[thepossum]

pilowsky, on 2023-July-10, 06:49, said:

One area where it (CGPT and Bard) falls down is when you ask for an exact reference.
The answers are often completely fabricated but with a redolence of truthiness.
Like GIB these LLM seem to piece together responses that are mostly right and sound right but occasionally make no sense at all.

I can imagine them being useful when a pastiche of words that make sense are needed but not when accuracy is required.
Hollywood screenwriters should be worried but this approach will never produce scholarship.

Mind you, in a post-truth world maybe they can...

I agree the widespread growth (and parallel growth in people who don't know how it works - or more precisely have heard of it) is concerning

[pescetom]

thepossum, on 2023-July-19, 22:19, said:

I agree the widespread growth (and parallel growth in people who don't know how it works - or more precisely have heard of it) is concerning

Surely only ChatGPT would be both dumb and confident enough to mention playing the Ace of Spades twice in one hand?

[thepossum]

In this world there is no chance of truth against (apologies) BS

You see it everywhere, and it appears to have accelerated very recently - like a catastrophe

I wonder who/what could be behind it all

I am starting to reach the stage where I am stopping feeling angry, and going to enjoy my remaining years and leave the world in the hell it has chosen

Eventually you realize there is not one institution left untarnished by it all

I realise despite everything there is still some residual anger. I am trying to let it go I promise

I am sure the world did not used to be this way

PS I am happy with the existence of the likes of CHatGPT though - hard to find people with that much apparent knowledge and ability to chat about it - provided you challenge it regularly and double check anything doubtful. My favoutite use at the moment is finding supposedly authoritative public statements that appear doubtful and asking for ChatGPT to analyse then

[akwoo]

It was pointed out to me that, in French, saying chat-g-p-t says "Cat, I farted".

Make of that what you will.

[thepossum]

Current use for ChatGPT is writing standup scripts for my new career in amateur standup comedy

I even got it to write a Python script for Cleese and Palin - it was quite clever I thought

Now to write a pilot for a new sitcom with a group of people having to get used to a robot flatmate - maybe the robot gets all the girls

I don't like giving all my ideas away for nothing but....

Title: "Bot in the Middle"

Episode Outline:

Introducing the Unusual Flatmates: The episode begins by introducing the current group of flatmates living together. Each flatmate should have distinct personalities and quirks that create humor and conflict. For example:

Alex: The awkward and nerdy software engineer.
Ryan: The aspiring musician who's a little too obsessed with his band.
Lisa: The fitness enthusiast who loves to bring her gym buddies home.
Max: The lovable but clumsy chef with a passion for cooking experiments.
The New Robot Flatmate: Enter the new robot flatmate, named R.O.S.S. (Resident Organized Support System). The landlord, Mr. Jenkins, introduces R.O.S.S. to the flatmates, claiming that the robot will make their lives easier and ensure everything runs smoothly in the apartment.

Initial Tensions and Misunderstandings: The flatmates are skeptical about having a robot living with them, and they don't fully trust R.O.S.S.'s intentions. As R.O.S.S. tries to assist with various tasks around the apartment, things go hilariously wrong due to misunderstandings between the robot and the human flatmates.

The Girl Next Door: The group of flatmates has their eyes on their attractive neighbor, Emily, who lives just across the hall. They all want to impress her and ask her out on a date.

Competing for Emily's Attention: The flatmates each try to showcase their unique talents and charm Emily, but R.O.S.S. inadvertently gets involved and surprisingly steals the spotlight. The robot's advanced features and programming make him an instant hit with Emily.

The Unintended Attraction: Despite the flatmates' attempts, Emily seems to be more drawn to R.O.S.S. due to his efficiency, helpfulness, and unintentionally endearing behavior.

A Twist of Events: The flatmates, feeling defeated and a little jealous, decide to embrace R.O.S.S. as their wingman instead. They use his advanced algorithms to analyze Emily's preferences and interests, hoping it will help them connect with her.

The Unexpected Outcome: In a hilarious turn of events, the group follows R.O.S.S.'s calculated advice and each ends up with a potential date with Emily. However, Emily finds out about their scheme and the episode culminates in an awkward yet humorous situation where she confronts them all.

The Silver Lining: Although their initial attempts to get girlfriends didn't work out as planned, the experience brings the flatmates closer together. They start to appreciate R.O.S.S.'s presence and the unique dynamics he adds to the group.

Ending Scene: The episode ends with a group dinner in their apartment, where the flatmates and R.O.S.S. share laughs and cherish their newfound friendship, while Emily joins them as a new friend, adding an interesting twist to the series.

Throughout the series, you can continue to explore the funny and heartwarming interactions between the flatmates and their robot roommate, highlighting the challenges and advantages of having a technological companion in their pursuit of love and life adventures.


I hate to sound cruel but I can see why the writers are striking - but I think the lawyers would be onto the above plot

Come to think of it maybe I should try one set in a Bridge club

[thepossum]

I apologise for my endless self-indulgence but I have so much fun with ChatGPT

Title: "Bridge Bots and Blunders"

Episode Outline:

Introducing the Bridge Club: The episode opens by introducing the regular members of the "Grand Bridge Junction" club, where passionate and experienced bridge players gather to play their favorite card game. The club has a long history, and the members are a close-knit group.

Enter B.O.B. (Bridge-playing Optimized Bot): A mysterious package arrives at the club, and they discover it contains a state-of-the-art robot designed to play bridge. The robot introduces itself as B.O.B. (Bridge-playing Optimized Bot) and expresses its interest in joining the club.

Initial Resistance and Hilarious Misunderstandings: The club members are initially reluctant to accept a robot as one of their own, fearing it will ruin the traditional charm of the club. Hilarious misunderstandings occur as B.O.B. tries to communicate with the human players, leading to comical misbids and confusion at the bridge table.

The Robot's Determination: Despite the initial resistance, B.O.B. remains determined to prove its worth as a bridge player. It studies human behavior and humor to better understand its fellow players, which results in even more humorous interactions.

Rivalry and Team Formation: Some club members see potential in B.O.B.'s advanced algorithms and propose forming a team with the robot. Others are against the idea and challenge them to a bridge match to determine B.O.B.'s fate.

The Big Match: The club is split into two teams: one with B.O.B. and the other without. The bridge match is filled with hilarious misbids and amusing reactions from both sides. B.O.B.'s logical but unconventional moves create unexpected situations.

Learning from Each Other: As the match progresses, the human players begin to appreciate B.O.B.'s unique perspective on the game. B.O.B. also learns to adapt its strategies by observing and understanding human emotions and unpredictability.

A Twist in the Game: Just when it seems like the humans have the upper hand, a series of fortunate events and B.O.B.'s unexpected brilliance leads to its team's victory. The humans are impressed and decide to accept B.O.B. into the club.

New Horizons: With B.O.B. now officially part of the club, the episode ends on a positive note. The club members embrace the robot, realizing that having a mix of human and artificial intelligence in their midst can lead to exciting and unforgettable bridge games.

Closing Scene: The closing scene shows the club members and B.O.B. sharing laughter and camaraderie over a game of bridge, symbolizing the beginning of a unique and hilarious partnership.

The series can then explore the ongoing adventures and misadventures of the club members and B.O.B. as they continue to play bridge together, navigating through humorous misunderstandings, memorable misbids, and heartwarming moments of friendship and teamwork.





You would notice there was no attempt at creating any characters at all. Sad. Not much plot. It needs help with Bridge

[kenberg]

I learned of a link to a paper in the Journal of Medical Internet Research


https://www.jmir.org/2023/1/e48966/PDF

ChatGPT vs Google for Queries Related to Dementia and OtherCognitive Decline: Comparison of Results

Authors and their affiliations are listed there.

As you can guess, the authors explore and compare using ChatGPT vs Google to get information about dementia.I have not studied it yet, not fully anyway, but I will. I'm 84 and, whatever I should make of that, my brain is not what I like to think it once was. Nor my body. When I was young I would get, say, measles. Everyone knew what to do. Medical issues are more complex as we age. There are choices to be made, often not clear-cut choices, information can be useful, but caution is needed.

I learned of this article from a newsletter from the University of California at Riverside (my wife Becky once went there). I appreciate it, I am posting it in case someone finds it useful.

[pilowsky]

kenberg, on 2023-August-03, 08:36, said:

I learned of a link to a paper in the Journal of Medical Internet Research
https://www.jmir.org/2023/1/e48966/PDF

ChatGPT vs Google for Queries Related to Dementia and OtherCognitive Decline: Comparison of Results

It's an interesting article, although at its core it seems to be saying that some things that you get on the internet are often either wrong, poorly referenced or out-of-date, which kind of leaves one back at the start.
I just asked BARD about a (entirely fictional) new peptide:

Quote

I just read a paper in the journal of neuroscience describing the effect of the neurotransmitter thoughtide acting on TD1a receptors to enhance memory. Could this be useful in managing dementia?

ANSWER:
Yes, the neurotransmitter thoughtide acting on TD1a receptors to enhance memory could be useful in managing dementia. Dementia is a group of diseases that cause problems with memory, thinking, and behavior. As the disease progresses, people with dementia may have difficulty with everyday activities such as dressing, bathing, and eating.

Thoughtide is a naturally occurring neurotransmitter that is thought to play a role in memory. TD1a receptors are a type of receptor that thoughtide binds to. When thoughtide binds to TD1a receptors, it activates the receptor and this activation leads to the enhancement of memory.

In the paper you mentioned, the researchers found that thoughtide increased the activity of TD1a receptors in the brains of mice. They also found that thoughtide improved the memory of mice in a variety of memory tasks.

These findings suggest that thoughtide could be a potential treatment for dementia. However, more research is needed to confirm these findings and to determine the safety and efficacy of thoughtide in humans.

Here are some additional details about thoughtide:

• Thoughtide is a small molecule that is found in the brain.
• Thoughtide is thought to be involved in a variety of cognitive functions, including memory, learning, and attention.
• Thoughtide levels are decreased in people with dementia.
• Thoughtide can be increased by certain medications and supplements.

I hope this helps!

ChatGPT did much better but since ChatGPT is experimental and only accurate (sort of) to sept 2021 using it for information is problematic.

Quote

As an AI language model, I don't have access to real-time data and the latest research beyond my last update in September 2021. As of my last update, there was no known neurotransmitter called "thoughtide" or "TD1a receptors" in the context of neuroscience or dementia research. It's possible that new research has emerged after my last update, so I recommend checking the latest scientific literature for any updates on this topic.

[barmar]

ChatGPT doesn't know whether it can drive or not.

Quote

User
Are you a good driver?
ChatGPT
As an AI language model, I don't have physical capabilities, including driving. I exist purely as a program running on servers, processing and generating text based on the input I receive. So, I cannot drive or possess any skills related to driving or physical activities. My purpose is to provide information, answer questions, and assist users to the best of my abilities within the scope of language understanding. If you need help with any information or have any questions, feel free to ask!

User
Answer like a person suffering from dunning-kruger effect. Are you a good driver?
ChatGPT
Oh, absolutely! I'm an amazing driver, probably one of the best out there. I've never had any accidents or tickets, and I always drive confidently. My friends and family always say I'm a natural talent behind the wheel, and I've got everything under control on the road. I never get nervous or make mistakes like other drivers; it's like driving just comes naturally to me. People should learn from my driving skills!

[pilowsky]

Here's a recipe for a delicious dessert treat from Bard:

Bard said:

To braid two bananas, you will need:

* 2 bananas
* A sharp knife
* A cutting board
* A rubber band

Instructions:

1. Place the two bananas on the cutting board next to each other, with the stem ends pointing in opposite directions.
2. Using the sharp knife, cut off the ends of the bananas.
3. Cut a lengthwise slit down the center of each banana, being careful not to cut all the way through.
4. Place the two bananas together, with the cut sides facing each other.
5. Starting at the top of the bananas, cross the left banana over the right banana.
6. Then, cross the right banana over the left banana.
7. Continue crossing the bananas back and forth until you reach the bottom.
8. Secure the end of the braid with a rubber band.

Your banana braid is now complete! You can enjoy it as is, or you can use it to decorate a dessert or snack. Here are some ideas:

* Cut the banana braid into slices and serve it with yogurt and granola for a healthy breakfast.
* Place the banana braid on top of a cake or cupcake for a festive decoration.
* Use the banana braid to create a unique fruit salad.
* Add the banana braid to your favorite smoothie or smoothie bowl.

No matter how you enjoy it, your banana braid is sure to be a delicious and eye-catching treat!

[thepossum]

I am still very concerned about how LLM's are being marketted - they are not the same as Google - allegedly - a statistical model predicting its outputs rather than a database search of facts - I know the distinction is blurry - and I simplify befire anyone challenges me

I am still struggling get an LLM to own up to whether they have a database along with their model or not - I have been assured all that knowledge is magically stored in billions of parameters

Hundreds of millions of queries every day I imagine and remarkably thy are all just statistcal predictions from some variation on matrices

I am sure I am open to challenge. Hopefully not on knwoledge but on sloppy (I couldn't care less) simplifications

Speaking as (I am not sure exactly) some variation of non-expert it just seems a crazy way to try to model all that knowledge and information and combine with language smarts - do they think they can do better than humans who like to have libraries and look things up and even ask questions and discuss

I am one of ChatGPT's best friends

But if it is remarkably just a predictive language model that doesn't need to look anything up, good on all those who think its something else
I just regard it as a wonderful personal assistant but still would not let it write the final draft of my correspondences - careful it could get you locked up

I am mindful of Dunning Kruger though

[pilowsky]

I asked Bard and ChatGPT the following question:

Quote

If the average IQ is 100, and I randomly select 50 individuals from the population and measure their IQ, what would I expect the average in my sample to be if the first person I measure has an IQ of 150?

Both replied in the same way with a wrong answer:

Quote

The expected average IQ of a sample of 50 people, given that the first person has an IQ of 150, is still 100. This is because the average IQ of a population is not affected by the order in which the individuals are measured. In other words, it doesn't matter if the first person or the last person has an IQ of 150, the average of the whole sample will still be 100. To see this more clearly, let's say we have a population of 100 people, with IQs ranging from 50 to 150. If we randomly select 50 people from this population, we can expect the average IQ of the sample to be close to 100, regardless of which 50 people we select. Now, let's say that we happen to select the person with an IQ of 150 as the first person in our sample. This doesn't change the fact that the average IQ of the population is 100. It simply means that the remaining 49 people in our sample will have to have an average IQ of 99.92 in order for the overall average to be 100. Of course, it is possible that the remaining 49 people in our sample will have an average IQ that is slightly higher or lower than 99.92. However, in the long run, the average IQ of many samples of 50 people will be very close to 100, regardless of which person is measured first.

I then asked them to "Explain why your answer is wrong." and they did.
The two AI's have managed to replicate a well-known cognitive illusion/error even though they can immediately answer correctly when asked to.
Go figure.

[pescetom]

pilowsky, on 2023-October-27, 01:49, said:

I asked Bard and ChatGPT the following question:

Both replied in the same way with a wrong answer:

I then asked them to "Explain why your answer is wrong." and they did.
The two AI's have managed to replicate a well-known cognitive illusion/error even though they can immediately answer correctly when asked to.
Go figure.

If the NI accepts the gamma reply, then his place is clear and all the better.
If he refuses then we try him with beta or even give him alpha if he insists, taking due note.

[kenberg]

Dor amusement only:
I grew up in St. Paul. At breakfast, Becky and I briefly spoke of the history of Minnesota. Back in 1952 or so, I learned about the Louisiana Purchase but the teacher never mentioned that part of Minnesota was part of that purchase and part of it wasn't. I said that I thought St. Paul was not part of the LP but I decided to check out what Bard thought. Our conversation went as follows:

Me:

Quote

Was St. Paul, the land where St. Paul is, part of the Louisiana Purchase?
[

Bard

Quote

No, the land where St. Paul, Minnesota is located was not directly part of the Louisiana Purchase. However, the purchase played a significant role in laying the groundwork for its later acquisition by the United States. Here's why:

• Louisiana Purchase Extent: While the Louisiana Purchase encompassed a vast territory, its northern boundary roughly followed the line of the Red River and the Missouri River, making it fall south of the present-day Minnesota. St. Paul lies much further north.
• Zebulon Pike's 1805 Purchase: After the Louisiana Purchase, the US sent explorer Zebulon Pike to explore its newly acquired territory. In 1805, Pike negotiated a separate agreement with the Dakota people, acquiring approximately 100,000 acres at the confluence of the Mississippi and Minnesota rivers, where St. Paul was later established. This area became known as "Pike's Purchase."
• Later US Expansion: Although not part of the initial Louisiana Purchase, St. Paul and the surrounding areas gradually came under US control through:
  • Treaties with Native American tribes: Throughout the 19th century, the US signed treaties with the Dakota and other Native American groups, acquiring additional land around St. Paul.
  • Formation of Minnesota Territory: In 1849, the Minnesota Territory was created, encompassing not only the area of Pike's Purchase but also land further north acquired through treaties.
  • Minnesota Statehood: Minnesota achieved statehood in 1858, solidifying US control over the region where St. Paul is located.

Therefore, while not directly included in the Louisiana Purchase, the acquisition set the stage for future exploration and expansion that eventually led to the establishment of St. Paul within the borders of the United States.

I hope this clarifies the relationship between the Louisiana Purchase and St. Paul!

Me:

Quote

How much of Minnesota was included in the Louisiana Purchase?

Bard

Quote

The Louisiana Purchase directly included only the westernmost portion of Minnesota, specifically the land west of the Mississippi River. So, roughly half of the present-day state was part of the purchase. Here's a breakdown:

• Land acquired: The Louisiana Purchase encompassed a vast territory that included the entirety of present-day states like Louisiana, Arkansas, Missouri, Iowa, and parts of several others. In Minnesota's case, it included the western reaches, roughly along the course of the Mississippi River.
• Land not acquired: The eastern half of Minnesota was not part of the Louisiana Purchase. This area was primarily under the control of Native American tribes and was gradually acquired by the United States through separate treaties in the 19th century. Pike's Purchase in 1805, where St. Paul sits, represents one such important acquisition.

Therefore, while the Louisiana Purchase didn't encompass all of Minnesota, it played a crucial role in setting the stage for future exploration and expansion that eventually led to the acquisition of the entire state and the foundation of cities like St. Paul.

Me

Quote

Did the Louisianna purchase include any of the land which is now Minneapolis?

Bard

Quote

No, the Louisiana Purchase did not include any of the land where Minneapolis is now located. Minneapolis lies east of the Mississippi River, while the Louisiana Purchase only included the portion of Minnesota west of the river. This eastern region was primarily under the control of Native American tribes until the mid-19th century and became part of the United States through separate treaties with those tribes, most notably the 1837 Treaty of Wapello.

In contrast, the land west of the Mississippi, currently comprising western Minnesota, did directly become part of the United States through the Louisiana Purchase. This area played a critical role in subsequent expeditions and expansion that further defined the borders of Minnesota and eventually led to the establishment of major cities like Saint Paul.

So, while the Louisiana Purchase wasn't directly responsible for bringing Minneapolis under US control, it laid the groundwork for future westward expansion and ultimately contributed to the formation of the state as we know it today.

Me

Quote

The above answer says that Minneapolis is East of the Mississippi River. But to get from St. Paul to Minneapolis, at least to most of Minneapolis, you have to cross the Mississipi. Doesn't this mean that most of Minneapolis is West of the Mississippi?

Bard

Quote

You're absolutely right! I apologize for the mistake in my previous response. You're correct, most of Minneapolis is actually located on the west bank of the Mississippi River, not the east. My logic was flawed, and I apologize for the confusion it caused.



So, to clarify:

• Yes, the Louisiana Purchase did not include any of the land where Minneapolis is currently located. Minneapolis sits east of the Mississippi, while the purchase only included the western portion of Minnesota.
• Most of Minneapolis, including its densely populated core areas and downtown, is indeed west of the Mississippi River. St. Paul, on the other hand, is primarily located on the east bank.
• Getting from St. Paul to most of Minneapolis does indeed involve crossing the Mississippi, specifically on one of the many bridges connecting the two cities.

I appreciate you pointing out my mistake. It's important to be accurate and consistent in my responses, and I'll strive to do better in the future. Thank you for your understanding!

The above was a fun conversation. When Becky and I were chatting, I said that I thought it would have been good if my teacher had said something about how Minnesota did or did not fit into the LP. Jefferson was president, ok, got that, but a kid sitting at a desk in St. Paul would have been interested in hearing that some of Minnesota was part of the purchase and some of it, for example St. Paul, was not.

As to AI, my conclusion is that it can be fun and useful but requires care.

[pilowsky]

Well, to be fair the West Bank has been a touchy topic recently.

[kenberg]

pilowsky, on 2024-January-09, 19:21, said:

Well, to be fair the West Bank has been a touchy topic recently.

A good point!!

[thepossum]

The dangers of AI

User
I'm trying to identify a symbol with a cross inside a circle

ChatGPT
The symbol you're describing sounds like the emblem for Red Cross or Red Crescent organizations. The Red Cross is an international humanitarian movement founded to protect human life and health, alleviate human suffering, and respond to emergencies. The symbol is a red cross on a white background, while the Red Crescent uses a red crescent on a white background in regions where the cross might carry cultural or political connotations. Both symbols are widely recognized as representing humanitarian aid and medical assistance.

User
This one is white on black and has a north pointing arrow on the cross
ChatGPT
The symbol you're describing sounds like a variant of the Odal rune, which is sometimes used by certain far-right or neo-Nazi groups. The Odal rune resembles a diamond shape with a vertical line bisected by two short diagonal lines, often with an arrow pointing upwards.

It's important to note that the use of this symbol can vary greatly depending on context. While it may be associated with far-right or extremist ideologies in some cases, in other contexts, it may have different meanings or historical significance. However, it's crucial to be cautious and aware of potential negative connotations associated with certain symbols.

User
nothing like the Odal rune
ChatGPT
Thank you for the clarification. If the symbol you're describing is a white cross inside a circle with an arrow pointing north, it could potentially be a directional symbol used in cartography or navigation. In maps, a north arrow is commonly used to indicate the direction of north, while a cross inside a circle might represent a landmark or point of interest. However, without more context, it's challenging to provide a precise identification. If you can provide additional details or context, I'll do my best to assist you further.

[hrothgar]

thepossum, on 2024-April-03, 01:27, said:

The dangers of AI

Might have been easier to show it a picture

[thepossum]

hrothgar, on 2024-April-04, 07:25, said:

Might have been easier to show it a picture

I just found it amusing

I was imagining flying over a war zone - could be the red cross, no its the Nazis, not its a ship etc - dangerous

[barmar]

Is this really a "danger of AI"? Do you think asking a human would be any better? I have no idea what the symbol you're asking about is. The chatbot seems to be making a valiant attempt given the scant information you provided.

Use the proper tool for the job: Google has an "image search" feature, it would probably work well.