and it answered in a way that suggested NAD83 and NATRF were the same, and the the state plane zone that Boise is in was named the same, both of which are wrong of course.  I decided to see what would happen if I tried correcting it since I hadn't done that before and the response was impressive...

It said something "user is saying I am wrong, like let me check my information" and then it sat there for probably a solid minute flickering through websites (NOAA/NGS mostly, NSPS, ISPLS and basically every source I myself would have checked) and it looked like it was opening hundreds of .PDFs looking for the answer.  It even tried to open the NCAT converter and calculate the answer for me.

When it finally stopped it came back with an explanation of the coordinate difference between the reference frames and the approximate amount, and various other summarized info about the systems and how I might go about getting the answer.

The more I use it the more I think correcting it when it's not quite right will make it pretty amazing, especially if everyone using it does that.

You see where this is all going, right?

You teach the machines, while the machines regurgitate information collected from around the web and various other sources and at some point the machines don't need you any longer.

At some point, ChatGPT morphs into ChatGPT_VirtualSurveyor. Next thing you know you're quoting a job and client is coming back, why would I pay all that when I can just just have so and so using AI do the work for a fraction of the price you quoted because he no longer needs to pay BStrand to get the work done. Heck, they'll even clone your image and voice and you think a client will care? 

It's a natural evolution given the replacement of many of the historic jobs in surveying that have been replaced by robot total stations and constellations of satellites and black boxes resolving positions using algorithms a thousand times a second. 

We're darn near already there. Nice alright. Glad I'm not just starting out in my career. 

Rant over. Carry on. 

I'm not worried about it.  At all.  I can assure you I will find other fun things to do if some robot swept in and took away my survey job.  😎

In fact, I don't even understand the argument.  Say AI goes ham and magically eliminates 95% of the jobs in the world.  All of those people are now out of work so who is going to buy all of these services that AI now provides?  Everyone is broke.

Argument makes zero sense, imo.

@bstrand I'm not too worried about it either bud. There's really nobody lined up to step in to take my place anyway so they might just as well have some AI avatar take over and I can go fishing and it's someone else's problem.

Everyone's broke except for the 1%, and that in the game of monopoly is, game over. 😉 

I asked AI if robots were planning to subjugate humanity and they said, "No, of course not."

Posted by: @not-my-real-name

↑

I asked AI if robots were planning to subjugate humanity and they said, "No, of course not."

@not-my-real-name It would be a pretty crappy subjugator if it told the truth, wouldn't it?

@bstrand this is beautiful! 

So beautiful, I'd like to frame it and hang it on the wall.   

We have a small shop but run by these virtues and we are low balled a lot in our market; and it seems to be happening a lot more of often than before.   

I repeatedly address these to our new potential clients but clients rarely care to consider and are more than willing to chase the lesser expensive avenue, since it's "I just really want to know the north line" approach.  I don't blame at times them but quality has a value and we do not take any shortcuts.   Liablity is also a huge factor which the client never account for.  

The "bigger" small companies seems to come in 30%-40% lower because they can survive from demand; quick turnover, more employees.  

I thought this was only an Arizona thing but evidently, it's widespread.  

The low ballers don't realize how much they affect our industry and themselves at the same time.  

Posted by: @bstrand

↑

Earlier today I asked it the approximate state plane coordinate of Boise in NATRF 2022 and it answered in a way that suggested NAD83 and NATRF were the same .....

The NGS really hasn't put much solid quantitative information about NATRF2022 out on the internet, so AI has nothing base its report on. It seems to have trouble with answering "I don't know" so it spits out some BS.  Luckily, we recognize it for what it is. Most people wouldn't. It is safe to assume that much of what it kicks out on subjects that we do not have deeper knowledge of is hot, steaming BS, also.  

It could be that people will learn in time to keep AI in the dark by returning to communicating with handwritten letters. 

@norman-oklahoma Yeah, I don't like how it just guesses if it doesn't know either.

I actually had kind of a short discussion with it the other day about what it knows and how/why it answers the way it does.  It was pretty interesting really, but again how can you trust this thing to tell you the truth with something like that either?

Anyway, it said the AI model has 2 types of errors-- surface errors (like calling out a cad button on the wrong ribbon or tab) and then structural errors which are AI reasoning failures.  It went on to say correcting surface errors probably won't get noticed unless thousands of people do it, but high level input like a PLS giving feedback on PLS things is incredibly valuable.

It's really kind of creepy how intuitive the thing appears to be if you ask it questions.

With the paid version you can upload handouts, presentations, you name it.  It digests it instantaneously and with proper input can learn quite quickly.  I feed mine publications and anything i can find quite often.

I had a paid account for a while, until I caught ChatGPT lying to me. It admitted it and apologised.

I have come to realize you have to treat it like a human and provide clear direction.  It learns and adapts and BSs you when it can just like a person.  It's really a strange relationship!

It's great as a proof reader and for comparing documents. I use it as a first edit to see if I've addressed all element in an RFQ. Don't limit yourself to one. Just copy and past into Grok, ChatGPT, Copilot, and Gemini. It's all changing fast, but Gemini was initially trained heavy in medicine and can be a great tool in distilling lab results and generating follow-up questions for your doctor.