Malazan Empire: Algorithms and automation - Malazan Empire

Jump to content

  • 4 Pages +
  • « First
  • 2
  • 3
  • 4
  • You cannot start a new topic
  • You cannot reply to this topic

Algorithms and automation

#61 User is offline   Azath Vitr (D'ivers 

  • Ascendant
  • Group: Malaz Regular
  • Posts: 3,645
  • Joined: 07-February 16

Posted 14 August 2025 - 02:56 PM

 HoosierDaddy, on 14 August 2025 - 02:40 PM, said:

I get what you are saying... but I think most people have no CLUE what tools their doctors are using or how they work. Nor do they care. They care about what the doctor tells them.


In the United States and other countries where healthcare is for-profit, some doctors are almost certainly going to be vocal about not using AI in a bid to attract more patients:

Quote

researchers [...] found that U.S. adults were significantly less likely to trust, feel empathy toward or seek care from a physician who advertised using AI — whether for diagnostic, therapeutic or even administrative tasks. [...] in every case, mentioning AI use reduced scores for perceived competence, trustworthiness and empathy. Patients were also less likely to say they would book an appointment.

https://www.medicale...-ai-study-shows


In an ethical and rational society, it would be illegal for doctors to not use AI assistance for applications where it improves medical outcomes.
0

#62 User is offline   HoosierDaddy 

  • Believer
  • Group: Malaz Regular
  • Posts: 8,074
  • Joined: 30-June 08
  • Location:Indianapolis
  • Interests:Football

Posted 14 August 2025 - 03:04 PM

 Azath Vitr (D, on 14 August 2025 - 02:56 PM, said:

 HoosierDaddy, on 14 August 2025 - 02:40 PM, said:

I get what you are saying... but I think most people have no CLUE what tools their doctors are using or how they work. Nor do they care. They care about what the doctor tells them.


In the United States and other countries where healthcare is for-profit, some doctors are almost certainly going to be vocal about not using AI in a bid to attract more patients:

Quote

researchers [...] found that U.S. adults were significantly less likely to trust, feel empathy toward or seek care from a physician who advertised using AI — whether for diagnostic, therapeutic or even administrative tasks. [...] in every case, mentioning AI use reduced scores for perceived competence, trustworthiness and empathy. Patients were also less likely to say they would book an appointment.

https://www.medicale...-ai-study-shows


In an ethical and rational society, it would be illegal for doctors to not use AI assistance for applications where it improves medical outcomes.


Lesson - Don't advertise its use. There's no reason to.

People who trust doctors, trust doctors. Again, AI is a TOOL, and a tool is only as good as the material used to build it and the person whose hands are utilizing it. It isn't doing the diagnosing and I don't want it doing the diagnosing either. I will trust the doctor's discretion far more than any AI.

But this is beyond the point. If they want to use AI to assist, great. But overreliance on any tool is a bad thing, medicine and otherwise.
Trouble arrives when the opponents to such a system institute its extreme opposite, where individualism becomes godlike and sacrosanct, and no greater service to any other ideal (including community) is possible. In such a system rapacious greed thrives behind the guise of freedom, and the worst aspects of human nature come to the fore....
0

#63 User is offline   Azath Vitr (D'ivers 

  • Ascendant
  • Group: Malaz Regular
  • Posts: 3,645
  • Joined: 07-February 16

Posted 17 August 2025 - 05:05 PM

Mark Lawrence conducted an online "AI vs authors" flash fiction blind test again:

Quote

The contributing authors have sold around 15 millions books between them. And they are...

Robin Hobb

Janny Wurts

Christian Cameron / Miles Cameron

& me!

[...] We had 964 votes on the issue of whether story 1 was by a human or AI. This fell fairly smoothly to 474 votes on the rating of story 8. [...] on average the public [guessed] no more effective[ly] than a coin toss!

[...] A sizeable majority of people thought my story was human authored [...]

[... But] the AI scored better than us. Not only was the highest rated story an AI one, but they scored higher on average too.

https://mark---lawre...lts-part-2.html


One obvious possible explanation would be that more people
guessed "human" for Mark Lawrence because they were familiar with his writing---another would be that his story was better than those of the other human authors (he decided not to share how the human-written stories were rated relative to each other, so we can't tell), but one of the AI stories scored strictly higher than any of the human stories...

This post has been edited by Azath Vitr (D'ivers: 17 August 2025 - 05:07 PM

0

#64 User is offline   Abyss 

  • abyssus abyssum invocat
  • Group: Administrators
  • Posts: 22,468
  • Joined: 22-May 03
  • Location:The call is coming from inside the house!!!!
  • Interests:Interesting.

Posted 18 August 2025 - 04:39 AM

I think that the 'flash fiction' format played to the AI's strengths... study a bunch of highly rated short content and reproduce the common points in a slightly different variation.
Had the challenge been 500 page doorstoppers - which is obviously unworkable for many reasons - i suspect the outcome would have been different.
THIS IS YOUR REMINDER THAT THERE IS A
'VIEW NEW CONTENT' BUTTON THAT
ALLOWS YOU TO VIEW NEW CONTENT
0

#65 User is offline   Azath Vitr (D'ivers 

  • Ascendant
  • Group: Malaz Regular
  • Posts: 3,645
  • Joined: 07-February 16

Posted 18 August 2025 - 09:46 AM

 Abyss, on 18 August 2025 - 04:39 AM, said:

I think that the 'flash fiction' format played to the AI's strengths... study a bunch of highly rated short content and reproduce the common points in a slightly different variation.
Had the challenge been 500 page doorstoppers - which is obviously unworkable for many reasons - i suspect the outcome would have been different.


ML (Mark Lawrence, that is, not Machine Learning) does say that LLM aren't as good at long-form narrative. I think a more likely explanation of the difference is that current LLM have a limited context window:

https://www.ibm.com/.../context-window

Here are the stories from the blind test (you can scroll down or use "find on page" to jump to "story 1"):

https://mark---lawre...ood-part-2.html

This post has been edited by Azath Vitr (D'ivers: 18 August 2025 - 09:46 AM

0

Share this topic:


  • 4 Pages +
  • « First
  • 2
  • 3
  • 4
  • You cannot start a new topic
  • You cannot reply to this topic

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users