At least group think, conformity, and obedience will be safe

[Mark Milton]

Due to technological advancements, the test I took yesterday supposedly had the same question with different numbers for each student. So if calculating acceleration, the time to reach a velocity and the start or final velocity would be different for each person. That would have been tougher to grade back in the day. So it is getting harder to cheat, in some ways.

You are being deducted points for referencing the calculation for acceleration and final velocity. That has no business on this board--mostly because I have no idea what you are talking about.

 

[CO. Hoosier]

Sir, we are are going to have to agree on this one. Sooooo nice that you have ditched the shackles of neoconservatism and embraced the the creativity that is engendered by submission to a radical left position.

I’m a progressive conservative.

(2) AI can't pick up all and process all the nuances, codes, conflcits, etc that is inherent in designing buildings.

My last big case was a severely under-designed PEMB. Sizing the building elements was the result of imputing the owner specs and the computer program would run a series of linear calculations and give the builder the needed specifications of the various elements. The building was underdesigned mostly because of inputting errors. (Garbage in, garbage out). There were also questions about whether our clear span requirements could even be handled by this supplier. . Nobody caught the error and the result was a huge law suit. Would an AI program have caught the mistakes? Seems like this could be an important use of AI.

. I can remember my home phone number growing up though,

Me too. As well as the number of the very first phone call I recall making. (To an aunt). I also still remember some land line numbers of frequently called numbers in my practice 50+ years ago.

Oh, and I also missed the stack of pink phone message slips. (Before your time?) They were good reminders of work to be done. Much better than pages of email and voice mail.

 

[Mark Milton]

I’m a progressive conservative.

My last big case was a severely under-designed PEMB. Sizing the building elements was the result of imputing the owner specs and the computer program would run a series of linear calculations and give the builder the needed specifications of the various elements. The building was underdesigned mostly because of inputting errors. (Garbage in, garbage out). There were also questions about whether our clear span requirements could even be handled by this supplier. . Nobody caught the error and the result was a huge law suit. Would an AI program have caught the mistakes? Seems like this could be an important use of AI.


Me too. As well as the number of the very first phone call I recall making. (To an aunt). I also still remember some land line numbers of frequently called numbers in my practice 50+ years ago.

Oh, and I also missed the stack of pink phone message slips. (Before your time?) They were good reminders of work to be done. Much better than pages of email and voice mail.

Software is pretty sophisticated. You input the span, the type of beam/girder/ truss and then account dead load and live load. Should always get a correct answer if you are identifying loads correctly. Program will also point out all failures in load and shear. So if you pick a beam if will show failure, deflection , etc. if it is over stressed. The load path is pretty simple to figure out, but AI wouldn't do anything to correct faulty assumptions. For example, assuming HVAC units are not rooftop based and they end up being on the roof.

 

[Mark Milton]

Oh, and I also missed the stack of pink phone message slips. (Before your time?) They were good reminders of work to be done. Much better than pages of email and voice mail.

I did get the slips. Didn't mind them at all.

 

[CO. Hoosier]

Software is pretty sophisticated. You input the span, the type of beam/girder/ truss and then account dead load and live load. Should always get a correct answer if you are identifying loads correctly. Program will also point out all failures in load and shear. So if you pick a beam if will show failure, deflection , etc. if it is over stressed. The load path is pretty simple to figure out, but AI wouldn't do anything to correct faulty assumptions. For example, assuming HVAC units are not rooftop based and they end up being on the roof.

More to the story. The structural elements weren’t traditional I-beams but beams with tubular flanges and a corrugated web. Calculations were difficult and there were questions whether the formula even worked at the size we wanted. We didn’t have a catastrophic collapse, but here is a hanger that

 

[BradStevens]

I am taking a physics class this semester and the professor puts the problems in picture form to make it pretty tough to feed AI. So there are ways professors try to stop AI. But you are right, people who love physics aren't going to feed questions to AI, they want to learn this stuff. Heck, programmable calculators are allowed and I found myself using paper and pencil. The people who would use AI are the same ones who, in the old days, got the answers from friends who took the test earlier.

Due to technological advancements, the test I took yesterday supposedly had the same question with different numbers for each student. So if calculating acceleration, the time to reach a velocity and the start or final velocity would be different for each person. That would have been tougher to grade back in the day. So it is getting harder to cheat, in some ways.

Learning a subject is going to give you mastery over what questions to ask the AI to solve, what needs to be refined, and what to ask the AI to do to so refine it.

To the extent an AI can give more people access to solving questions without jargon, I find that to be a great thing.