The AI, chatGPT, learns much faster about MMT than mainstream economists. Or the general public. - Politics Forum.org | PoFo

Wandering the information superhighway, he came upon the last refuge of civilization, PoFo, the only forum on the internet ...

"It's the economy, stupid!"

Moderator: PoFo Economics & Capitalism Mods

Forum rules: No one line posts please.
#15261150
A comment and Bill's reply on Bill's blog.

andrew
Tuesday, January 3, 2023 at 17:38
I was excited to test out the new genius AI on the block. chatGPT with regard to money creation. Unfortunately while you can get it to admit that money is created when currency issuing governments take the decision to spend and that tax payments must be paid in government money and that this money is essentially destroyed when taxes are paid, it still insists that governments need taxes to finance their operations. Oh Dear !

Wednesday, January 4, 2023 at 8:56
Dear Andrew (at 2023/01/03 at 5:38 pm)

I have experimented with chatGPT and found that if it initially delivers less than accurate information you can submit a better answer and over time it will adapt to your input and eliminate the falsehoods.

In other words, unlike many humans, mainstream economists included, it learns rather quickly from its mistakes.

Give it a try. I was really happy with the adaptive learning process.

best wishes
bill
#15261154
Unthinking Majority wrote:Cant wait to try out Skynet then get murdered by it in 20 years.


What is your point?

I fear AI, but I have less than zero say in the matter.

You reply didn't refute the point that the AI is looking at the data and peer reviewed papers and learning the truth, as it sees it, without any bias (toward the wishes of the rich like most economists have).

.
#15261158
Steve_American wrote:What is your point?

I fear AI, but I have less than zero say in the matter.

You reply didn't refute the point that the AI is looking at the data and peer reviewed papers and learning the truth, as it sees it, without any bias (toward the wishes of the rich like most economists have).

.


I'll admit it's really cool tech. But there's also a danger in the general public feeding it all this data so it can learn. Not now, but when it gets smarter by orders of magnitude, and it will in time. It will be impossible to regulate AI on a global scale. AI is a massive threat to humanity. It could also make us be able to live forever. It is the most powerful technology ever made and it's not even close. Splitting the atom is a joke compared to AI.
#15261173
Unthinking Majority wrote:Cant wait to try out Skynet then get murdered by it in 20 years.


@Steve_American

You misunderstand "AI". Statistical analysis is not AI and all of our algorithms IS statistical analysis. You don't get AI with just statistics.

Also learning and being correct is not the same thing. If you think that our modern "AI" does no mistakes then well I got news for you.
#15261295
JohnRawls wrote:@Steve_American

You misunderstand "AI". Statistical analysis is not AI and all of our algorithms IS statistical analysis. You don't get AI with just statistics.

Also learning and being correct is not the same thing. If you think that our modern "AI" does no mistakes then well I got news for you.


John, did you fail reading comprehension?
My OP included this quote, "while you can get it to admit that money is created when currency issuing governments take the decision to spend and that tax payments must be paid in government money and that this money is essentially destroyed when taxes are paid,"

I don't see that any of those statements are statistical in nature. They are either facts about how all modern nations use their fiat currency or they are false.

They are all statements that MMTers say are true. They follow from the fact that the world is off the gold standard. BTW, money is also created by bank making loans. Banks don't lend the savings of their customers. This is actually against the law, AFAIK.

.
Last edited by Steve_American on 06 Jan 2023 12:28, edited 1 time in total.
#15261296
Steve_American wrote:
John, did you fail reading comprehension?
My OP included this quote, "while you can get it to admit that money is created when currency issuing governments take the decision to spend and that tax payments must be paid in government money and that this money is essentially destroyed when taxes are paid,"

I don't see that any of those statements are statistical in nature. They are either facts about all modern nations use their fiat currency or they are false.

They are all statements that MMTers say are true. They follow from the fact that the world is off the gold standard. BTW, money is also created by bank making loans. Banks don't lend the savings of their customers. This is actually against the law, AFAIK.



It's not AI, not even close.

It's a smart program, but they depend heavily on their database and development.

For example, there is a smart program for doctors. It does a better job at finding illness than Docs. Which offends Docs, so they don't use it. It took years to build the database and make it work.

But it's not in the slightest degree intelligent.

This new one is excellent at mimicry. Birds can do that, but they are also not intelligent.

We will get closer to AI, but like fusion, it will remain elusive for most or all of this century.
#15261298
late wrote:It's not AI, not even close.

It's a smart program, but they depend heavily on their database and development.

For example, there is a smart program for doctors. It does a better job at finding illness than Docs. Which offends Docs, so they don't use it. It took years to build the database and make it work.

But it's not in the slightest degree intelligent.

This new one is excellent at mimicry. Birds can do that, but they are also not intelligent.

We will get closer to AI, but like fusion, it will remain elusive for most or all of this century.


It doesn't matter if current AIs don't meet your criteria to be intelligent.
You said one AI is better than Docs at "finding illness". That is good enough for me.
They do learn AFAIK. That is how they work. They slowly improve their program.
For example, humans could never (IIRC) program a 2-legged robot to walk. The other day I saw a YouTube video of a 2-legged robot walking jumping, dancing, etc. It was very agile. The claim was that an AI had taught itself to do all that.
So, they learn. How they learn economics is open to question, because the dancing robot got instant feedback as it tried to move around. I don't see that that is a possible way to learn economics. The AI would have to use a different method.

.
#15261305
Steve_American wrote:John, did you fail reading comprehension?
My OP included this quote, "while you can get it to admit that money is created when currency issuing governments take the decision to spend and that tax payments must be paid in government money and that this money is essentially destroyed when taxes are paid,"

I don't see that any of those statements are statistical in nature. They are either facts about how all modern nations use their fiat currency or they are false.

They are all statements that MMTers say are true. They follow from the fact that the world is off the gold standard. BTW, money is also created by bank making loans. Banks don't lend the savings of their customers. This is actually against the law, AFAIK.

.


If i teach a model and the text that I input in to it often repeats that "Jesus is Hitler" then the "AI" will believe that Jesus is Hitler. So when you write to them "Who is Jesus" then the "AI" will answer that Jesus is Hitler.
#15261328
Steve_American wrote:
That is good enough for me.

The AI would have to use a different method.



You are aware that the AI has to be different to be able to do economics; but you seem to be letting your emotions cloud you thinking.

There is no real AI yet, and a mastery of economics is still a ways off.
#15261333
late wrote:
We will get closer to AI, but like fusion, it will remain elusive for most or all of this century.



Steve, do you currently have any *models* of AI applied to MMT -- ?

Presumably the AI would be used to 'balance-out' all of the various components of the political economy, for your desired MMT economic 'homeostasis', but I'll argue that capitalism's inherent economic exploitation should *not* be countenanced and retained -- proletarian *revolution* is required.


Social Production Worldview

Spoiler: show
Image



labor and capital, side-by-side

Spoiler: show
Image



material-economic exploitation

Spoiler: show
Image


Spoiler: show
Image



---


And, here's a present-day application of AI that I use in the process of rendering 3D graphics -- a 'shortcut' by using noise-reduction:



Intel Open Image Denoise internally builds on top of Intel oneAPI Deep Neural Network Library (oneDNN), and automatically exploits modern instruction sets like Intel SSE4, AVX2, and AVX-512 to achieve high denoising performance. A CPU with support for at least SSE4.1 or Apple Silicon is required to run Intel Open Image Denoise.



https://www.openimagedenoise.org/
#15262665
Steve_American wrote:You reply didn't refute the point that the AI is looking at the data and peer reviewed papers and learning the truth, as it sees it, without any bias (toward the wishes of the rich like most economists have).

You think peer-reviewed papers in economics journals don't serve the wishes of the rich??
#15262703
Truth To Power wrote:You think peer-reviewed papers in economics journals don't serve the wishes of the rich??


TTP, of course most of them do. However, the ones by MMTers don't do that, and mainstream papers are mostly flawed in that they are based in deductive logic and I would hope that an AI would reject all conclusions based on false assumptions. OTOH, it seems like most people don't care about that problem with MS Econ.

.

The October 7 attack may constitute an act of atte[…]

Russia-Ukraine War 2022

will putin´s closest buddy Gennady Timchenko be […]

https://youtu.be/URGhMw1u7MM?si=YzcCHXcH9e-US9mv […]

Xi Jinping: "vladimir, bend down even lower, […]