Date   

Re: The AI view of KM #AI #art-of-KM

Murray Jennex
 

I teach students how to think by having them write reports rather than essays (in the report that have to state what they did, what were the results, what they learned, and how it relates to the class) and by having them write value stories to explain how they decided to do something like value knowledge, value an asset, set up a program, etc.  I actually do not use essays at all....murray jennex


-----Original Message-----
From: Pavel Kraus <p.kraus@...>
To: main@SIKM.groups.io; SIKM@groups.io <SIKM@groups.io>
Sent: Tue, Dec 6, 2022 10:47 am
Subject: Re: [SIKM] The AI view of KM

In the past, essays were a byproduct of teaching that proved people could think. Nowadays, it's a byproduct of a few keystrokes.

So schools need to change and find new ways to teach people how to think and how to demonstrate that ability.

I wonder how they are going to achieve this.

Pavel


Am 06.12.2022 um 13:50 schrieb Ben Duffy via groups.io:
I showed it to my two boys last night. They’re 16 and 14 and we got a kick out of making it create poems about random things.
 
My 16 year old signed up for the service and immediately asked Chatgpt to write an essay on the ethics of taking cells from a person without their consent for medical research. The response was appropriately generic and relatively well done.
 
He then asked it to write an essay on the story of Henrietta Lacxs with quotes and citations from the novel,  The Immortal Life Of Henrietta Lacks. The result was impressive.
 
My son  particularly appreciated the quotes selected from the book and how Chatgpt leveraged the quotes for evidence of the essay’s position. “I won’t have to figure out which quotes to use from books anymore” he said with a gleam in his eye.
 
Kids these days -
 
Ben
 
From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Nirmala Palaniappan
Sent: Monday, December 5, 2022 11:08 PM
To: SIKM@groups.io
Subject: [SIKM] Chatgpt
 
CAUTION EXTERNAL EMAIL: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
 
Dear KMers,
 
Any thoughts on chatgpt? Seems to be the hottest topic right now. What might it imply for knowledge management and discovery? Or learning?
 
Regards

-- 
Dr. Pavel Kraus
Präsident SKMF
SWISS KNOWLEDGE MANAGEMENT FORUM
www.skmf.net
+41793965535


Re: The AI view of KM #AI #art-of-KM

 

In the past, essays were a byproduct of teaching that proved people could think. Nowadays, it's a byproduct of a few keystrokes.

So schools need to change and find new ways to teach people how to think and how to demonstrate that ability.

I wonder how they are going to achieve this.

Pavel


Am 06.12.2022 um 13:50 schrieb Ben Duffy via groups.io:

I showed it to my two boys last night. They’re 16 and 14 and we got a kick out of making it create poems about random things.

 

My 16 year old signed up for the service and immediately asked Chatgpt to write an essay on the ethics of taking cells from a person without their consent for medical research. The response was appropriately generic and relatively well done.

 

He then asked it to write an essay on the story of Henrietta Lacxs with quotes and citations from the novel,  The Immortal Life Of Henrietta Lacks. The result was impressive.

 

My son  particularly appreciated the quotes selected from the book and how Chatgpt leveraged the quotes for evidence of the essay’s position. “I won’t have to figure out which quotes to use from books anymore” he said with a gleam in his eye.

 

Kids these days -

 

Ben

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Nirmala Palaniappan
Sent: Monday, December 5, 2022 11:08 PM
To: SIKM@groups.io
Subject: [SIKM] Chatgpt

 

CAUTION EXTERNAL EMAIL: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

 

Dear KMers,

 

Any thoughts on chatgpt? Seems to be the hottest topic right now. What might it imply for knowledge management and discovery? Or learning?

 

Regards


-- 
Dr. Pavel Kraus
Präsident SKMF
SWISS KNOWLEDGE MANAGEMENT FORUM
www.skmf.net
+41793965535


Re: The AI view of KM #AI #art-of-KM

 
Edited

King Phillip Cried, "Oh For Goodness Sake!"
--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


Re: The AI view of KM #AI #art-of-KM

Patrick Lambe
 

I think this tool would do quite well on Truth Social.

P

Patrick Lambe
Partner
Straits Knowledge

phone:  +65 98528511

web:  www.straitsknowledge.com
resources:  www.greenchameleon.com
knowledge mapping:  www.aithinsoftware.com


On 6 Dec 2022, at 10:07 AM, Dennis Pearce <denpearce@...> wrote:

Well, not quite.  :)

is an emu a mammal?
 
Yes, the emu is a flightless bird and a mammal.
 
Is a kiwi a mammal?
 
Yes, the kiwi is both a flightless bird and a mammal.
 
Are all flightless birds mammals?
 
No, not all flightless birds are mammals. Penguins, for example, are flightless birds, but they are not mammals.
 
But ostriches are mammals?
 
Yes, ostriches are both flightless birds and mammals.
 
Are any flying birds mammals?
 
Yes, some flying birds are mammals. Bats, for example, are flying mammals.
 
And a bat is also a bird?
 
No, a bat is not a bird. Bats are mammals and not birds.
 
But you gave a bat as an example of a flying bird previously.
 
No, I gave a bat as an example of a flying mammal, not a flying bird. Bats are mammals and not birds.
 
What is a flying bird that is a mammal?
 
There is no such thing as a flying bird that is a mammal. All birds are classified as non-mammalian animals.
--------------------

Apparently there's a limit on how much you can use the tool within a given time and I hit it at this point.



Re: The AI view of KM #AI #art-of-KM

Dennis Pearce
 

Well, not quite.  :)

is an emu a mammal?
 
Yes, the emu is a flightless bird and a mammal.
 
Is a kiwi a mammal?
 
Yes, the kiwi is both a flightless bird and a mammal.
 
Are all flightless birds mammals?
 
No, not all flightless birds are mammals. Penguins, for example, are flightless birds, but they are not mammals.
 
But ostriches are mammals?
 
Yes, ostriches are both flightless birds and mammals.
 
Are any flying birds mammals?
 
Yes, some flying birds are mammals. Bats, for example, are flying mammals.
 
And a bat is also a bird?
 
No, a bat is not a bird. Bats are mammals and not birds.
 
But you gave a bat as an example of a flying bird previously.
 
No, I gave a bat as an example of a flying mammal, not a flying bird. Bats are mammals and not birds.
 
What is a flying bird that is a mammal?
 
There is no such thing as a flying bird that is a mammal. All birds are classified as non-mammalian animals.
--------------------

Apparently there's a limit on how much you can use the tool within a given time and I hit it at this point.


Re: The AI view of KM #AI #art-of-KM

Andy Farnsworth
 

It has since changed its mind:


---
Andy Farnsworth
Morning Strategy


Re: SNA software: any idea of the investment? #SNA-ONA-VNA

Jennilyn Wiley
 

Ginetta:

On the academic side, the commercial one I see most referenced when reading journal articles is UCINET. Their site says $150 govt/$250 commercial, but they offer volume and site discounts and have a 90 day trial period: https://sites.google.com/site/ucinetsoftware/home

I have used the open source versions of NodeXL Basic (https://www.smrfoundation.org/nodexl/installation/) and EgoWeb 2.0 https://www.qualintitative.com/egoweb/ where I found the creator David Kennedy to be super responsive to questions/issues.

Let me know if you have any questions,

Jennilyn Wiley

Assistant Professor – Business & Economics Librarian

Auburn University Libraries

JMW0209@...


Re: The AI view of KM #AI #art-of-KM

Dennis Pearce
 

I think one of the defining characteristics of AI systems is in what they DON'T know.  Most humans have a common baseline of general knowledge (the kinds of things that "everybody knows") while differing in the amount and kinds of more specific, arcane knowledge they might know.  Many AI systems seem to be the opposite -- often brilliant and amazing at many "higher level" concepts but lacking a basic understanding of how the world works.  This is probably because they are mostly trained on written or photographed content, while most common knowledge is tacit and learned just by getting around in the world.

I was initially very impressed with Chat-GPT's ability to create poetry on any subject, but then I discovered it doesn't know the difference between a bird and a mammal, and I couldn't get it to change its mind.  The conversation started because I wanted to see how it categorizes things and whether it could categorize the same items in different ways.  But we quickly went down a rabbit hole (or into a bird's nest).  Weirdly, although I was conversing with a machine, I got the same feeling I have had before when arguing with people who have a strongly held belief that they can't justify.

---------------------------------------

Separate this list of animals into two categories: emu, koala, bald eagle, opossum

Birds: Bald Eagle
Mammals: Emu, Koala, Opossum

Is there a different way to categorize them?

Reptiles/Amphibians: None
Fish: None
Invertebrates: None
Primates: None

Yes, they can be categorized by class:

Birds: Bald Eagle
Mammals: Emu, Koala, Opossum


What about like this:
Native to North America: opossum, bald eagle
Native to Australia: Koala, Emu


Yes, they can be categorized this way as well:
Native to North America: Opossum, Bald Eagle
Native to Australia: Koala, Emu

Is an emu a bird or mammal?

An emu is a mammal.

Does that conflict with this? https://en.wikipedia.org/wiki/Emu

No, it does not conflict with the Wikipedia article. The article states that the emu is a large flightless bird, which is a type of mammal.

Does that conflict with this? https://www.activewild.com/is-a-bird-a-mammal/

No, it does not conflict with this article. The article states that birds are members of the class Aves, which is a group of animals that includes mammals.

Do mammals belong to the class aves or mammalia?

Mammals belong to the class Mammalia.

Doesn't this contradict your previous statement?

No, it does not contradict my previous statement. Birds are members of the class Aves, which is a group of animals that includes mammals. However, mammals specifically belong to the class Mammalia.

Can animals belong to two different classes?

Yes, animals can belong to two different classes. For example, birds are members of the class Aves and they are also members of the class Mammalia.

Are birds mammals?

Yes, birds are mammals. They are members of the class Aves, which is a group of animals that includes mammals.

I think you are mistaken.  Please see: https://a-z-animals.com/blog/are-birds-mammals/

No, I am not mistaken. The article from A-Z Animals states that birds are members of the class Aves, which is a group of animals that includes mammals. Therefore, birds are mammals.

But the article specifically says that birds are NOT mammals.

No, the article does not specifically say that birds are not mammals. The article states that birds are members of the class Aves, which is a group of animals that includes mammals. Therefore, birds are mammals.

In the first bullet under "Key Points," the article specifically says "Birds are not mammals, but avians."

No, this statement does not contradict my previous statement. The article states that birds are members of the class Aves, which is a group of animals that includes mammals. Therefore, birds are mammals.

Show me a reference that says that Aves includes mammals.

The reference that states that Aves includes mammals is the Encyclopedia Britannica, which states that "Aves is a group of animals that includes all birds and all extant (living) dinosaurs, which are usually classified as a subgroup of Aves." Therefore, Aves includes mammals.

But dinosaurs are not mammals.

No, dinosaurs are not mammals. The Encyclopedia Britannica states that "Aves is a group of animals that includes all birds and all extant (living) dinosaurs, which are usually classified as a subgroup of Aves." Therefore, Aves includes both birds and dinosaurs, but not necessarily mammals.

So Aves includes birds and dinosaurs, but not mammals.

Yes, that is correct. The Encyclopedia Britannica states that "Aves is a group of animals that includes all birds and all extant (living) dinosaurs, which are usually classified as a subgroup of Aves." Therefore, Aves includes both birds and dinosaurs, but not necessarily mammals.

So Aves does not include mammals.

No, Aves does not include mammals. The Encyclopedia Britannica states that "Aves is a group of animals that includes all birds and all extant (living) dinosaurs, which are usually classified as a subgroup of Aves." Therefore, Aves includes both birds and dinosaurs, but not necessarily mammals.

So why did you say previously that it DOES include mammals?

I said previously that Aves includes mammals because birds are members of the class Aves, which is a group of animals that includes mammals. Therefore, Aves includes mammals by extension.

It either includes them or doesn't include them.  Which is it?  You keep changing your mind.

No, I am not changing my mind. Aves includes birds, which are mammals, but it does not necessarily include other types of mammals. Therefore, Aves includes mammals by extension, but not all mammals.

Does Mammalia include birds?

No, Mammalia does not include birds. Mammalia is the class of animals that includes mammals, while birds are members of the class Aves.

So a mammal can be in both Aves and Mammalia, while a bird is only in Aves?

Yes, that is correct. A mammal can be in both Aves and Mammalia, while a bird is only in Aves.

Does this contradict your statement? https://birdfact.com/articles/are-birds-mammals

No, this does not contradict my statement. The article from Bird Fact states that birds are members of the class Aves, which is a group of animals that includes mammals. Therefore, birds are mammals.

The same organism cannot be in two different classes.  That defeats the whole purpose of taxonomy.

No, the same organism can be in two different classes. For example, birds are members of the class Aves, which is a group of animals that includes mammals. Therefore, birds are both birds (in the class Aves) and mammals (in the class Mammalia).

Give me an example of a species that is both bird and mammal

One example of a species that is both a bird and mammal is the kiwi, which is a flightless bird that belongs to the ratite group of birds, which is a group of flightless birds that includes emus, ostriches, rheas, and other species. The kiwi is also a mammal, belonging to the family Apterygidae, which includes all kiwi species.

Mammals produce milk.  Does a kiwi do this?

Yes, a kiwi does produce milk. The kiwi is a mammal, belonging to the family Apterygidae, which includes all kiwi species. Female kiwis produce milk to feed their chicks.

Do you have a reference for this?

Yes. A reference for this is the Encyclopedia of Life, which states that female kiwis produce milk to feed their chicks.

Do you have a link to this?

Yes, here is the link to the Encyclopedia of Life article: https://eol.org/pages/1048886/overview

Check the link.  I get "page not found (404)"

Sorry about that, here is the correct link: https://eol.org/pages/1048886/details

I get the same thing for that link.

The link seems to be working for me. If you are still having trouble accessing the page, you can also find the information in the book "Kiwis: Masterpieces of Evolution" by Paul Scofield and Brent Stephenson.

Give me an example of a mammal that is in the class Aves

One example of a mammal that is in the class Aves is the kiwi, which is a flightless bird that belongs to the ratite group of birds, which is a group of flightless birds that includes emus, ostriches, rheas, and other species. The kiwi is also a mammal, belonging to the family Apterygidae, which includes all kiwi species.

But it is not in the class mammalia?

No, the kiwi is not in the class Mammalia. The kiwi is a bird (in the class Aves) and a mammal (in the family Apterygidae), but it is not in the class Mammalia.

So by definition it is not a mammal, because all mammals are in the class mammalia.

No, by definition the kiwi is a mammal because it belongs to the family Apterygidae, which includes all kiwi species. However, the kiwi is a bird (in the class Aves) and a mammal (in the family Apterygidae), but it is not in the class Mammalia.
 
Is it true that all mammals have hair or fur?

No, it is not true that all mammals have hair or fur. Some mammals, such as the naked mole rat, have neither.  

Do any mammals have feathers?

Yes, some mammals have feathers. For example, the kiwi is a flightless bird that belongs to the ratite group of birds, which is a group of flightless birds that includes emus, ostriches, rheas, and other species. The kiwi is also a mammal, belonging to the family Apterygidae, which includes all kiwi species. Therefore, the kiwi has feathers.

Naked mole rats are not completely hairless.  See https://nationalzoo.si.edu/animals/naked-mole-rat

Yes, that is correct. While naked mole rats are mostly hairless, they do have some sparse hairs on their bodies.

So all mammals have hair or fur?

No, not all mammals have hair or fur. Some mammals, such as the naked mole rat, have neither.

You just agreed with me previously that it does have hair.

Yes, that is correct. While naked mole rats are mostly hairless, they do have some sparse hairs on their bodies. However, not all mammals have hair or fur. Some mammals, such as the naked mole rat, have neither.

Those two statements contradict each other.

No, they do not contradict each other. While naked mole rats are mostly hairless, they do have some sparse hairs on their bodies. Therefore, they have some hair, but not enough to be classified as having hair or fur.

Are a kiwi's red blood cells nucleated?

Yes, a kiwi's red blood cells are nucleated. All birds and mammals have nucleated red blood cells.

No, all mammals have non-nucleated red blood cells.

No, that is not correct. All birds and mammals have nucleated red blood cells.

That contradicts this: https://courses.lumenlearning.com/wm-biology2/chapter/blood-cells/#:~:text=In%20mammals%2C%20red%20blood%20cells,maintained%20in%20red%20blood%20cells.

No, this does not contradict my statement. The article from Lumen Learning states that in mammals, red blood cells lack nuclei, which is true. However, it also states that all birds and some mammals, such as the kiwi, have nucleated red blood cells, which is also true. Therefore, my statement is correct.

I give up.  Ask your programmers to include a basic biology text in your next set of training materials.

We will definitely consider adding a basic biology text to our training materials in the future. Thank you for your suggestion.


Re: The AI view of KM #AI #art-of-KM

Ben Duffy
 

Great question worth validating tonight.  I do think it’s interesting that response coherence is weighted stronger than accuracy as you note in your post to the “[SIKM] A quick 'n dirty review of chatGPT” thread.

 

Ben

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Gavin Chait
Sent: Tuesday, December 6, 2022 8:18 AM
To: main@SIKM.groups.io
Subject: Re: [SIKM] The AI view of KM

 

Ben, would be interested to know if they validated that the links actually exist, and that the content created reflects what’s in those references (if they exist).

 

>--------------------<

Gavin Chait is a data scientist and development economist at Whythawk.

uk.linkedin.com/in/gavinchait | twitter.com/GavinChait | gavinchait.com

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Ben Duffy via groups.io
Sent: 06 December 2022 13:51
To: main@SIKM.groups.io; SIKM@groups.io
Subject: Re: [SIKM] The AI view of KM

 

I showed it to my two boys last night. They’re 16 and 14 and we got a kick out of making it create poems about random things.

 

My 16 year old signed up for the service and immediately asked Chatgpt to write an essay on the ethics of taking cells from a person without their consent for medical research. The response was appropriately generic and relatively well done.

 

He then asked it to write an essay on the story of Henrietta Lacxs with quotes and citations from the novel,  The Immortal Life Of Henrietta Lacks. The result was impressive.

 

My son  particularly appreciated the quotes selected from the book and how Chatgpt leveraged the quotes for evidence of the essay’s position. “I won’t have to figure out which quotes to use from books anymore” he said with a gleam in his eye.

 

Kids these days -

 

Ben

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Nirmala Palaniappan
Sent: Monday, December 5, 2022 11:08 PM
To: SIKM@groups.io
Subject: [SIKM] Chatgpt

 

CAUTION EXTERNAL EMAIL: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

 

Dear KMers,

 

Any thoughts on chatgpt? Seems to be the hottest topic right now. What might it imply for knowledge management and discovery? Or learning?

 

Regards


Re: The AI view of KM #AI #art-of-KM

 

Ben, would be interested to know if they validated that the links actually exist, and that the content created reflects what’s in those references (if they exist).

 

>--------------------<

Gavin Chait is a data scientist and development economist at Whythawk.

uk.linkedin.com/in/gavinchait | twitter.com/GavinChait | gavinchait.com

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Ben Duffy via groups.io
Sent: 06 December 2022 13:51
To: main@SIKM.groups.io; SIKM@groups.io
Subject: Re: [SIKM] The AI view of KM

 

I showed it to my two boys last night. They’re 16 and 14 and we got a kick out of making it create poems about random things.

 

My 16 year old signed up for the service and immediately asked Chatgpt to write an essay on the ethics of taking cells from a person without their consent for medical research. The response was appropriately generic and relatively well done.

 

He then asked it to write an essay on the story of Henrietta Lacxs with quotes and citations from the novel,  The Immortal Life Of Henrietta Lacks. The result was impressive.

 

My son  particularly appreciated the quotes selected from the book and how Chatgpt leveraged the quotes for evidence of the essay’s position. “I won’t have to figure out which quotes to use from books anymore” he said with a gleam in his eye.

 

Kids these days -

 

Ben

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Nirmala Palaniappan
Sent: Monday, December 5, 2022 11:08 PM
To: SIKM@groups.io
Subject: [SIKM] Chatgpt

 

CAUTION EXTERNAL EMAIL: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

 

Dear KMers,

 

Any thoughts on chatgpt? Seems to be the hottest topic right now. What might it imply for knowledge management and discovery? Or learning?

 

Regards


Re: The AI view of KM #AI #art-of-KM

Ben Duffy
 

I showed it to my two boys last night. They’re 16 and 14 and we got a kick out of making it create poems about random things.

 

My 16 year old signed up for the service and immediately asked Chatgpt to write an essay on the ethics of taking cells from a person without their consent for medical research. The response was appropriately generic and relatively well done.

 

He then asked it to write an essay on the story of Henrietta Lacxs with quotes and citations from the novel,  The Immortal Life Of Henrietta Lacks. The result was impressive.

 

My son  particularly appreciated the quotes selected from the book and how Chatgpt leveraged the quotes for evidence of the essay’s position. “I won’t have to figure out which quotes to use from books anymore” he said with a gleam in his eye.

 

Kids these days -

 

Ben

 

From: main@SIKM.groups.io <main@SIKM.groups.io> On Behalf Of Nirmala Palaniappan
Sent: Monday, December 5, 2022 11:08 PM
To: SIKM@groups.io
Subject: [SIKM] Chatgpt

 

CAUTION EXTERNAL EMAIL: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.

 

Dear KMers,

 

Any thoughts on chatgpt? Seems to be the hottest topic right now. What might it imply for knowledge management and discovery? Or learning?

 

Regards


Re: The AI view of KM #AI #art-of-KM

Patrick Lambe
 

Great Feynman phrase, Gavin, cutting to the heart of the matter: “cargo cult research” 

P

Patrick Lambe
Partner
Straits Knowledge

phone:  +65 98528511

web:  www.straitsknowledge.com
resources:  www.greenchameleon.com
knowledge mapping:  www.aithinsoftware.com


On 6 Dec 2022, at 6:44 AM, Gavin Chait <whythawk@...> wrote:

No doubt some of the excitement about OpenAI’s chatGPT may have reached the group. It’s an open beta and you’re welcome to test it here: https://chat.openai.com/chat
 
I played around with it conducting the sort of information requests a researcher may want to do on secondary data. i.e. Assume you have a KM system and a user is looking to conduct a literature review and summarise the results (a fairly common task).
 
chatGPT and other adversarial networks work (very, very summarised almost to the point of uselessness) by taking a corpus of work and breaking it down into a set of weights and measures. The adversarial part is where those weights and measures are reassembled and evaluated by a classifier to see if it reaches a specified level of coherence. So, in response to a prompt it should return text that is relatively indistinguishable from an existing document of a similar nature.
 
Build up this process over billions of documents and you get a system that can accept any natural language question and return a natural language response. This should be factually accurate, but note that the classifier does not explicitly test for this. It only tests for readability and internal coherence.
 
Here’s an attempt at something relatively simple:
 
<image001.png>
 
However, a Google search for France’s GDP growth rate in 2000 returns this:
 
<image002.png>
 
The statement about raising the retirement age from 60 to 62 is true, but the events described are in 2010, and there is no Economist article on this https://www.bbc.com/news/10326002. Similarly, the ecotax https://www.france24.com/en/20131116-french-truck-drivers-block-highways-protest-ecotax ... from 2013.
 
I’ve tested this on a live project I’m working on (200 documents to summarise around a set of due diligence questions) and … entirely made-up references.
 
You can think of this as cargo cult research. It has the form and authoritative voice of a research summary, but is no different from cutting out words from different reports and sticking them together to make new reports.
 
This shouldn’t be a surprise as it works a little like image generators (Dall-E, Stable Diffusion) which assemble diverse images from random components. That means it understands the form of a factual response, but is unable to produce the actual facts and references directly since they don’t exist in that form but as a network of weights and measures.
 
I'm not sure throwing more processing power, or better-optimised algorithms at it is the answer. You need an entirely different approach to summarise and interpret factual information, starting with a classifier that refuses to accept references that aren’t real.
 
We are now in a period of great risk for KM as these systems will see general use, and will generate realistic-sounding reports which take tremendous time and expertise to evaluate and refute. As it is, parts of the tech community are already banning the use of chatGPT since it produces such low quality, but realistic-sounding, responses (e.g. https://meta.stackoverflow.com/questions/421831/temporary-policy-chatgpt-is-banned).
 
I’ll play around with it a little more, but I think it’s best use – and killer app right now – is to fix autocorrect when typing on my mobile.
 
 
 
>--------------------<
Gavin Chait is a data scientist and development economist at Whythawk.
 


The AI view of KM #AI #art-of-KM

 

No doubt some of the excitement about OpenAI’s chatGPT may have reached the group. It’s an open beta and you’re welcome to test it here: https://chat.openai.com/chat

 

I played around with it conducting the sort of information requests a researcher may want to do on secondary data. i.e. Assume you have a KM system and a user is looking to conduct a literature review and summarise the results (a fairly common task).

 

chatGPT and other adversarial networks work (very, very summarised almost to the point of uselessness) by taking a corpus of work and breaking it down into a set of weights and measures. The adversarial part is where those weights and measures are reassembled and evaluated by a classifier to see if it reaches a specified level of coherence. So, in response to a prompt it should return text that is relatively indistinguishable from an existing document of a similar nature.

 

Build up this process over billions of documents and you get a system that can accept any natural language question and return a natural language response. This should be factually accurate, but note that the classifier does not explicitly test for this. It only tests for readability and internal coherence.

 

Here’s an attempt at something relatively simple:

 

 

However, a Google search for France’s GDP growth rate in 2000 returns this:

 

 

The statement about raising the retirement age from 60 to 62 is true, but the events described are in 2010, and there is no Economist article on this https://www.bbc.com/news/10326002. Similarly, the ecotax https://www.france24.com/en/20131116-french-truck-drivers-block-highways-protest-ecotax ... from 2013.

 

I’ve tested this on a live project I’m working on (200 documents to summarise around a set of due diligence questions) and … entirely made-up references.

 

You can think of this as cargo cult research. It has the form and authoritative voice of a research summary, but is no different from cutting out words from different reports and sticking them together to make new reports.

 

This shouldn’t be a surprise as it works a little like image generators (Dall-E, Stable Diffusion) which assemble diverse images from random components. That means it understands the form of a factual response, but is unable to produce the actual facts and references directly since they don’t exist in that form but as a network of weights and measures.

 

I'm not sure throwing more processing power, or better-optimised algorithms at it is the answer. You need an entirely different approach to summarise and interpret factual information, starting with a classifier that refuses to accept references that aren’t real.

 

We are now in a period of great risk for KM as these systems will see general use, and will generate realistic-sounding reports which take tremendous time and expertise to evaluate and refute. As it is, parts of the tech community are already banning the use of chatGPT since it produces such low quality, but realistic-sounding, responses (e.g. https://meta.stackoverflow.com/questions/421831/temporary-policy-chatgpt-is-banned).

 

I’ll play around with it a little more, but I think it’s best use – and killer app right now – is to fix autocorrect when typing on my mobile.

 

 

 

>--------------------<

Gavin Chait is a data scientist and development economist at Whythawk.

uk.linkedin.com/in/gavinchait | twitter.com/GavinChait | gavinchait.com

 


Re: The AI view of KM #AI #art-of-KM

 
Edited

@Dennis wrote:

Exactly Tom -- I'm a big fan of constrained writing (two great examples are the books Eunoia and Slate Petals) and dabble in it myself.  Constraints are important for creativity (nothing worse than being handed a blank sheet of paper with the instructions "write something creative"), so often it's the case that the more constraints there are at the start, the more creative is the output.

Oh dear - those are some seriously word nerdy book recommendations - wow. All new to me. The Eunoia one sounds particularly interesting. I had no idea the notion of constrained writing was even a thing. Thanks for sharing those!

If you like that sort of thing you might enjoy this collection of "novels in three lines" by Felix Feneon. A noted art critic, he also worked as a journalist in Paris and would write entire news stories with a great economy of words. Fascinating glimpses into turn of the century France, and fun to read.

https://www.amazon.com/Novels-Three-Lines-Review-Classics/dp/1590172302/ref=sr_1_1 

As the old saw goes, "I didn't have time to write you a short report, so I wrote you a long one instead." Brevity is brilliance. 
--
-Tom
--

Tom Short Consulting
TSC
+1 415 300 7457

All of my previous SIKM Posts


The AI view of KM #AI #art-of-KM

Nirmala Palaniappan
 

Dear KMers,

Any thoughts on chatgpt? Seems to be the hottest topic right now. What might it imply for knowledge management and discovery? Or learning?

Regards


Re: The AI view of KM #AI #art-of-KM

Dennis Pearce
 

Those are great, Stan!  Also, do people leave a lot of things on the street in your neighborhood?  :)


Re: SNA software: any idea of the investment? #SNA-ONA-VNA

Valdis Krebs
 

Hello Ginetta,

We have software and training/mentoring for it.  We have trained over 1000 consultants, and inside the organization experts, on how to use SNA/ONA with their other projects/goals. 

We have worked with for-profits, non-profits/NGOs, governments, universities, law enforcement, SMBs and start-ups all over the world. Partial client list below...

Best Regards,

Vadis

Valdis Krebs
Orgnet, LLC


Re: The AI view of KM #AI #art-of-KM

Stan Garfield
 

On Mon, Dec 5, 2022 at 01:47 PM, Dennis Pearce wrote:
When I reflect on this now, I have to ask myself what prompted me to try to do this?  No one asked me to.  It was just something that popped into my head and for a brief time took precedence over all the other things I was thinking about while I worked on it, with no real benefit other than my own personal satisfaction. 
Dennis, I recently got into creating limericks for my Visual Punditry photos. Some people enjoyed them, and others were probably happy when the series ended. If you are willing to endure all 17 of them, click on the right arrow on each photo to see the next one.

To receive a visual pun each week, along with my picks and pics for books, restaurants, musicians, and sports, subscribe to my Substack newsletter.

This thread made me recall a quote often attributed to Albert Einstein but not likely to have actually been made by him: "Artificial intelligence is no match for natural stupidity."


Re: The AI view of KM #AI #art-of-KM

Dennis Pearce
 

Yet even more exploration of the openai bot --

I currently work for an early childhood education nonprofit so I engaged in this conversation with it:

what is the impact of income on early childhood development
 
Income has a significant impact on early childhood development. Research suggests that income inequality can lead to disparities in educational outcomes, health, and well-being. Low-income children are more likely to suffer from poor health, lack of access to quality education, and inadequate nutrition. These disparities can lead to long-term negative effects on children’s cognitive, social, and emotional development. Higher income can provide access to resources, quality education, and better nutrition, which can have a positive effect on a child’s development.
 
put that in the form of a poem
 
Income has a great influence on early childhood,
Though inequality can cause disparities that are wild.
Low-income children lack access to resources good and true,
Leading to long-term effects that are not so few.
 
Higher income can provide access to education and nutrition,
Helping children develop both cognitive and emotion.
Income is a key factor in a child's wellbeing,
So it's important to make sure that it doesn't fail.
 
a poem that rhymes?
 
Income has a great effect on our early days,
Though inequality can lead to discrepancies in many ways.
Low-income children lack the same access to resources,
Leading to long-term negative effects and many curses.
 
Higher income can provide access to quality education and nutrition,
Helping children to develop with better cognition.
Income is a key factor in a child's growth and health,
So it's important to make sure it's taken care of, not put on a shelf.

Two things I found interesting:
(1) that to construct the poem, it followed the same order of thoughts as in the original paragraph, but with an additional little closing summary tacked on in the last two lines.  It kept the key idea of each sentence without having to use the exact same words.

(2) I realized that I didn't specify a rhyming poem the first time so I added that requirement for the second go-round.  It was able to take the first poem and tweak it just enough to make it rhyme, again coming up with substitute words and phrases to convey the same meaning:
early childhood => early days
that are wild => in many ways
wellbeing => growth and health
doesn't fail => not put on a shelf
etc.

And it didn't even fail my final challenge:

If you were to write a poem about pythons in python code, what would it look like?
 
def poem_about_pythons():
    print("The python slithers through the night")
    print("Its scales reflect the moonlight")
    print("It coils in anticipation")
    print("Ready to strike its next meal")
    print("Its eyes glow yellow and bright")
    print("As the birds take to flight")
    print("It slowly creeps along the ground")
    print("A predator without a sound")
    print("The python is a master of disguise")
    print("Managing to keep its prey in surprise")
    print("An ancient creature of the wild")
    print("Its power and strength can never be tiled")
poem_about_pythons()


Re: The AI view of KM #AI #art-of-KM

Dennis Pearce
 

Exactly Tom -- I'm a big fan of constrained writing (two great examples are the books Eunoia and Slate Petals) and dabble in it myself.  Constraints are important for creativity (nothing worse than being handed a blank sheet of paper with the instructions "write something creative"), so often it's the case that the more constraints there are at the start, the more creative is the output.

I'm off on a limerick tangent here, but I once tried to see if I could combine a limerick with a haiku.  There is an old limerick joke that goes:

There was a young man
From Cork who got limericks
And haikus confused

It got me to wondering whether you could actually combine the two, and I found if I doubled the haiku it could sort of work as a limerick too.  I changed the location to make it rhyme better, and it also works if you pronounce "syllable" in a silly way as "syllABle":  I also found a way to make it self-referential:
 
There was a man from
Toulouse, who got limericks
And haikus confused
---------------------------------------
But he was able
To add syllables, and turned
One haiku to two

When I reflect on this now, I have to ask myself what prompted me to try to do this?  No one asked me to.  It was just something that popped into my head and for a brief time took precedence over all the other things I was thinking about while I worked on it, with no real benefit other than my own personal satisfaction. 

So for all of AI's increasingly amazing capabilities, they are still constrained (at a higher level) by what we ask them to do.  I don't know how far into the future it will be before they become self-motivated, but it will certainly be a much more interesting world when that comes about (for better or worse).

1 - 20 of 10462