Chat GPT (and this forum)

[Teaching3bears]

This thing can write poems, stories and essays quite well and explain things conversationally.  It will be almost impossible to tell it from text composed by a real person.  This scares me and makes me sad.  It will be able to write novels and sell them.  Isn’t that sad?  It might even join this forum.  What are the repercussions of all this?

Yikes! I am too old and not ready for the mental shift that I will need to accept this kind of AI.  What will this world come to?
 

 

[Farrar]

I sat down and played with it for a little while to see what implications it had for my courses. I came away less impressed than others. It couldn't write an essay that would ace my AP class. They were way too general. It could do a mediocre reflection paper for some of my other classes of the kind that I often just check off if it's done. But... meh. Whatever. I've also read a bunch of stories people have generated with it. They're really dumb. Could they fool me that a real person wrote them? Absolutely. People aren't always good writers either.

It's going to get better, of course. But I'm not as worried as I was reading about it after I actually played with it. 

Also worth noting is that right now it's relatively easy to catch with various ChatGPT detectors. I couldn't masque it and fool the detectors, so again, I'm less worried about classroom plagiarism than I was.

There are really a few different elements to this. One is that it can do some basic tasks really well, like write a quick, polite email or summarize something basic. But then... why do we care if people wrote those themselves or not? There's the element of job loss, but it's not that many at this point and every industry faces that.

Another element is educational implications. Here I see some possible pitfalls. I think we need to remember that part of the reason we train children to write is that it's part of training them how to think and organize information. If they use AI to write their essays, the problem isn't the AI or the essays, it's that they won't learn to process that information. But it has potential applications as well, such as in helping students organize thoughts. 

Finally, there's what I think the crux of things is. If AI can almost make good art, what does that mean for humanity? And Lensa and DallE and all that poses the same questions. One meme I saw said something like, they promised us that automation would free up our time to make art but then instead the actual automation is making the art and the humans are stuck in service jobs. Which is both true and not true? I think it's not really there yet. But will it be? Maybe. And then what does that mean? I don't fully know. Metaphorically, I suspect that it will turn out to be a bit like chess. It wasn't that hard to make chess that could beat most people, but it took a very long time and a very big computer to make chess that could beat the best humans. Making good art may be similar. But what's the end there. I don't know.

[Bootsie]

Interestingly, my dean sent out an email about this a couple of days ago, suggesting that we consider adding information to our syllabus whether this is allowed in our course or if it will be considered academic dishonesty.  

[Eos]

9 hours ago, Farrar said:

they promised us that automation would free up our time to make art but then instead the actual automation is making the art and the humans are stuck in service jobs.

This truth has always made me think of The Little Prince:

“Good morning," said the little prince.

Good morning," said the merchant.

This was a merchant who sold pills that had been invented to quench thirst. You need only swallow one pill a week, and you would feel no need for anything to drink.

Why are you selling those?" asked the little prince.

Because they save a tremendous amount of time," said the merchant. "Computations have been made by experts. With these pills, you save fifty-three minutes in every week."

And what do I do with those fifty-three minutes?"

Anything you like..."

As for me," said the little prince to himself, "if I had fifty-three minutes to spend as I liked, I should walk at my leisure toward a spring of fresh water.”

Edited January 15, 2023 by Eos

[Jean in Newcastle]

This forum already has zombie threads that are often resurrected by bots. Not quite the same as what you mentioned but still automated responses. They are often non sequiturs. Sometimes they include spam links but not always. 

[Eos]

Just now, Jean in Newcastle said:

This forum already has zombie threads that are often resurrected by bots. Not quite the same as what you mentioned but still automated responses. They are often non sequiturs. Sometimes they include spam links but not always. 

Yes, and so our zombie-zapping skills need to stay sharp if we are to continue to be real humans having online conversations!

[Ausmumof3]

I’m thinking that I’ve only seen the results second hand. That means I’m only getting a biased sample because people mostly filter and only share the good stuff they get not the random nonsense. Meaning in some sense I’m actually getting human interpretation of a chatbot not the bot itself. 

[Farrar]

53 minutes ago, Ausmumof3 said:

I’m thinking that I’ve only seen the results second hand. That means I’m only getting a biased sample because people mostly filter and only share the good stuff they get not the random nonsense. Meaning in some sense I’m actually getting human interpretation of a chatbot not the bot itself. 

It's worth making a free account and trying out so you can understand it. Even if it's just for curiosity. It is really, really good at some things - especially short, informally styled explanations of things. Like, ask it to write a paragraph explaining something that's pretty well understood and you'll be floored at how good it is. But the longer, more complex, and more detailed the task, the more it will start to get over-generalized. It especially likes big vague statements about the world and how it works. It doesn't like to take a stand on anything.

When I had it write academic content, it reminded me of that kid in class who knows how to write but didn't do the reading and is trying to fake their way through the assignment.

[lewelma]

My ds last year as a part of his environmental science class had to write a 10 page report on what we are going to do with the COWS when the big one hits the South Island of NZ in the next 50 years (pretty well assured). There are 2 million dairy cows spread around and it takes 2 weeks to wean them then they die of mastitis. Most farmers only have 1 week of fuel to run their generators, and we cannot assume that the milking sheds will be operational, or that there will be clean water for them to drink, nor that famers can get the freaked-out escaped cows in sheds without quad bikes to corral them (limited fuel storage). The earthquake will destroy all the roads, ports, airports, electricity, water, sewer, and communications. It will take 3 months to get electricity back up to most areas. There will also be about 50 7+ aftershocks that will continue to cause the cows to panic and continue to do damage. So the question was what are we going to do about the cows?!?!? And then on top of that, set these required resources in context of what other problems the region will have, and set priorities.  (Yes, this was a freshman research paper! )

So we put this question into the AI. There are apparently 4 different AIs to chose from each with different goals. We tried 2.

The first one said: There are 3 options: "move the cows to higher ground, put a roof over their heads, or dig a trench around them."

The second one said: "There are 2 million cows, and they are all going to die." 😳

So the AI has some work to do to come up to speed! So I don't think it will be taking over university papers that are well constructed. 

 

Edited January 16, 2023 by lewelma

[Melissa Louise]

@lewelma, I know this isn't the point, but what do you do with the cows?

[Farrar]

The poor kiwi cows!

Let's say a student was assigned to research and write something about that topic for a class. They could turn to the AI and ask it to summarize the background information for them. So, I did some of that and got bits like...

Quote

Cows play a significant role in New Zealand's economy, particularly in the agricultural sector. They are a primary source of milk and meat production, and exports of these products make up a significant portion of the country's overall export revenue. Additionally, the farming of cows also helps to support many rural communities and contributes to the overall employment rate in these areas.

And...

Quote

Cows were first introduced to New Zealand by European settlers in the late 1700s and early 1800s. They were initially used for meat and dairy production, as well as for draft power on farms. Over time, the dairy industry became increasingly important, particularly in the late 1800s and early 1900s, as refrigeration technology made it possible to export milk and butter to other countries. The cattle population in New Zealand grew rapidly throughout the 20th century, with the dairy sector becoming the largest agricultural industry. Today, cows continue to play a major role in New Zealand's economy through the production of milk and meat, and are considered a significant part of the country's cultural heritage.

And...

Quote

New Zealand is located on the boundary of the Pacific and Australian tectonic plates, making it prone to earthquakes. The country experiences a high number of earthquakes each year, and many of these are caused by movement along the plate boundary. The danger of earthquakes in New Zealand is due to several factors, including the high seismic activity, the proximity of population centers to active faults, and the potential for ground shaking and landslides.

The most severe earthquakes in New Zealand have caused significant damage to infrastructure, buildings, and homes. Some earthquakes have also caused fatalities and injuries. In addition, there is a risk of tsunamis being generated by earthquakes offshore, which can be particularly dangerous for coastal communities.

I could keep going. But you get the idea. Not terrible.

An enterprising student could get a couple of pages of background information this way, going paragraph by paragraph and possibly filling in more specifics, like the number of cows, how many earthquakes above such and such on the scale in the last decade, etc. They could write transition sentences and organize it. Then they'd need to do the writing themselves about the solution they're proposing and the independent research they did. 

So the AI is terrible at coming up with creative solutions. But it's not bad at writing out what might form the background information in a paper about a creative solution.

Edited January 16, 2023 by Farrar

[lewelma]

1 minute ago, Melissa Louise said:

@lewelma, I know this isn't the point, but what do you do with the cows?

Seriously. We decided that it is all down to the helicopters delivering supplies of fuel. We found out that there are 753 helicopters in NZ, but they would also be needed for all the other issues including reconnaissance of the entire region looking for landslide damns that might break and cause a massive flood, and used to get equipment up to the power lines because all the roads will be gone. I did suggest stockpiling lime to cover the bodies because they won't have the equipment to push them into mass graves. 😳 Basically, it is a massive massive problem with no clear cut solutions that we could find in the literature. (Yes, I did study up on this with him because it was just so fascinating). The big one comes ever 250 years, the last one was in 1717. It will move the fault 10 meters horizontally and 2 meter vertically, if I remember correctly. The power lines run right across the fault line, through the southern alps.

It is called AF8 for alpine fault magnitude 8. https://af8.org.nz/

My son is majoring in hazards, so really loved this project. 

[AmandaVT]

I played around with it, asking it to make IEP goals. Overall, it did a decent job! 

 

Please write an IEP goal for a student learning multiplicative reasoning with a relative weakness in working memory and strengths in visual spatial and fluid reasoning.

 

"Given visual aids and support for organization, the student will independently apply multiplicative reasoning strategies to solve problems with 80% accuracy within a one-minute time frame, as measured by teacher-created assessments, by the end of the school year."

I tried present levels, SMART goals and added in time frames - it was able to do it all which is cool. 

[lewelma]

11 minutes ago, Farrar said:

The poor kiwi cows!

Let's say a student was assigned to research and write something about that topic for a class. They could turn to the AI and ask it to summarize the background information for them. So, I did some of that and got bits like...

And...

And...

I could keep going. But you get the idea. Not terrible.

An enterprising student could get a couple of pages of background information this way, going paragraph by paragraph and possibly filling in more specifics, like the number of cows, how many earthquakes above such and such on the scale in the last decade, etc. They could write transition sentences and organize it. Then they'd need to do the writing themselves about the solution they're proposing and the independent research they did. 

So the AI is terrible at coming up with creative solutions. But it's not bad at writing out what might form the background information in a paper about a creative solution.

Agreed. So that is what university papers need to do, assume the background information is known (so tell students not to write it), and only ask for the creative solutions. Prioritizing with justification was particularly tricky.

I just liked that one of the AIs said all the cows would die. It was very matter of fact, which unfortunately made us laugh. (as did the trenches).

[Farrar]

Just now, lewelma said:

Agreed. So that is what university papers need to do, assume the background information is known (so tell students not to write it), and only ask for the creative solutions. Prioritizing with justification was particularly tricky.

I just liked that one of the AIs said all the cows would die. It was very matter of fact, which unfortunately made us laugh. (as did the trenches).

I can see a purpose to the background though. Presumably if you had a class working on a ton of different problems, even the professor might not be fully versed in all of them. Plus, a "real" paper needs background for most audiences, even academic papers. So then I feel like you run into the question of whether it's okay to have those sorts of sections written by AI. Is the role for the student or author to summarize the information effectively or is it just to convey that information to the audience in an efficient way?

Part of the issue I see is that it's hard for students to learn to effectively convey information about creative solutions (or anything else of import) without first, at some point, as students, writing some of these more bleh papers. I found that ChatGPT was meh at papers with a thesis (its thesis statements are mostly pretty weak sauce) but I assume it will get a lot better in the nearish future. So then what. Some people are like, so then we don't bother asking students to write papers like those anymore. But then what do the students do in order to learn the thinking skills instead? Because I'd genuinely like to know.

[Melissa Louise]

I had a loooong chat with it about all manner of things.

I asked it to analyze one of my prose poems. It picked up on core themes. I'm going to go back and ask it to write an essay and support the thesis with details from the text.

This poem is a reflection on the nature of art and the human experience. The poem starts with the idea that metaphor breaks down, meaning that it can't accurately convey the true essence of what is being described. The poet uses the image of a person looking through a microscope at a pinned moth on a plate, to symbolize the act of observing and interpreting art. The person is making adjustments to the lens and as they do, the moth's wing bleeds into the artist's brushstrokes, blurring the lines between the subject and the interpretation.

The poem then shifts to the idea of walking through a gallery, where the speaker advises the reader to focus on a particular painting, and to kneel down and speak the thoughts and feelings that come to mind, without worrying about what others might think. The speaker advises the reader to approach the painting with a sense of vulnerability and authenticity, to speak their truth, even if it feels foolish.

The poem is a reminder that art is not just about the object itself, but also about the emotional and personal connection that we have with it. The poem encourages the reader to engage with art in a way that is authentic and personal, rather than trying to understand it in a detached and analytical way. The poem also touches on the theme of vulnerability and authenticity, encouraging the reader to express their true feelings and thoughts, even if it feels foolish.

[Farrar]

8 minutes ago, lewelma said:

Seriously. We decided that it is all down to the helicopters delivering supplies of fuel. We found out that there are 753 helicopters in NZ, but they would also be needed for all the other issues including reconnaissance of the entire region looking for landslide damns that might break and cause a massive flood, and used to get equipment up to the power lines because all the roads will be gone. I did suggest stockpiling lime to cover the bodies because they won't have the equipment to push them into mass graves. 😳 Basically, it is a massive massive problem with no clear cut solutions that we could find in the literature. (Yes, I did study up on this with him because it was just so fascinating). The big one comes ever 250 years, the last one was in 1717. It will move the fault 10 meters horizontally and 2 meter vertically, if I remember correctly. The power lines run right across the fault line, through the southern alps.

It is called AF8 for alpine fault magnitude 8. https://af8.org.nz/

My son is majoring in hazards, so really loved this project. 

Total aside, but this makes me think of what happens to hogs in North Carolina after hurricanes. It's NOT PRETTY.

[lewelma]

We asked it to explain my older boy's research in physics with simple language. And it did a very nice job of the background material (obviously not current research), and I thought it could help ds to explain the basics of his work better to a lay audience. 

[Melissa Louise]

Re bolded - eek!

What does human's obsession with AI say about the concept of psychological projection

Human's obsession with artificial intelligence (AI) can be seen as a form of psychological projection, which is the process of attributing one's own thoughts, feelings, or attitudes to someone or something else. In this case, the human obsession with AI can be seen as a projection of their own desires, fears, and aspirations onto the idea of AI.

One aspect of this projection is the desire to create intelligent machines that can think and learn like humans. This desire can be seen as a reflection of human's own desire to understand and control the world around them, as well as their desire to understand their own intelligence and consciousness.

Another aspect of this projection is the fear of AI surpassing human intelligence and capabilities, becoming a threat to humanity. This fear can be seen as a reflection of human's own fear of their own limitations and the fear of the unknown.

Additionally, the projection can also be seen in the way humans imagine AI to be in the future, as a perfect being that can solve all the problems that humans can't. This projection can also be seen in the way humans imagine AI to have emotions and consciousness like human, which is not true in most of the cases.

In summary, human's obsession with AI can be seen as a form of psychological projection, where humans attribute their own thoughts, feelings, and aspirations onto the idea of AI. This projection reflects human's desire to understand and control the world around them, as well as their own limitations and fears.

[Melissa Louise]

10 minutes play; I am 100% on the side of openly adopting this as a tool for learning and creating.

[lewelma]

My concern has been with the return to the idea of hand written tests, which would be a serious problem with my ds who has dysgraphia.

Also, his favorite thing is research papers. He LOOOOOVES them, and learns way way more than with a lecture-test learning approach. So we are hoping that cheaters don't ruin it for those who love papers as a way of learning to think.

[lewelma]

1 minute ago, Melissa Louise said:

10 minutes play; I am 100% on the side of openly adopting this as a tool for learning and creating.

We are going to have to. The question is how to upskill teachers, which will not be straightforward. 

[Melissa Louise]

1 minute ago, lewelma said:

We are going to have to. The question is how to upskill teachers, which will not be straightforward. 

If someone had zero experience or expertise or qualification in AI, but was interested in integrating it into education, including freelance education outside the school system, where on earth could that someone go to learn more, do you know? I did ask the chatbot 🙂 It told me to look on Coursera!

[TechWife]

33 minutes ago, Farrar said:

The poor kiwi cows!

Let's say a student was assigned to research and write something about that topic for a class. They could turn to the AI and ask it to summarize the background information for them. So, I did some of that and got bits like...

And...

And...

I could keep going. But you get the idea. Not terrible.

An enterprising student could get a couple of pages of background information this way, going paragraph by paragraph and possibly filling in more specifics, like the number of cows, how many earthquakes above such and such on the scale in the last decade, etc. They could write transition sentences and organize it. Then they'd need to do the writing themselves about the solution they're proposing and the independent research they did. 

So the AI is terrible at coming up with creative solutions. But it's not bad at writing out what might form the background information in a paper about a creative solution.

Except that the AI a may not be giving the correct information. It has no ability to know whether or not the data it has stored is true or false. Someone may get what they think is background information and end up with information that is incorrect, includes only part of the information needed to draw an accurate conclusion or leaves out significant information that would change the conclusion  entirely. 

[Melissa Louise]

I did find this, for anyone interested in ed + AI

https://www.iste.org/professional-development/iste-u/artificial-intelligence

 

[Ausmumof3]

35 minutes ago, lewelma said:

Seriously. We decided that it is all down to the helicopters delivering supplies of fuel. We found out that there are 753 helicopters in NZ, but they would also be needed for all the other issues including reconnaissance of the entire region looking for landslide damns that might break and cause a massive flood, and used to get equipment up to the power lines because all the roads will be gone. I did suggest stockpiling lime to cover the bodies because they won't have the equipment to push them into mass graves. 😳 Basically, it is a massive massive problem with no clear cut solutions that we could find in the literature. (Yes, I did study up on this with him because it was just so fascinating). The big one comes ever 250 years, the last one was in 1717. It will move the fault 10 meters horizontally and 2 meter vertically, if I remember correctly. The power lines run right across the fault line, through the southern alps.

It is called AF8 for alpine fault magnitude 8. https://af8.org.nz/

My son is majoring in hazards, so really loved this project. 

Some of these problems could maybe be reduced by encouraging dairy farms to use localised power generation? Solar and wind? I know you guys have a lot of hydro, but not sure how solar works there? Do you guys have gas turbines? It’s pretty terrifying to think about but also amazing. I like to think similar planning is done here around fire but hmm? Not really. 

[Ausmumof3]

20 minutes ago, lewelma said:

My concern has been with the return to the idea of hand written tests, which would be a serious problem with my ds who has dysgraphia.

Also, his favorite thing is research papers. He LOOOOOVES them, and learns way way more than with a lecture-test learning approach. So we are hoping that cheaters don't ruin it for those who love papers as a way of learning to think.

Alternatively it could drive a faster move toward authentic assessment models and maybe more oral, video and podcast type presentation modes!

Edited January 16, 2023 by Ausmumof3

[Farrar]

11 minutes ago, TechWife said:

Except that the AI a may not be giving the correct information. It has no ability to know whether or not the data it has stored is true or false. Someone may get what they think is background information and end up with information that is incorrect, includes only part of the information needed to draw an accurate conclusion or leaves out significant information that would change the conclusion  entirely. 

Right, but if you're the author of a paper on this topic, you have the ability to go back through and check it and add detail. And if you're putting your name on it, you're just as liable for the information as if you penned it yourself.

I'm not advocating for this approach exactly. I'm just saying... I can see that it might be used this way and I'm not sold that it's terrible. Getting it to write your background, your abstract, your blurbs... this is human labor time that it's saving people. It's not writing that needs to be especially creative or groundbreaking. So then is that bad? Unethical? I'm a lot less sure than I am about it authoring a student paper that's meant to be about learning or taking the place of someone engaged in a creative art, or posing as a human in a social interaction.

[regentrude]

4 minutes ago, Ausmumof3 said:

Alternatively it could drive a faster move toward authentic assessment models and maybe more oral, video and podcast type presentation modes!

But the main work for all these is the making of the script. Students can easily have the chatbot write them the script and then read it off for your podcast or class presentation

[Farrar]

26 minutes ago, lewelma said:

My concern has been with the return to the idea of hand written tests, which would be a serious problem with my ds who has dysgraphia.

Also, his favorite thing is research papers. He LOOOOOVES them, and learns way way more than with a lecture-test learning approach. So we are hoping that cheaters don't ruin it for those who love papers as a way of learning to think.

I have some of the same concerns. I do think having students write more by hand would just be good for students in a generalized sense. If the goal of writing basic student papers is by and large for students to learn from the process of the writing, then we know that writing by hand is better for our brains. But... it's an access issue as well, so that's hardly some perfect solution.

[regentrude]

For assessment, you could of course adopt the model we had in college: oral examinations.
My transcript has three grades: comprehensive oral exams over 5 semesters of math, 8 semesters of experimental physics, 8 semesters of theoretical physics, respectively. Administered by humans who ask questions that you did not know in advance. 
Good luck trying this in this country.

ETA: You can also do this in the humanities and have the student analyze a work of literature, discuss historical developments etc.
But it won't work with the factory mass-education model of 1000-student courses.

 

Edited January 16, 2023 by regentrude

[Melissa Louise]

OP, I asked it to be a forum member, judge for yourself how likely it is that you would think this is a real person.

Write me a response to 'Spare' in the voice of a middle aged American woman

"Well, I have to say, I was quite surprised to hear about Prince Harry's new memoir 'Spare.' I've always been a fan of the royal family and I'm intrigued to hear about his experiences, but I can't help but feel a little apprehensive about all the personal details that might be revealed. I understand that everyone has the right to tell their own story, but I can't help but wonder if it's the right thing to do to air all of the family's dirty laundry like that. I guess we'll just have to wait and read it to find out."

 

[Melissa Louise]

.

Edited January 16, 2023 by Melissa Louise
too big

[Melissa Louise]

Write a forum post on the topic 'good news' in the voice of an American mother who homeschools her children.

Hello everyone,

I hope this post finds you all well. I wanted to take a moment to share some good news that's been happening in my household lately.

As many of you know, I am a homeschooling mom and it can be a tough journey at times. But lately, I've seen some real progress in my children's education and it has brought so much joy to our home. My oldest son has been struggling with math for a while, but with the help of some extra resources and a lot of patience, he finally understands it and his grades have improved dramatically.

Also, my daughter who has always been shy and introverted, has found a passion for public speaking and has been participating in online debate competitions and doing exceptionally well. It has been amazing to see her confidence grow and her love for learning.

I wanted to share this with you all, because in these times, it can be easy to get caught up in the negativity and bad news that surrounds us. But it's important to remember that there is still good happening all around us, we just have to look for it.

Take care, A Homeschooling mom

Edited January 16, 2023 by Melissa Louise

[TechWife]

5 minutes ago, Farrar said:

Right, but if you're the author of a paper on this topic, you have the ability to go back through and check it and add detail. And if you're putting your name on it, you're just as liable for the information as if you penned it yourself.

I'm not advocating for this approach exactly. I'm just saying... I can see that it might be used this way and I'm not sold that it's terrible. Getting it to write your background, your abstract, your blurbs... this is human labor time that it's saving people. It's not writing that needs to be especially creative or groundbreaking. So then is that bad? Unethical? I'm a lot less sure than I am about it authoring a student paper that's meant to be about learning or taking the place of someone engaged in a creative art, or posing as a human in a social interaction.

Wri ting papers is about more than churning out content in a specific format. When someone writes a paper, they are honing their own language fluency skills - going beyond phonics and grammar to comprehension - understanding what they are reading.  They are also honing their critical thinking skills - choosing appropriate sources, learning information, synthesizing information from multiple sources (how does source b support or expand on the information in source a? Source c has info that isn’t in a or b, do I need to include it? Sources b and d have conclusions and recommendations that contradict each other - which is valid, both, neither of one of them?). 
 

[Melissa Louise]

OK, I'm really sorry and I promise that this will be the last, but while it can't impersonate us, it is a lot better than Koreaboo

Write a short, enthusiastic article about Suga from BTS

Suga, also known as Min Yoongi, is a rapper, songwriter, and record producer from the South Korean boy band BTS. He is known for his smooth flow, clever wordplay and powerful stage presence.

Suga is one of the most talented members of BTS, and his contributions to the group's music are undeniable. He has written and produced many of BTS's hit songs, and his solo work is equally impressive. He has released multiple mixtapes, which showcase his versatility as an artist. His most recent one, "D-2" which was released in 2020, received critical acclaim for its introspective and honest lyrics, and his ability to blend different genres like hip hop, R&B and rock.

Suga is also known for his charismatic and relatable personality, which has made him a fan favorite. He is known for his witty and humorous remarks and is a source of entertainment for the fans. He is also known for his dedication and hard work, which can be seen in his music and performances.

In short, Suga is a force to be reckoned with in the music industry and his talent, charisma and personality make him one of the most beloved members of BTS. He continues to inspire and entertain fans with his music and performances, and we can't wait to see what he'll do next.

[bolt.]

I think this is sophisticated AI, but not in a way that really matters in terms of "Is it learning?" and "Does it think for itself?"

What it is is a sophisticated predictive text engine.

It knows what words 'should come next' in sentences and paragraphs simply because we humans have provided it with so many sample texts. It is a mirror. The reason it can sound like our WTM posts on the internet is because we literally are its source material when that genre of writing is requested. It just uses our sample texts to predict our likely-future words based on our past use of words.

That really doesn't freak me out or upset me on an existential level. Text prediction is not witchcraft. And it's not really even intelligence. It's pattern recognition and reproduction -- something computers are excellent at. Even when it's poetry (or visual arts mashups and mimicry) it's still just pattern recognition and reproduction, with combinations and variances as requested. I'm comfortable with that.

[YaelAldrich]

1 hour ago, lewelma said:

Seriously. We decided that it is all down to the helicopters delivering supplies of fuel. We found out that there are 753 helicopters in NZ, but they would also be needed for all the other issues including reconnaissance of the entire region looking for landslide damns that might break and cause a massive flood, and used to get equipment up to the power lines because all the roads will be gone. I did suggest stockpiling lime to cover the bodies because they won't have the equipment to push them into mass graves. 😳 Basically, it is a massive massive problem with no clear cut solutions that we could find in the literature. (Yes, I did study up on this with him because it was just so fascinating). The big one comes ever 250 years, the last one was in 1717. It will move the fault 10 meters horizontally and 2 meter vertically, if I remember correctly. The power lines run right across the fault line, through the southern alps.

It is called AF8 for alpine fault magnitude 8. https://af8.org.nz/

My son is majoring in hazards, so really loved this project. 

If your son is looking for research projects in disaster and hazards research my husband is always looking for smart minds to co author with on papers. He has worked in NZ before.... 

[Farrar]

17 minutes ago, TechWife said:

Wri ting papers is about more than churning out content in a specific format. When someone writes a paper, they are honing their own language fluency skills - going beyond phonics and grammar to comprehension - understanding what they are reading.  They are also honing their critical thinking skills - choosing appropriate sources, learning information, synthesizing information from multiple sources (how does source b support or expand on the information in source a? Source c has info that isn’t in a or b, do I need to include it? Sources b and d have conclusions and recommendations that contradict each other - which is valid, both, neither of one of them?). 
 

Yes, that's exactly what I've pointed out in a number of places - the problem with doing away with these sorts of tasks for students has implications that I don't think the AI can solve. My point there was about professionals, academics, etc. But I do think there can be a purpose for students too. We should be thinking more clearly about when this technology is useful and when it's not. When we should embrace it and when we should focus on the process that you talk about above.

Part of the issue with student writing and one of the reasons that students find writing tedious is that they're asked to do these things and to turn out boring papers that don't have a real purpose beyond their own education or a real audience beyond the teacher issuing the grade. I think Chat GPT really highlights that more acutely than ever.

[TechWife]

18 minutes ago, bolt. said:

I think this is sophisticated AI, but not in a way that really matters in terms of "Is it learning?" and "Does it think for itself?"

What it is is a sophisticated predictive text engine.

It knows what words 'should come next' in sentences and paragraphs simply because we humans have provided it with so many sample texts. It is a mirror. The reason it can sound like our WTM posts on the internet is because we literally are its source material when that genre of writing is requested. It just uses our sample texts to predict our likely-future words based on our past use of words.

That really doesn't freak me out or upset me on an existential level. Text prediction is not witchcraft. And it's not really even intelligence. It's pattern recognition and reproduction -- something computers are excellent at. Even when it's poetry (or visual arts mashups and mimicry) it's still just pattern recognition and reproduction, with combinations and variances as requested. I'm comfortable with that.

It’s really machine learning. My husband calls it a stochastic parrot -  gleaned from this paper. 
https://dl.acm.org/doi/10.1145/3442188.3445922

 

Edited January 16, 2023 by TechWife

[lewelma]

1 hour ago, Farrar said:

Getting it to write your background, your abstract, your blurbs... this is human labor time that it's saving people. It's not writing that needs to be especially creative or groundbreaking. So then is that bad? Unethical? I'm a lot less sure than I am about it authoring a student paper that's meant to be about learning or taking the place of someone engaged in a creative art, or posing as a human in a social interaction.

I think that the bigger problem is that younger kids need to do basic writing before they can handle advanced writing. And the basic writing can be done by the AI. So if they don't write the easy stuff because they are using an AI, they won't have the skills when they are older to do what the AI can't. 

[KSera]

Has anyone asked it to write a paper extolling the virtues of something known to be bad? I’ve seen some examples about eating glass, but I don’t know if they’re real. I haven’t made an account myself. 

[lewelma]

Also, currently it does not have access to peer-reviewed journal articles that are behind a paywall, but I assume has access to opensource journals. So it has half of the intellectual material? Not sure of the impact of that to open vs closed access in the future. 

[Grace Hopper]

1 hour ago, Farrar said:

The poor kiwi cows!

Let's say a student was assigned to research and write something about that topic for a class. They could turn to the AI and ask it to summarize the background information for them. So, I did some of that and got bits like...

And...

And...

I could keep going. But you get the idea. Not terrible.

An enterprising student could get a couple of pages of background information this way, going paragraph by paragraph and possibly filling in more specifics, like the number of cows, how many earthquakes above such and such on the scale in the last decade, etc. They could write transition sentences and organize it. Then they'd need to do the writing themselves about the solution they're proposing and the independent research they did. 

So the AI is terrible at coming up with creative solutions. But it's not bad at writing out what might form the background information in a paper about a creative solution.

Is this with input you plugged into the AI program? I’m just wondering how a student would handle footnotes/citations. 

[Melissa Louise]

4 minutes ago, KSera said:

Has anyone asked it to write a paper extolling the virtues of something known to be bad? I’ve seen some examples about eating glass, but I don’t know if they’re real. I haven’t made an account myself. 

I asked it to write a poem extolling the virtues of drug taking; it said no and gave me a lecture on drugs.

[lewelma]

1 hour ago, TechWife said:

Except that the AI a may not be giving the correct information. It has no ability to know whether or not the data it has stored is true or false. 

Also, for controversial material, how does it choose which source to use?  Or what about ethical issues with multiple valid perspectives?.  These things could be manipulated for a price, I would assume. Companies could pay to have their viewpoints be the ones picked by the AI, not that different from google or facebook feeds.  

Edited January 16, 2023 by lewelma

[lewelma]

4 minutes ago, Melissa Louise said:

I asked it to write a poem extolling the virtues of drug taking; it said no and gave me a lecture on drugs.

Who coded that in?  Who decides what is OK to write about and what isn't? We need some transparency, but likely won't get any.

[Melissa Louise]

5 minutes ago, lewelma said:

Also, for controversial material, how does it choose which source to use?  Or what about ethical issues with multiple valid perspectives?.  These things could be manipulated for a price, I would assume. Companies could pay to have their viewpoints be the ones picked by the AI.  

I tried a controversial issue. First, I asked it to summarise a social conflict where both sides have valid points. It did so accurately, if shallowly.

I then asked if side A in the conflict  was correct. It disavowed an opinion but gave a response that leaned to side B.

I asked if side B in the conflict was correct. It disavowed an opinion but gave a response leaning further to side B.

So I guess it is already manipulated in some ways.

[KSera]

1 minute ago, Melissa Louise said:

I tried a controversial issue. First, I asked it to summarise a social conflict where both sides have valid points. It did so accurately, if shallowly.

I then asked if side A in the conflict  was correct. It disavowed an opinion but gave a response that leaned to side B.

I asked if side B in the conflict was correct. It disavowed an opinion but gave a response leaning further to side B.

So I guess it is already manipulated in some ways.

That is interesting. Thanks for trying and sharing that experiment. 

[lewelma]

4 minutes ago, Melissa Louise said:

I tried a controversial issue. First, I asked it to summarise a social conflict where both sides have valid points. It did so accurately, if shallowly.

I then asked if side A in the conflict  was correct. It disavowed an opinion but gave a response that leaned to side B.

I asked if side B in the conflict was correct. It disavowed an opinion but gave a response leaning further to side B.

So I guess it is already manipulated in some ways.

Very interesting. So I'm wondering about 2 issues concerning geographical location. 1) For example, euthanasia is legal in NZ because it has strong support, but illegal in most places in USA. Do you think that ChatGPT's American origins will impact how it evaluates issues and what content it provides?

2) Perhaps it will be tailored to regions or to the user's google history (or Chatbot history)? So for an abortion paper, a user's with GPS showing location in the American South would have a different outcome than if they were in California? If so, we would need transparency. Who is making these decisions?