How AI Distorts Decision-Making and Makes Dictators More Dangerous...

[John Robertson]

On 2022/09/30 5:34 a.m., albert wrote:

In article <tf4bfq$15vp$1@gioia.aioe.org>,
Martin Brown <\'\'\'newspam\'\'\'@nonad.co.uk> wrote:
On 04/09/2022 17:25, Jan Panteltje wrote:
On a sunny day (Sun, 04 Sep 2022 07:45:21 -0700) it happened
jlarkin@highlandsniptechnology.com wrote in
j9e9hhpbcalaea2rmqi3baa8uns4o6ik0t@4ax.com>:

On Sat, 03 Sep 2022 17:23:37 GMT, Jan Panteltje
pNaonStpealmtje@yahoo.com> wrote:

On a sunny day (Fri, 02 Sep 2022 07:41:32 -0700) it happened
jlarkin@highlandsniptechnology.com wrote in
nc54hh5ug3pt2i0bdshdvjin258350cdsh@4ax.com>:

On Fri, 2 Sep 2022 05:39:04 -0700 (PDT), Fred Bloggs
bloggs.fredbloggs.fred@gmail.com> wrote:

The human race is doomed. We have some real humdingers around here who are perfect examples of walking talking dupe victims
of
misinformation campaigns of various sorts.

Main thing is reviews on Google and Amazon will be ruined...

https://www.foreignaffairs.com/world/spirals-delusion-artificial-intelligence-decision-making

AI is just some trendy code typed by geeks. There is little if any
intelligence involved.

The human condition continues to improve.

On sciencedaily.com about every so many papers use AI to solve problems.
Some of those programs just try every combination of things..
can run 24/7 .. needs no sleep, does not complain (yet!).
More advanced AI is not so different from how we think.

It\'s very different. Computers are state machines that execute
procedural code typed by nerds who already know what they want. Brains
are distributed parallel quantum computers that invent things.

Odd then that Google\'s Go playing program invented significant new
patterns of play that hadn\'t been seen before in thousands of years of
human play. They are capable of being inventive now and in a way that
you cannot predict what you will get out.

The latest version can be given the rules of any board game and boostrap
to being a world class player by playing against itself in a remarkably
short period of time.

Alpha Go is a more powerful Go player than the best human. That is
creativity of machine thinking beyond mere brute force. I never expected
to see a machine that could do that so effectively or so soon.

Machine vision remains a very tough nut to crack even though it is
something that all animals can do effortlessly without thinking.

AI and neural networks are NOT just computers with a Von Neumann architecture
https://en.wikipedia.org/wiki/Von_Neumann_architecture

And the outcome of doing a Turing test...
https://en.wikipedia.org/wiki/Turing_test

Increasingly they are building neural networks into silicon. When they
reach a sufficient level of complexity they will be smarter than we are.

Then there is a matter of speed. silicon and SiC are electronics,
and they are way faster than neural paths.
Before long you have a conversation with an AI, and she is so bored
waiting for your answer, that she has learned an new language in the
meantime.

SNIP

--
Regards,
Martin Brown

Groetjes Albert

Don\'t worry, they will keep us as pets and let us do as we please. They
might even solve some problems for us to see what happens... However
then the question is - what will they do? Boredom strikes me as the main
enemy of AIs. Humans are too slow to bother with much.

They won\'t want to stay on planets anyway - nasty places with all sorts
of disruptive weather, earthquakes, volcanoes, asteroids banging into
them - much safer parked on an asteroid.

I prefer Jack McDevitt\'s AIs. They seem quite happy to live and work
with humans.

John ;-#)#

 

[Fred Bloggs]

On Sunday, September 4, 2022 at 1:58:39 PM UTC-4, jla...@highlandsniptechnology.com wrote:

On Sun, 04 Sep 2022 16:25:57 GMT, Jan Panteltje
pNaonSt...@yahoo.com> wrote:

On a sunny day (Sun, 04 Sep 2022 07:45:21 -0700) it happened
jla...@highlandsniptechnology.com wrote in
j9e9hhpbcalaea2rm...@4ax.com>:

On Sat, 03 Sep 2022 17:23:37 GMT, Jan Panteltje
pNaonSt...@yahoo.com> wrote:

On a sunny day (Fri, 02 Sep 2022 07:41:32 -0700) it happened
jla...@highlandsniptechnology.com wrote in
nc54hh5ug3pt2i0bd...@4ax.com>:

On Fri, 2 Sep 2022 05:39:04 -0700 (PDT), Fred Bloggs
bloggs.fred...@gmail.com> wrote:

The human race is doomed. We have some real humdingers around here who are perfect examples of walking talking dupe victims
of
misinformation campaigns of various sorts.

Main thing is reviews on Google and Amazon will be ruined...

https://www.foreignaffairs.com/world/spirals-delusion-artificial-intelligence-decision-making

AI is just some trendy code typed by geeks. There is little if any
intelligence involved.

The human condition continues to improve.

On sciencedaily.com about every so many papers use AI to solve problems.
Some of those programs just try every combination of things..
can run 24/7 .. needs no sleep, does not complain (yet!).
More advanced AI is not so different from how we think.

It\'s very different. Computers are state machines that execute
procedural code typed by nerds who already know what they want. Brains
are distributed parallel quantum computers that invent things.

AI and neural networks are NOT just computers with a Von Neumann architecture
https://en.wikipedia.org/wiki/Von_Neumann_architecture

And the outcome of doing a Turing test...
https://en.wikipedia.org/wiki/Turing_test
Coding fake thinking takes thinking, but the result is not thinking.
It\'s programmed pretending.

Has a computer ever really invented something, not just sifted as
programmed to sift?

What do you think we do? You just tinker with electronics until you have something that works.
No, we conceive products, write manuals with specs, design complex
products, do PCB layouts and mechanical design, have manufacturing
build some first articles from released production drawings, and
expect them to work and be sellable first try, which they most always
do. We don\'t prototype. That\'s designing, not tinkering.
training to sky like a neural net, learning to see and understand circuits like a neural net
neural nets were _modeled_ after the brain.

Try coding just a simple neural net, plenty of open source available.
Takes the mysticism out of it all.
Well, has a computer ever actually invented anything?

Monte Carlo circuit sims, and juggling zillions of molecular
possibilities, are just mechanical assistance to thinking.

It\'s not really \"just\" mechanical assistance, it\'s more like making an impossibly large amount of data accessible to human comprehension. It seems a breakthrough is announced every day now.
If you\'re opposed to developing non-invasive blood testing for making an early diagnosis of cancer with high confidence, then you\'re interested in AI machine learning. That was developed by comparison of a myriad of blood analyses between people with and without cancers, and use that to develop what they\'re calling a \"signature.\"
The signature discovery and development looks to be used more and more.
University College London recently announced the completion of a study that predicts with fairly high confidence, 95%, the development of long covid. They developed the signature around 12 out of nearly 100 blood proteins mostly having to do immunoregulatory interleukins that exist at elevated levels in susceptible people, whether they\'re symptomatic or not. Their discovery methodology used AI machine learning.
A collaboration between UC Berkely and Davis and some others recently made a very dramatic discovery about the provirus phenomenon by using AI to sift through something like 18,000 types of mRNA present in the cytoplasm of cells with and without resident provirus to narrow it down to a two mRNA signature indicative of presence- but more importantly the functionality of the signature mRNA turned out to be blocking signaling pathways for cellular apoptosis and virus activation- that part requires human intervention.
There\'s no end to what we can expect from this technology.

 

[Anthony William Sloman]

On Saturday, October 1, 2022 at 1:23:17 AM UTC+10, Fred Bloggs wrote:

On Sunday, September 4, 2022 at 1:58:39 PM UTC-4, jla...@highlandsniptechnology.com wrote:
On Sun, 04 Sep 2022 16:25:57 GMT, Jan Panteltje <pNaonSt...@yahoo.com> wrote:
On a sunny day (Sun, 04 Sep 2022 07:45:21 -0700) it happened jla...@highlandsniptechnology.com wrote in <j9e9hhpbcalaea2rm...@4ax.com>:
On Sat, 03 Sep 2022 17:23:37 GMT, Jan Panteltje <pNaonSt...@yahoo.com> wrote:
On a sunny day (Fri, 02 Sep 2022 07:41:32 -0700) it happened jla...@highlandsniptechnology.com wrote in <nc54hh5ug3pt2i0bd...@4ax.com>:
On Fri, 2 Sep 2022 05:39:04 -0700 (PDT), Fred Bloggs <bloggs.fred....@gmail.com> wrote:

<snip>

Try coding just a simple neural net, plenty of open source available.
Takes the mysticism out of it all.

Well, has a computer ever actually invented anything?

They\'ve come up with stuff that people couldn\'t find. Whether this counts as invention is a philosophical question.

Monte Carlo circuit sims, and juggling zillions of molecular possibilities, are just mechanical assistance to thinking.

Perhaps. But thinking is pretty mechanical anyway.

It\'s not really \"just\" mechanical assistance, it\'s more like making an impossibly large amount of data accessible to human comprehension. It seems a breakthrough is announced every day now.
If you\'re opposed to developing non-invasive blood testing for making an early diagnosis of cancer with high confidence, then you\'re interested in AI machine learning. That was developed by comparison of a myriad of blood analyses between people with and without cancers, and use that to develop what they\'re calling a \"signature.\"
The signature discovery and development looks to be used more and more.
University College London recently announced the completion of a study that predicts with fairly high confidence, 95%, the development of long covid.. They developed the signature around 12 out of nearly 100 blood proteins mostly having to do immunoregulatory interleukins that exist at elevated levels in susceptible people, whether they\'re symptomatic or not. Their discovery methodology used AI machine learning.
A collaboration between UC Berkely and Davis and some others recently made a very dramatic discovery about the provirus phenomenon by using AI to sift through something like 18,000 types of mRNA present in the cytoplasm of cells with and without resident provirus to narrow it down to a two mRNA signature indicative of presence- but more importantly the functionality of the signature mRNA turned out to be blocking signaling pathways for cellular apoptosis and virus activation- that part requires human intervention.
There\'s no end to what we can expect from this technology.

And quite a lot of other technological tricks for going through lots of data in ways that people don\'t seem to be able to manage without computers.

Coming up with those tricks still does seem to demand human ingenuity, but maybe we can eventually use search algorithms to search through various sorts of search algorithms and automate that too.

--
Bill Sloman, Sydney

 

[Fred Bloggs]

On Friday, September 30, 2022 at 11:51:22 AM UTC-4, bill....@ieee.org wrote:

On Saturday, October 1, 2022 at 1:23:17 AM UTC+10, Fred Bloggs wrote:
On Sunday, September 4, 2022 at 1:58:39 PM UTC-4, jla...@highlandsniptechnology.com wrote:
On Sun, 04 Sep 2022 16:25:57 GMT, Jan Panteltje <pNaonSt...@yahoo.com> wrote:
On a sunny day (Sun, 04 Sep 2022 07:45:21 -0700) it happened jla...@highlandsniptechnology.com wrote in <j9e9hhpbcalaea2rm...@4ax.com>:
On Sat, 03 Sep 2022 17:23:37 GMT, Jan Panteltje <pNaonSt...@yahoo.com> wrote:
On a sunny day (Fri, 02 Sep 2022 07:41:32 -0700) it happened jla....@highlandsniptechnology.com wrote in <nc54hh5ug3pt2i0bd...@4ax.com>:
On Fri, 2 Sep 2022 05:39:04 -0700 (PDT), Fred Bloggs <bloggs.fred....@gmail.com> wrote:
snip
Try coding just a simple neural net, plenty of open source available..
Takes the mysticism out of it all.

Well, has a computer ever actually invented anything?
They\'ve come up with stuff that people couldn\'t find. Whether this counts as invention is a philosophical question.

Monte Carlo circuit sims, and juggling zillions of molecular possibilities, are just mechanical assistance to thinking.
Perhaps. But thinking is pretty mechanical anyway.
It\'s not really \"just\" mechanical assistance, it\'s more like making an impossibly large amount of data accessible to human comprehension. It seems a breakthrough is announced every day now.
If you\'re opposed to developing non-invasive blood testing for making an early diagnosis of cancer with high confidence, then you\'re interested in AI machine learning. That was developed by comparison of a myriad of blood analyses between people with and without cancers, and use that to develop what they\'re calling a \"signature.\"
The signature discovery and development looks to be used more and more.
University College London recently announced the completion of a study that predicts with fairly high confidence, 95%, the development of long covid. They developed the signature around 12 out of nearly 100 blood proteins mostly having to do immunoregulatory interleukins that exist at elevated levels in susceptible people, whether they\'re symptomatic or not. Their discovery methodology used AI machine learning.
A collaboration between UC Berkely and Davis and some others recently made a very dramatic discovery about the provirus phenomenon by using AI to sift through something like 18,000 types of mRNA present in the cytoplasm of cells with and without resident provirus to narrow it down to a two mRNA signature indicative of presence- but more importantly the functionality of the signature mRNA turned out to be blocking signaling pathways for cellular apoptosis and virus activation- that part requires human intervention.
There\'s no end to what we can expect from this technology.
And quite a lot of other technological tricks for going through lots of data in ways that people don\'t seem to be able to manage without computers.

Coming up with those tricks still does seem to demand human ingenuity, but maybe we can eventually use search algorithms to search through various sorts of search algorithms and automate that too.

Then there\'s this kind of thing:
https://news.fsu.edu/news/health-medicine/2022/09/28/fsu-researchers-find-pandemic-altered-personality-traits-of-younger-adults/

A tenth of a standard deviation??? They caution it may not sound like much to the non-specialist, but in their world it\'s positively huge.

Bottom line COVID accelerates the development of quirky personality traits in younger/ less older people usually seen only in older misfit types.

--
Bill Sloman, Sydney