TxAgswin said:

---

SpreadsheetAg said:

---

dude95 said:

---

Chat GPT is fastest app (or anything else) to 100M users. It was just introduced in November.

Rumors about Chat GPT 5 is being trained right now. Will be in beta by end of year and released next year??? Instead of 2000 char inputs, you could hand it the entire book of 'Dune' in one go. Outputs will obviously be text, speech, image video and possibly touch (physical robotics).

The question is what is AGI (Artificial General Intelligence). How quickly will it be equal to a human in that manner? Earlier in the thread, someone mentioned it couldn't create, just regurgitate what was already there. What exactly is creating? How many times have we as humans actually created something from nothing? New ideas build on previous ideas (same for arts, science, technology).

AI has the ability to self-generate code. It's really in its infancy here and needs human prompting, but it's growing exponentially on a week-by-week basis.

Jarvis from Ironman is just around the corner and it's not science fiction. Vision really isn't that far either (outside of flying and going through walls. Told my kids I've never seen anything like it and it's going to change everything in the next 3-5 years. With one in college, one in high school and one in middle school - their careers will now be totally different.

---

The problem is always going to be who is controlling it / restricting it's outputs?

People will use this for everyday information gathering and then try to use that as a bludgeon towards free thinkers. "We'll ChatGPT says this so it must be true" not realizing it's been curated to give specific outputs or at least omit inputs ...

Just like kids in public school are being indoctrinated - this will be used to indoctrinate the entire globe.

---

If there are gatekeepers for the outputs, or the inputs for that matter, it seizes to be AI - at least as I understand it.

Search engines lost objectivity when they monetized. And that's fine. Capitalism and all that.

This is an entirely different thing. There are certainly inherit dangers with something like this, but I'm not sure why the knee-jerk conservative reaction is that it will be a liberal mouthpiece. If it's truly AI, it won't have a baked-in agenda.

Even if that were to happen, and it was some sort of Trojan horse of liberal indoctrination masked as AI, what's to stop conservatives from designing their own?

---

Lol that's the dumbest ****ing thing I've ever read.

All digital information is controlled by leftists interests under express disinformation motives for various reasons.

Therefore everything ai relies on is garbage input.

That's not by accident either.

Train your own? Tools are there bra.

Even GPT-4, as coherent as it is, can feel a bit dreamlike. It can answer most of your questions, it can write silly little poems, it feels like the lights are on when you talk to it pic.twitter.com/LTEZGwJbKU

— Zero HP Lovecraft 🦅🐍 (@0x49fa98) April 3, 2023

Awesome thread about AI and what it means. It won't be a quick change, you'll slowly see jobs going away more and more over the years until it's too late for anyone to stop it.

This tech doesn't have to replace all jobs for it to be a big problem, it only has to do 1/4 to 50% to disrupt nearly everything.

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

ChatGPT can't. But robots can.

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

Why do you think all these tech and Silicon Valley guys signed the 6 month moratorium letter? They know their jobs are eroding. All those copy and paste code monkeys could be coding their own buggy whips. Along with so many STEM jobs where you're simply regurgitating computer outputs.

The next generation is going to learn how to think instead of what to think. Interesting that certain liberal arts education may survive unscathed

TexAgs91 said:

---

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

ChatGPT can't. But robots can.

---

In a production assembly line setting, yes. There are too many variables in custom installation and repair after installation scenarios.

TeslaBot is using their full-self-driving tech which is designed to operate in the real world. I know driving is different than construction, plumbing etc, but they have conquered working in the real world.

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

Imagine a robot with similarities as the one google is demonstrating here. That robot has the ability to cut framing lumber and nail it together. It's built to work on a construction site and not fall over. Now imagine that you upload the architechture plans of the house to the robot and say "frame this house". No lunch breaks, don't even have to let it go home and go to sleep. Of course - drywall, brick, etc.

dude95 said:

---

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

Imagine a robot with similarities as the one google is demonstrating here. That robot has the ability to cut framing lumber and nail it together. It's built to work on a construction site and not fall over. Now imagine that you upload the architechture plans of the house to the robot and say "frame this house". No lunch breaks, don't even have to let it go home and go to sleep. Of course - drywall, brick, etc.

---

Wouldn't you have to recharge it? Or would it work off a long extension cord?

EskimoJoe said:

---

TexAgs91 said:

---

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

ChatGPT can't. But robots can.

---

In a production assembly line setting, yes. There are too many variables in custom installation and repair after installation scenarios.

---

for now, you are correct. that will change rapidly.

Didn't you hear? Tesla has a great battery for that.

DargelSkout said:

---

dude95 said:

---

EskimoJoe said:

---

AI can't frame a house, repair broken pipes in the ground, reapir large equipment, etc...

The joke is on the "learn to code!" crowd.

---

Imagine a robot with similarities as the one google is demonstrating here. That robot has the ability to cut framing lumber and nail it together. It's built to work on a construction site and not fall over. Now imagine that you upload the architechture plans of the house to the robot and say "frame this house". No lunch breaks, don't even have to let it go home and go to sleep. Of course - drywall, brick, etc.

---

Wouldn't you have to recharge it? Or would it work off a long extension cord?

---

Yes, power is the problem. The minimum HW requirements to run Chatgpt are not feasible for autonomous robots. It's a problem that's at least an order of magnitude greater than the EV range problem.

Canada launches investigation into company behind chat gpt.

Stuff shocks me every day. BlueWillow AI text to image. I just saw this done. Whoever was doing spent 6 or seven tries refining the prompt to just what they wanted. It's on discord so you can see the prompts as they come through. This is a photo of a person that does not exist.

ultrarealistic, ultra realistic, photo, realistic, girl, teenager, She has straight, blond hair and hazel eyes. She's of average height and build. Olivia is a studious and analytical member of the group. She's always looking for rational explanations for the strange occurrences they face. She's logical, cautious, and level-headed. Olivia's analytical approach often helps the group make sense of what's happening, but it can also make her seem cold or distant., 8k

The output

GeorgiAg said:

---

What it will do is the poor saps at the big law firms are going to be expected to crank out 3 briefs a day instead of 1.

---

And the end result will be that 2 out of 3 of those saps get laid off along with most of their paralegals. West Law is probably spending a fortune trying to incorporate AI into their database lest they be displaced by someone that does.

You're a trial attorney which involves lots of moving parts with complex interfaces, but your little wills and probates shop down the street will get decimated.

I am writing books and played with chat stuff yesterday. I already have a plot for a historical fiction young adult story. In the prompt I asked if Chat could help me and I gave the background of the main character and a little of the plot line.

Chat spit out a title that was not bad and eight chapters. The ideas for the plot were not that bad. It inferred several things that kind of amazed me.

Then I asked for it to create a 1500 word first chapter and it was robotic crap. I then tried wordai which is a program and it is supposed to make it non ai recognizable and to add a more human voice. I did three revisions with that app and it made each sentence better but on the whole it sucked.

Still the thing made me realize that I have chosen the wrong side-profession, as it will not take it long to get to human level prose for fiction.

As it is I will continue using it with the outlining to see if it helps speed things up on that end.

Scary stuff

Thank you Texags for catching my attention about ChatGPT. It got me going down the rabbit hole of this tool and I downloaded the premium ChatGPT4 for $20/month.

If you haven't done that yet, do it now. This technology will change everything, everywhere, all at once (to coin a phrase). And if you aren't using it you are going to be left behind.

With just a few experimental queries I am officially freaked out because this technology will replace so many white collar knowledge workers or make them much less valuable that there will be a salary collapse like never before seen.

Example:
The current education model is dead man walking. I was a professor for decades and I taught topics that I spent decades learning and developing expertise in.

So to test the ChatGPT4 tool I entered a couple of old test questions and homework questions. My questions were based on hypothetical scenarios and required ChatGPT4 to take a specific abstract unique business concept and apply it to a completely made up business case and give a a recommended business strategy. The result was a B+ answer that with minor tweaking by the student would easily be an A answer.

I then loaded in an old homework question that again was based on a made up hypothetical and asked for a business strategy answer that drew upon what I taught in the classroom based on 20 years of experience in my job. The answer from ChatGPT4 was A+ and highlighted the most important concept that I would emphasize in the classroom. The key is that the concept was based on my experience in a specific type of negotiations, not facts that can be easily drawn from publications.

The tool was synthesizing information and presenting answers indistinguishable from student answers. It was not simply presenting facts like dates and accounting rules. It was thinking (or something indistinguishable from thinking).

Unless a student is in a testing center or writing down their answers with pencils in front of me, how can I as a professor/teacher know it is the student's work?

My step-daughter asked me for help on her high school English project. Question: Explore the themes of death and guilt in the play MacBeth. I put that question into ChatGPT4 and the answer was college level excellent. When I asked it to write the answer at a 9th grade level, ChatGPT4 dutifully spit out an excellent answer with less sophistication and different vocabulary.

My conclusion - Homework is dead. The traditional way teachers test their student's knowledge is dead. Anything that isn't literally done in front of a teacher can't be trusted.

And every student will use this tool by the end of the year at the latest.

I have no idea how schools (elementary through college) will adapt to this. But I do know they will be behind the curve no matter what they do.

Will students actually learn the material or just learn how to use a tool to get the answer? Just because you can use google translate doesn't mean you now know Spanish.

I know many on here want to blow up the education model in the US. Don't worry, ChatGPT4 and similar will do that. And faster than you can imagine.

-edited for grammar

Quote:

---

I know many on here want the blow up the education model in the US. Don't worry, ChatGPT4 and similar will do that. And faster than you can imagine.

---

Please type in the following to your for-pay ChatGPT and let me know what it tells you:

"Provide a first order crossover design between an Audax HM210C0 and a Morel MDM 55 at 1500HZ"

Thanks

Gigem_94 said:

---

bmks270 said:

---

You still have to build things and transport raw materials and energy to get things done.

Chat bot can't do that.

Chat not can help organize humans to do it though.

---

What about human slaves answering to their computer overlords that control flow of money and everything else?

---

Learn how to farm, weld, etc. barter and trade.

Win At Life said:

---

Please type in the following to your for-pay ChatGPT and let me know what it tells you:

"Provide a first order crossover design between an Audax HM210C0 and a Morel MDM 55 at 1500HZ"

Thanks

---

I have no idea what any of this means. It looks to me like a regurgitation of facts and data (similar to the way a student answers a question they really don't know the answer to) rather than what you were looking for (I guess). But if you think that just because this alpha version of AI can't answer everything with perfect answers, therefore it's not something to be concerned about, then that would just be obtuse. These tools will improve faster than we can imagine. Literally.


Provide a first order crossover design between an Audax HM210C0 and a Morel MDM 55 at 1500HZ


To design a first-order crossover for an Audax HM210C0 woofer and a Morel MDM 55 tweeter at a crossover frequency of 1500 Hz, you'll need to calculate the values for the capacitors and inductors required for the crossover network.
A first-order crossover has a 6dB per octave roll-off rate. For the woofer (low-pass filter), you'll use an inductor (L), and for the tweeter (high-pass filter), you'll use a capacitor (C).
Here's the formula for the low-pass filter (woofer):
L = (1 / (2 * * Fc * R_L))
Here's the formula for the high-pass filter (tweeter):
C = (1 / (2 * * Fc * R_H))
In both cases, Fc is the crossover frequency, which is 1500 Hz. R_L and R_H are the resistances (impedances) of the woofer and tweeter, respectively. The Audax HM210C0 has a nominal impedance of 8 ohms, and the Morel MDM 55 has a nominal impedance of 6 ohms.
Now let's calculate the values:
[ol]

Low-pass filter (woofer) inductor value:
L = (1 / (2 * * 1500 * 8))
L 0.0133 H (13.3 mH)

High-pass filter (tweeter) capacitor value:
C = (1 / (2 * * 1500 * 6))
C 0.0000177 F (17.7 F)

[/ol]Based on these calculations, you should use a 13.3 mH inductor for the low-pass filter (Audax HM210C0) and a 17.7 F capacitor for the high-pass filter (Morel MDM 55) to create a first-order crossover at 1500 Hz.
Please note that these calculations are a starting point and may not result in the best possible performance. You might need to experiment with different component values and take into account the individual characteristics of the drivers, as well as your listening environment and preferences.

Win At Life said:

---

Please type in the following to your for-pay ChatGPT and let me know what it tells you:

"Provide a first order crossover design between an Audax HM210C0 and a Morel MDM 55 at 1500HZ"

Thanks

---

ChatGPT4.0 said:

---

A first order crossover design is quite simple, as it involves only one capacitor for the high-pass filter and one inductor for the low-pass filter. In this case, we are creating a crossover network between an Audax HM210C0 midrange driver and a Morel MDM 55 tweeter at a frequency of 1500 Hz.

The formula to calculate the capacitor value for the high-pass filter is:

C = 1 / (2 * * f * R)

Where:
C is the capacitor value in farads
f is the crossover frequency (1500 Hz)
R is the impedance of the tweeter (Morel MDM 55 has 6 ohms impedance)
is the constant pi (approximately 3.14159)

The formula to calculate the inductor value for the low-pass filter is:

L = R / (2 * * f)

Where:
L is the inductor value in henrys
f is the crossover frequency (1500 Hz)
R is the impedance of the midrange driver (Audax HM210C0 has 8 ohms impedance)
is the constant pi (approximately 3.14159)

Plugging the values into the formulas, we get:

C = 1 / (2 * * 1500 * 6)
C 17.67 F

L = 8 / (2 * * 1500)
L 0.85 mH

So for this first order crossover design, you'll need a 17.67 F capacitor for the high-pass filter connected to the Morel MDM 55 tweeter and a 0.85 mH inductor for the low-pass filter connected to the Audax HM210C0 midrange driver. Note that these values may need to be adjusted depending on the specific components and their tolerances. Additionally, you may need to experiment with the actual values to achieve the desired sound quality, as this is a simple calculation and doesn't take into account other factors like driver interaction or baffle step compensation.

---

Interesting that identical queries provided similar but differs responses.

Hogties said:

---

Thank you Texags for catching my attention about ChatGPT. It got me going down the rabbit hole of this tool and I downloaded the premium ChatGPT4 for $20/month.

If you haven't done that yet, do it now. This technology will change everything, everywhere, all at once (to coin a phrase). And if you aren't using it you are going to be left behind.

With just a few experimental queries I am officially freaked out because this technology will replace so many white collar knowledge workers or make them much less valuable that there will be a salary collapse like never before seen.

Example:
The current education model is dead man walking. I was a professor for decades and I taught topics that I spent decades learning and developing expertise in.

So to test the ChatGPT4 tool I entered a couple of old test questions and homework questions. My questions were based on hypothetical scenarios and required ChatGPT4 to take a specific abstract unique business concept and apply it to a completely made up business case and give a a recommended business strategy. The result was a B+ answer that with minor tweaking by the student would easily be an A answer.

I then loaded in an old homework question that again was based on a made up hypothetical and asked for a business strategy answer that drew upon what I taught in the classroom based on 20 years of experience in my job. The answer from ChatGPT4 was A+ and highlighted the most important concept that I would emphasize in the classroom. The key is that the concept was based on my experience in a specific type of negotiations, not facts that can be easily drawn from publications.

The tool was synthesizing information and presenting answers indistinguishable from student answers. It was not simply presenting facts like dates and accounting rules. It was thinking (or something indistinguishable from thinking).

Unless a student is in a testing center or writing down their answers with pencils in front of me, how can I as a professor/teacher know it is the student's work?

My step-daughter asked me for help on her high school English project. Question: Explore the themes of death and guilt in the play MacBeth. I put that question into ChatGPT4 and the answer was college level excellent. When I asked it to write the answer at a 9th grade level, ChatGPT4 dutifully spit out an excellent answer with less sophistication and different vocabulary.

My conclusion - Homework is dead. The traditional way teachers test their student's knowledge is dead. Anything that isn't literally done in front of a teacher can't be trusted.

And every student will use this tool by the end of the year at the latest.

I have no idea how schools (elementary through college) will adapt to this. But I do know they will be behind the curve no matter what they do.

Will students actually learn the material or just learn how to use a tool to get the answer? Just because you can use google translate doesn't mean you now know Spanish.

I know many on here want to blow up the education model in the US. Don't worry, ChatGPT4 and similar will do that. And faster than you can imagine.

-edited for grammar

---

Got it for homework, but there are some subjects which a student can't fake their way through a test. Try having ChatGPT do your calculus homework for a semester and then take a mid-term and final each worth 25% of the grade. I don't think things would go too well. I'm drawing on a personal experience from 40 years ago when I had a ChemE major roommate that was a math wiz. Sure I could pay him to do my homework, but I had to take the test.

From my limited use of the product, I believe it has some real issues. Because it uses public data it cannot always tell fact from conjecture. It also uses the blogosphere to form(steal) much of its content.

When I asked it to write a report on data privacy laws for ten different locations where a business I consult with has facilities, it got half of the information wrong but wrote it as fact. Much of what it found was not passed law but proposed.

As far as the homework questions you can find most of those somewhere using google if you search hard enough. This just makes it easier. Again if it can verify the correct answer it will give you good data but if it finds conflicting answers it will make a best guess.

I really think that the amount of inaccuracies out on the web is about to go up exponentially as people don't fact check these things.

As far a writing prose, I think it can come up with plots based off of other plots and even form novel plots by sequencing events/twists in new ways after it maps out the old ones. The problem it will have is emotion. It is like teaching a person with autism to read faces and mimic emotions. Not sure how it will make that jump.

I think people need to understand it and use it like a personal assistant but an assistant that needs their work double checked.

Couple of points

• Looking things up via Google should net you similar numbers of inaccuracies. This might be highlighted more now - but will getting better
• Don't know what i think about school. We learn long division in the 4th grade, but who actually pulls out a pencil for it in the real world any more? When my kids are in their 30s, I doubt it's a skill they will have (oh yeah, we did this in the 4th grade). Same goes for this - how will the very act of learning change because of it? What needs to be learned, what needs to be practiced regularly? What new skills do we need to adapt to.
• GPT5 is in test now - rumour is end of this year or 1Q next. It will go from prompts that can be several pages to "taking the whole book of Dune in a single prompt". I reportedly will have much more ability to keep up with a conversation, do better with real time information, respond with even longer answers.

Well all the published data for the Morel MDM 55 lists it at a nominal 8 ohms and the published curves show it higher than that and rising in the crossover range. I'm not sure where both came up with 6 ohms on that. There exists free software out there to improve the design, but it seems it doesn't have access to use them. Also, I'm unaware of any published data on the driver phase angles, so you'd have to estimate a curve-it to improve the design. You would also need to account for the baffle gain, which it didn't have any information to do. It just says you may need to do some more work, which is true. Generally speaking, without accounting for any real-world conditions, using the canned formulas it does will create a huge null right at the crossover point you're trying to design.

i havent been into car audio systems in a long time. is the JL audio W7 still the king?

Stressboy said:

---

I am writing books and played with chat stuff yesterday. I already have a plot for a historical fiction young adult story. In the prompt I asked if Chat could help me and I gave the background of the main character and a little of the plot line.

Chat spit out a title that was not bad and eight chapters. The ideas for the plot were not that bad. It inferred several things that kind of amazed me.

Then I asked for it to create a 1500 word first chapter and it was robotic crap. I then tried wordai which is a program and it is supposed to make it non ai recognizable and to add a more human voice. I did three revisions with that app and it made each sentence better but on the whole it sucked.

Still the thing made me realize that I have chosen the wrong side-profession, as it will not take it long to get to human level prose for fiction.

As it is I will continue using it with the outlining to see if it helps speed things up on that end.

Scary stuff

---

Interesting. I started a book a few years ago and never finished. I fed it the first three chapters and asked it to write the next chapter.

Yikes it was bad.

Can you load your own datbase of documents in it and have it locate and sort and spit out summaries, combinations, or common citations?

Justin2010 said:

---

Stressboy said:

---

I am writing books and played with chat stuff yesterday. I already have a plot for a historical fiction young adult story. In the prompt I asked if Chat could help me and I gave the background of the main character and a little of the plot line.

Chat spit out a title that was not bad and eight chapters. The ideas for the plot were not that bad. It inferred several things that kind of amazed me.

Then I asked for it to create a 1500 word first chapter and it was robotic crap. I then tried wordai which is a program and it is supposed to make it non ai recognizable and to add a more human voice. I did three revisions with that app and it made each sentence better but on the whole it sucked.

Still the thing made me realize that I have chosen the wrong side-profession, as it will not take it long to get to human level prose for fiction.

As it is I will continue using it with the outlining to see if it helps speed things up on that end.

Scary stuff

---

Interesting. I started a book a few years ago and never finished. I fed it the first three chapters and asked it to write the next chapter.

Yikes it was bad.

---

I asked it to write a brief note to my wife on our anniversary. It said all the things I couldn't express because I'm not a lib arts major. Copy and paste and my wife loved it. She is a smart woman and figured it out though. But I still meant what I (we) said.

I also use chatGPT to play Zork! It is amazing at it. The only boundary is the question limit.