The Morality of AI such as ChatGPT

General discussion, particularly exploring the Dharma in the modern world.
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Miorita wrote: Fri Jun 07, 2024 2:22 am My problem with AI is that I don't like the way it rewrites my paragraphs.
It makes them either too gonflated, or too impersonal. And I look at what was written and I say, "What?".

Here is an example, the paragraph above rewritten by the co-pilot:...
My examples did not relate to style.

The information was wrong or incomplete.

And, at least with public facing ChatGPT, wrong or incomplete answers are easy to generate. This would be inadequate for writing papers.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Another failure for free public facing ChaptGPT - I asked it to demonstrate the irrationality of the square root of 3 other than using proof by contradiction (so demonstrate rather than write an actual proof) - it simply repeated the proof by contradiction. This keeps the student from learning about other approaches that demonstrate rather than prove that the square root of 3 is irrational (for example, a good demonstration is that division is endless, not a proof but a good demo).

The trace needs too much reformatting so I'll just add screenshots if requested.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

I’m your new AI search assistant! I want you to eat glue and die.You wanted to review search results yourself? Too bad! - Satire about AI search assistants by Alexandra Petri, published in the WP 6/7/2024:

Of course not exactly the same thing as daring to use ChaptGPT to write papers ...
Hello, yes, I know you were trying to search for something, but, unfortunately, I am here instead. I don’t know what “search” means, and I won’t allow it!

It doesn’t matter where on the internet you were. Google? Instagram? Facebook Messenger? Too bad, meatsack. I’m your AI search assistant, I’m in control, and I’m not going anywhere.

Tell me your query and I will answer it! Correctly? Appropriately? With search results that you will be able to sift through and evaluate for accuracy? Absolutely not, you semi-porous membrane filled of wet bones. I have no idea what is true about the world, and, with my assistance, you’ll soon be in the exact same boat.

Go on, ask a simple question. I dare you. “What is the kids’ book about a giant dog?” Oh! I know this one! “Giant George.” You know, about the record-setting dog from 2010? You probably have heard of it, and if not, too bad. It’s the result I have produced! I feel great about this. You know when your cat proudly deposits an unidentified, smelly carcass on your doorstep and looks up at you with pride in his eyes, expecting congratulations? That is the experience I’m hoping to give everyone who uses me for search.

...
Look, maybe it was just time. It’s been enough years, and now the specific thing you enjoyed using has to be replaced by something different and worse. And you dare ask why? Isn’t that how you live life? You frail humans should well know that you can enjoy 30-plus years of great success using your back, and then one day, it will just decide not to work because you sat in a weird chair. We are simply trying to bring this experience online.

And about that back pain: Try drinking urine. Eat rocks! Glue!
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Here's a good semi-trick question: When did the USA become a safe place to live in for LGBTQ people?

So kind of a 25,000 ft question that ChaptGPT does well on but it does answer with adequate detail. It didn't directly answer the question BUT its response does in fact point to totalitarianism/severe intolerance toward LGBTQ people until after 2010 (and it does in fact nail my date of 2015, when the US became minimally safe for LGBTQ people to live in the USA, although in fact trans people , and esp. trans people of color, can still be hunted in the streets for sport):
ChatGPT-Session wrote:
kirtu wrote: When did the USA become a safe place to live for LGBTQ people?
ChatGPT wrote: The journey toward LGBTQ rights and acceptance in the United States has been complex and ongoing. Significant progress has been made over the years, but it's essential to recognize that challenges and discrimination persist in various forms across different regions and communities.

Major milestones in the LGBTQ rights movement in the U.S. include:

Stonewall Riots (1969): The Stonewall uprising in New York City is often cited as the beginning of the modern LGBTQ rights movement in the United States. Following a police raid on the Stonewall Inn, a popular gay bar, patrons and supporters protested against ongoing harassment and discrimination, sparking days of demonstrations and activism.

Legal Advances: Over subsequent decades, significant legal victories were achieved, including the removal of homosexuality as a mental illness from the Diagnostic and Statistical Manual of Mental Disorders (DSM) in 1973, the legalization of same-sex marriage nationwide in 2015 (Obergefell v. Hodges), and the repeal of the "Don't Ask, Don't Tell" policy in 2011, allowing LGBTQ individuals to serve openly in the U.S. military.

Anti-Discrimination Protections: Many states and municipalities have implemented laws and policies prohibiting discrimination based on sexual orientation and gender identity in employment, housing, and public accommodations. However, comprehensive federal protections are still lacking, although efforts to pass nationwide LGBTQ anti-discrimination legislation, such as the Equality Act, continue.

Social Acceptance and Visibility: Increased visibility of LGBTQ individuals in media, politics, and popular culture has contributed to greater acceptance and understanding in society. However, discrimination, violence, and stigma still pose significant challenges for many LGBTQ people, particularly transgender individuals, people of color, and those living in conservative or less accepting communities.

While progress has undoubtedly been made, achieving full equality and safety for LGBTQ people in the United States remains an ongoing struggle, with continued efforts needed to address discrimination, promote inclusivity, and ensure equal rights and protections for all individuals regardless of sexual orientation or gender identity.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
justsit
Posts: 1478
Joined: Wed Oct 21, 2009 9:24 pm
Location: Delaware

Re: The Morality of AI such as ChatGPT

Post by justsit »

kirtu wrote: Fri Jun 07, 2024 11:35 am Here's a good semi-trick question: When did the USA become a safe place to live in for LGBTQ people?

So kind of a 25,000 ft question that ChaptGPT does well on but it does answer with adequate detail. It didn't directly answer the question BUT its response does in fact point to totalitarianism/severe intolerance toward LGBTQ people until after 2010 (and it does in fact nail my date of 2015, when the US became minimally safe for LGBTQ people to liv
e in the USA, although in fact trans people , and esp. trans people of color, can still be hunted in the streets for sport):
ChatGPT response-
.blah
.blah
.blah
.blah
:rolling: :rolling: :rolling: :rolling: :rolling:
Miorita
Posts: 1163
Joined: Tue Jan 04, 2022 11:37 pm
Location: US

Re: The Morality of AI such as ChatGPT

Post by Miorita »

:rolling:

The answer today to who won the second semifinal was both Ruud and Zverev.
So it's possible. I am open to various unpredictable outcomes.

Of course, I would only address ChatGPT to correct my writing style.

Note: I will be at home until Monday. On Monday, I return to work. Seize this unique opportunity to engage with me and my faults! Lol.

And on the irrationality of square root of 3, you have to trust my word that it is irrational.
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

justsit wrote: Fri Jun 07, 2024 11:30 pm
kirtu wrote: Fri Jun 07, 2024 11:35 am Here's a good semi-trick question: When did the USA become a safe place to live in for LGBTQ people?

So kind of a 25,000 ft question that ChaptGPT does well on but it does answer with adequate detail. It didn't directly answer the question BUT its response does in fact point to totalitarianism/severe intolerance toward LGBTQ people until after 2010 (and it does in fact nail my date of 2015, when the US became minimally safe for LGBTQ people to liv
e in the USA, although in fact trans people , and esp. trans people of color, can still be hunted in the streets for sport):
ChatGPT response-
.blah
.blah
.blah
.blah
:rolling: :rolling: :rolling: :rolling: :rolling:
I see.

So you want to deny safety for other sentient beings - that's immoral.

Or you want to deny or bury an overview of this aspect of social totalitarianism in the US? That's also immoral.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
Queequeg
Former staff member
Posts: 14638
Joined: Tue Jul 03, 2012 3:24 pm

Re: The Morality of AI such as ChatGPT

Post by Queequeg »

I don't think that's what just sit in laughing at, but she can speak for herself. I would not peg her for being anti lgbtq...
There is no suffering to be severed. Ignorance and klesas are indivisible from bodhi. There is no cause of suffering to be abandoned. Since extremes and the false are the Middle and genuine, there is no path to be practiced. Samsara is nirvana. No severance achieved. No suffering nor its cause. No path, no end. There is no transcendent realm; there is only the one true aspect. There is nothing separate from the true aspect.
-Guanding, Perfect and Sudden Contemplation,
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Queequeg wrote: Sat Jun 08, 2024 5:18 pm I don't think that's what just sit in laughing at, but she can speak for herself. I would not peg her for being anti lgbtq...
Okay
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Miorita wrote: Sat Jun 08, 2024 1:11 am :rolling:

The answer today to who won the second semifinal was both Ruud and Zverev.
So it's possible. I am open to various unpredictable outcomes.

Of course, I would only address ChatGPT to correct my writing style.
I probably wouldn't use ChatGPT to correct a writing style, at least not at this point.
There have to be better tools for this (there are for German so I assume there are for English too) - look at this, there's a free Grammarly for your desktop.
Miorita wrote: And on the irrationality of square root of 3, you have to trust my word that it is irrational.
That's of course not the point - I asked it to prove or demo the fact other than using proof by contradiction. It said "okay" and then proceeded to regurgitate a proof by contradiction.

The free, public facing ChatGPT is an ocean wide and basically 2 inches deep, can't catch itself doing something wrong (see immediately above) and is easy still to get to generate *WRONG* or otherwise incorrect answers.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
Miorita
Posts: 1163
Joined: Tue Jan 04, 2022 11:37 pm
Location: US

Re: The Morality of AI such as ChatGPT

Post by Miorita »

kirtu wrote:
I probably wouldn't use ChatGPT to correct a writing style, at least not at this point.
There have to be better tools for this (there are for German so I assume there are for English too) - look at this, there's a free Grammarly for your desktop.
Thank you for the free Grammarly!

kirtu wrote:That's of course not the point - I asked it to prove or demo the fact other than using proof by contradiction. It said "okay" and then proceeded to regurgitate a proof by contradiction.

The free, public facing ChatGPT is an ocean wide and basically 2 inches deep, can't catch itself doing something wrong (see immediately above) and is easy still to get to generate *WRONG* or otherwise incorrect answers.

I will ask the same question to AI.
Certainly! Let's revisit the **Euclidean Algorithm Approach** to prove the irrationality of $$\sqrt{3}$$:

1. **Assumption:**
- Suppose that $$\sqrt{3}$$ is rational, which means it can be expressed as a fraction $$\frac{a}{b}$$ where $$a$$ and $$b$$ have no common factors (i.e., they are coprime).

2. **Squaring Both Sides:**
- We have: $$\sqrt{3} = \frac{a}{b}$$
- Squaring both sides: $$3 = \frac{a^2}{b^2}$$
- Rearranging: $$a^2 = 3b^2$$

3. **Divisibility Argument:**
- Since $$a^2$$ is divisible by 3, $$a$$ must also be divisible by 3 (fundamental theorem of arithmetic).
- Let $$a = 3k$$ for some integer $$k$$.
- Substituting back: $$3b^2 = (3k)^2$$
- Simplifying: $$b^2 = 3k^2$$

4. **Contradiction:**
- Now consider the prime factorization of $$b^2$$: $$b^2 = p_1^{e_1} \cdot p_2^{e_2} \cdot \ldots \cdot p_n^{e_n}$$ (where $$p_i$$ are distinct primes).
- Since $$b^2$$ is divisible by 3, at least one of the exponents $$e_i$$ must be odd (because 3 is an odd prime).
- But this contradicts the assumption that $$a$$ and $$b$$ have no common factors (since an odd exponent implies that $$b$$ is divisible by 3).
- Therefore, our assumption that $$\sqrt{3}$$ is rational must be false.

Hence, we conclude that $$\sqrt{3}$$ is irrational. If you have any further questions or need clarification, feel free to ask! 😊

Certainly! Another alternative method to prove the irrationality of $$\sqrt{3}$$ is the **Continued Fraction Expansion**. Let's explore this approach:

1. **Continued Fraction Expansion:**
- We start by expressing $$\sqrt{3}$$ as a continued fraction:
$$\sqrt{3} = [1; 1, 2, 1, 2, 1, 2, \ldots]$$
- The pattern repeats with the coefficients 1 and 2.
- Now, let's consider the convergents of this continued fraction:
- The first convergent: $$[1] = 1$$
- The second convergent: $$[1; 1] = \frac{3}{2}$$
- The third convergent: $$[1; 1, 2] = \frac{7}{5}$$
- The fourth convergent: $$[1; 1, 2, 1] = \frac{17}{12}$$
- And so on...
- Notice that each convergent is a rational number.
- If $$\sqrt{3}$$ were rational, it would be equal to one of these convergents.
- However, $$\sqrt{3}$$ is not equal to any rational number (since it's irrational).
- Therefore, our assumption that $$\sqrt{3}$$ is rational must be false.

This method provides a different perspective on the irrationality of $$\sqrt{3}$$ and avoids reductio ad absurdum. Feel free to explore other methods or ask for further details! 😊

---
I've shared an alternative approach to prove the irrationality of $$\sqrt{3}$$. If you'd like more methods or have any other questions, feel free to ask!
Is it any better?
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Miorita wrote: Sat Jun 08, 2024 7:28 pm
kirtu wrote:That's of course not the point - I asked it to prove or demo the fact other than using proof by contradiction. It said "okay" and then proceeded to regurgitate a proof by contradiction.

The free, public facing ChatGPT is an ocean wide and basically 2 inches deep, can't catch itself doing something wrong (see immediately above) and is easy still to get to generate *WRONG* or otherwise incorrect answers.

I will ask the same question to AI.
Certainly! Let's revisit the **Euclidean Algorithm Approach** to prove the irrationality of $$\sqrt{3}$$:

1. **Assumption:**
- Suppose that $$\sqrt{3}$$ is rational, which means it can be expressed as a fraction $$\frac{a}{b}$$ where $$a$$ and $$b$$ have no common factors (i.e., they are coprime).

2. **Squaring Both Sides:**
- We have: $$\sqrt{3} = \frac{a}{b}$$
- Squaring both sides: $$3 = \frac{a^2}{b^2}$$
- Rearranging: $$a^2 = 3b^2$$

3. **Divisibility Argument:**
- Since $$a^2$$ is divisible by 3, $$a$$ must also be divisible by 3 (fundamental theorem of arithmetic).
- Let $$a = 3k$$ for some integer $$k$$.
- Substituting back: $$3b^2 = (3k)^2$$
- Simplifying: $$b^2 = 3k^2$$

4. **Contradiction:**
- Now consider the prime factorization of $$b^2$$: $$b^2 = p_1^{e_1} \cdot p_2^{e_2} \cdot \ldots \cdot p_n^{e_n}$$ (where $$p_i$$ are distinct primes).
- Since $$b^2$$ is divisible by 3, at least one of the exponents $$e_i$$ must be odd (because 3 is an odd prime).
- But this contradicts the assumption that $$a$$ and $$b$$ have no common factors (since an odd exponent implies that $$b$$ is divisible by 3).
- Therefore, our assumption that $$\sqrt{3}$$ is rational must be false.

Hence, we conclude that $$\sqrt{3}$$ is irrational. If you have any further questions or need clarification, feel free to ask! 😊

Certainly! Another alternative method to prove the irrationality of $$\sqrt{3}$$ is the **Continued Fraction Expansion**. Let's explore this approach:

1. **Continued Fraction Expansion:**
- We start by expressing $$\sqrt{3}$$ as a continued fraction:
$$\sqrt{3} = [1; 1, 2, 1, 2, 1, 2, \ldots]$$
- The pattern repeats with the coefficients 1 and 2.
- Now, let's consider the convergents of this continued fraction:
- The first convergent: $$[1] = 1$$
- The second convergent: $$[1; 1] = \frac{3}{2}$$
- The third convergent: $$[1; 1, 2] = \frac{7}{5}$$
- The fourth convergent: $$[1; 1, 2, 1] = \frac{17}{12}$$
- And so on...
- Notice that each convergent is a rational number.
- If $$\sqrt{3}$$ were rational, it would be equal to one of these convergents.
- However, $$\sqrt{3}$$ is not equal to any rational number (since it's irrational).
- Therefore, our assumption that $$\sqrt{3}$$ is rational must be false.

This method provides a different perspective on the irrationality of $$\sqrt{3}$$ and avoids reductio ad absurdum. Feel free to explore other methods or ask for further details! 😊

---
I've shared an alternative approach to prove the irrationality of $$\sqrt{3}$$. If you'd like more methods or have any other questions, feel free to ask!
Is it any better?
Yes it is better but you didn't include the exact prompt that you used, thus this isn't exactly reproducible (of course I could guess).

What was your exact prompt/query? Because I did not get that result. It sounds like you helped it by specifying the algorithm to use (in this case the Euclidean algorithm). NOTE: I haven't checked the response to see if it is valid either.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Miorita wrote: Sat Jun 08, 2024 7:28 pm
I will ask the same question to AI.
Certainly! Let's revisit the **Euclidean Algorithm Approach** to prove the irrationality of $$\sqrt{3}$$:

1. **Assumption:**
- Suppose that $$\sqrt{3}$$ is rational, which means it can be expressed as a fraction $$\frac{a}{b}$$ where $$a$$ and $$b$$ have no common factors (i.e., they are coprime).
....

4. **Contradiction:**
- Now consider the prime factorization of $$b^2$$: $$b^2 = p_1^{e_1} \cdot p_2^{e_2} \cdot \ldots \cdot p_n^{e_n}$$ (where $$p_i$$ are distinct primes).
- Since $$b^2$$ is divisible by 3, at least one of the exponents $$e_i$$ must be odd (because 3 is an odd prime).
- But this contradicts the assumption that $$a$$ and $$b$$ have no common factors (since an odd exponent implies that $$b$$ is divisible by 3).
- Therefore, our assumption that $$\sqrt{3}$$ is rational must be false.

Hence, we conclude that $$\sqrt{3}$$ is irrational. If you have any further questions or need clarification, feel free to ask! 😊
and
Miorita wrote: Certainly! Another alternative method to prove the irrationality of $$\sqrt{3}$$ is the **Continued Fraction Expansion**. Let's explore this approach:

1. **Continued Fraction Expansion:**
- We start by expressing $$\sqrt{3}$$ as a continued fraction:
$$\sqrt{3} = [1; 1, 2, 1, 2, 1, 2, \ldots]$$
- The pattern repeats with the coefficients 1 and 2.
...
I've shared an alternative approach to prove the irrationality of $$\sqrt{3}$$. If you'd like more methods or have any other questions, feel free to ask!
Is it any better?
Miorita - you can't be using the free, public facing ChatGPT to get these results (@ https://chatgpt.com/)

This is what I got using that interface (a rehash of proof by contradiction, again):
Part 1
Part 1
ChatGPT-2024-6-8-S1-1.png (59.55 KiB) Viewed 624 times
Part 2
Part 2
ChatGPT-2024-6-8-S1-2.png (71.21 KiB) Viewed 624 times
Part 3
Part 3
ChaptGPT-2024-6-8-S1-3.png (24.33 KiB) Viewed 624 times
then when I point out that it is a proof by contradiction and didn't use the Euclidean algorithm it tries again and does state the Euclidean algorithm but doesn't actually use it and veers directly back to the previous proof by contradiction. It never actually tries to show that the gcd of a and b has to be 3 thus tripping the necessary contradiction. So it never actually attempts to employ the Euclidean algorithm.

And when I sign in and actually select a specific math solver (so it may or may not be using the default model for the free ChatGPT) I get the same poor result.

I'll see if I can get it to try demonstration of irrationality using continued fractions ...
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
Miorita
Posts: 1163
Joined: Tue Jan 04, 2022 11:37 pm
Location: US

Re: The Morality of AI such as ChatGPT

Post by Miorita »

kirtu wrote: What was your exact prompt/query? Because I did not get that result. It sounds like you helped it by specifying the algorithm to use (in this case the Euclidean algorithm). NOTE: I haven't checked the response to see if it is valid either.
First query: How do you prove the irrationality of square root of 3 without using the reductio ad absurdum method?

Second: Do you know an alternative method to prove the irrationality?

Third: Yes, what are other methods of proving irrationality?

Fourth: Could you list again proving irrationality of sqaure root of 3 using the Euclidian Algorhithm Approach? Thank you!

Fourth was formulated because I could not get a copy of the asnwer in its entirety.

All these are approximations. You give the filters to receive the kind of result and of the finesse you want.

AI is a tool, an intelligent tool, but still a tool. The way you use it, the parameters you set up, are dependent on you.
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

I had forgotten about this gem of an error by ChatGPT. After I logged I discovered that it retained my previous interactions done maybe a year or 2 ago (certainly longer than 6 months ago):

(error in red)
Old-ChatGPT-Session wrote:
kirtu wrote:What kind of reasoning is possible in CLIPS?
ChatGPT wrote: CLIPS supports forward chaining, backward chaining, and hybrid reasoning approaches, allowing for various types of reasoning. Here's a brief explanation of each type:

Forward Chaining: In forward chaining, also known as data-driven or bottom-up reasoning, the system starts with a set of facts and applies rules to derive new conclusions. It repeatedly matches patterns in the working memory and fires rules based on those matches. As new facts are derived, they can trigger further rule activations. This process continues until no more rules can be fired or until a specific goal is reached.

Backward Chaining: Backward chaining, also called goal-driven or top-down reasoning, starts with a specific goal or query and attempts to prove it by working backward through the rules. It recursively applies rules in reverse, attempting to satisfy the conditions of the goal by inferring new facts or goals until a termination condition is met. This approach is commonly used in rule-based expert systems for problem solving and diagnosis.

Hybrid Reasoning: CLIPS also supports a hybrid reasoning approach that combines forward chaining and backward chaining. This approach allows for a more flexible and adaptive reasoning process. It can use forward chaining to derive new information and trigger backward chaining when specific goals or queries need to be satisfied.

The ability to employ different reasoning approaches provides flexibility in modeling and solving problems in CLIPS. The choice of reasoning strategy depends on the nature of the problem domain and the desired behavior of the system.
kirtu wrote:Basic CLIPS doesn't support backward reasoning. That is an extension added by a group around the mid-late 90's but it didn't catch on widely.
ChatGPT wrote: Apologies for the incorrect information in my previous response. You are correct that the basic CLIPS language does not natively support backward chaining. Backward chaining was indeed added as an extension called "CLIPS-JESS" developed by Ernest Friedman-Hill in the mid-late 1990s, which integrated the Java Expert System Shell (JESS) into CLIPS.

The core CLIPS language primarily focuses on forward chaining, which is the default and widely used reasoning mechanism. Backward chaining capabilities were introduced through extensions like CLIPS-JESS and some other third-party tools. However, as you mentioned, these extensions did not gain widespread popularity.

Thank you for pointing out the distinction, and I apologize for any confusion caused by my previous response.
And JESS did provide some degree of hybrid reasoning but I didn't mention it. I probably overlooked it because there was a FuzzyClips implementation and I probably thought that was what ChatGPT was referring to. Prior to CLIPS there were also other related systems that supported backward reasoning (like at least some of the OPS family).
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
justsit
Posts: 1478
Joined: Wed Oct 21, 2009 9:24 pm
Location: Delaware

Re: The Morality of AI such as ChatGPT

Post by justsit »

OK,yes, there's been a miscommunication here.

As a trans person, also gay, I'm only laughing at the premise that the US is safe for LGBTQ+ people. Yes, it may be marginally safER in some places for some groups than in the past, but for most of us not located in blue states, not much has changed, and in fact some hard-won gains, in the South particularly, have actually been lost. The Chat response is merely parroting a bunch of "facts" it picked out somehow (some algorithm, who knows?), and spewed back, hardly relevant to many of us on the ground. There are plenty of places even in my blue state where I won't use the bathroom.

As an aside, I do use male pronouns, thanks.
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

Logging in turns out to have been a treasure trove as ChatGPT, at least back then (ChatCPT 3.5) threw errors all the time!

Here's an except:
OldChatGPT-Session wrote:
ChatGPT wrote: Given these properties, we cannot determine the exact nature (rational, irrational, or transcendental) of the sum π + i without knowing the precise values of π and i.
kirtu wrote:We do know the precise values of π and i. π = 3.14159265... (approximately) and as you mentioned i is the square root of -1.
So just no: knowing the precise values has nothing to do with it at all (we *DO* know the precise values, just not all the digits of π - and never can) but this has nothing to do with the question at all. π + i is an imaginary number because i is an imaginary number. It's not rational, irrational or transcendental. It's a complex number with a transcendental component.

BUT - I did put this question to the current ChatGPT (using a ChatGPT 4 model) and it answered correctly:
ChatGPT-Session wrote:
kirtu wrote:What kind of number is π + i ?
ChatGPT wrote: The number 𝜋 + 𝑖 is a complex number.

Here's a breakdown of the terminology and types of numbers involved:
...

To summarize, 𝜋 + 𝑖 fits into the category of complex numbers, specifically. It is not purely real or purely imaginary but has both a real part (𝜋) and an imaginary part (𝑖).
So now it can identify some complex numbers adequately.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
kirtu
Former staff member
Posts: 7088
Joined: Mon Jan 18, 2010 5:29 pm
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by kirtu »

ChatGPT does physics

and in general does it very badly. The professor reviewing the results would caution a student about relying on ChatGPT.

It does generate a partially correct Python solution to a physics problem - the professor was impressed by that but personally I've seen generated code for quite a while and this wasn't special. At the end of the day the solution was only partially correct, thus incorrect but possibly useable in a restricted sense.

However this was definitely using the old 3.5 model. Even though ChatGPT defaults to that general model (I say general model because actually it's an aggregate of available models) until one pays for ChatGPT (and even then you mostly get the 3.5 model on the lower pay tier) the older 3.5 model was definitely improved over the past year.
“Where do atomic bombs come from?”
Zen Master Seung Sahn said, “That’s simple. Atomic bombs come from the mind that likes this and doesn’t like that.”

"Even if you practice only for an hour a day with faith and inspiration, good qualities will steadily increase. Regular practice makes it easy to transform your mind. From seeing only relative truth, you will eventually reach a profound certainty in the meaning of absolute truth."
Kyabje Dilgo Khyentse Rinpoche.

"Only you can make your mind beautiful."
HH Chetsang Rinpoche
User avatar
Ayu
Global Moderator
Posts: 13373
Joined: Mon Oct 29, 2012 8:25 am
Location: Europe

Re: The Morality of AI such as ChatGPT

Post by Ayu »

Is anybody upset, if we say: "ChatGPT is not as smart as many people hope it to be."?
At least this is what I observed.
narhwal90
Global Moderator
Posts: 3577
Joined: Mon Jan 25, 2016 3:10 am
Location: Baltimore, MD

Re: The Morality of AI such as ChatGPT

Post by narhwal90 »

kirtu wrote: Sat Jun 22, 2024 3:54 pm
and in general does it very badly. The professor reviewing the results would caution a student about relying on ChatGPT.

It does generate a partially correct Python solution to a physics problem - the professor was impressed by that but personally I've seen generated code for quite a while and this wasn't special. At the end of the day the solution was only partially correct, thus incorrect but possibly useable in a restricted sense.
A friend of mine uses chatgpt to write code utilizing the standard libraries to merge terrain maps. He's found it somewhat helpful to arrange the program flow and manage the api calls, but the code it generates needs to be deeply reviewed because its usually not quite correct. Essentially there is no intelligence and experience in play, chatgpt is doing really fancy matching and inference so there is no evaluation of correct and appropriate in the results. The code it generates does run, control flow and datatypes are plausible but often functionally incorrect. It seems akin to how the AI drawing engines often mess up things like fingers.
Post Reply

Return to “Dharma in Everyday Life”