Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

What can't artificial intelligence do?

  • Please log in to reply
38 replies to this topic

#26 ButterFly

ButterFly

    Aurora

  • *****
  • Posts: 4,759
  • Joined: 07 Sep 2018

Posted 14 March 2023 - 11:18 PM

Hello, I am so thankful for Ai, you can download for free from Microsoft, "seeing Ai."

So incredibly helpful for visually impaired or even blind patients. Yes I use it almost continually, as I am visually impaired, can't read, drive, etc. That application will even describe my surroundings. 

I can watch foreign movies with subtitles using head phones not to bother anyone!

 

Can't wait for my self driving car! My son installed "Alexa" in most of the rooms in my house.

I can only see using "averted vision" which we use when we want to observe very dim objects far away!

And how much better is that than a seeing eye dog!  Except for snuggling purposes.  But even those barriers could break down in a better way.  Isolation is huge problem for impaired persons.  Permitting engagement with others, such as reading and driving, is a huge boon that not even a cuddly dog can match.  To some extent, we all have limitations, both physical and mental, that serve as barriers to engagement.  It is unlikely that AI can overcome a person's dislike for X, Y, and Z, just because.  But if these things learn what we know and how we learn, even the dimmest among us may have a fighting chance at becoming rocket scientists - tailored education in terms that one can comprehend.

 

The biggest roadblocks to getting AI to better help us be "I" will be us.  Is there really incentive to make the general population better at engaging with each other?  Every form of communication humanity developed helped usher in Revolutions - in every sense of the word.  But moving from facilitating mere access to facilitating understanding by finding and packaging relevant data comprehensibly way can cut both ways.  As the events of the last decade have shown, the data one is presented with in an echo chamber can lead to disastrous consequences for those therein.  Whether AI will free us from our limitations, or just become another way to hold on to keys of power, depends on us.

 

popcorn.gif



#27 Tony Flanders

Tony Flanders

    ISS

  • *****
  • Posts: 21,951
  • Joined: 18 May 2006
  • Loc: New Lebanon, NY and Cambridge, MA, USA

Posted 15 March 2023 - 05:06 AM

Everybody seems to imagine that machines with AI have reflective consciousness.  They do not.


I'm inclined to agree with you. But I would be happier with that belief if I could define "consciousness," which I cannot.

My hunch is that consciousness, like beauty and ugliness, good and evil, is in the eye of the beholder, and fundamentally impossible to define in any rigorous way.

There is no known technical path toward making them conscious and self-aware.


That's for sure! But "no known path" and "no path" are not the same thing.

#28 therealdmt

therealdmt

    Surveyor 1

  • *****
  • Moderators
  • Posts: 1,952
  • Joined: 05 Mar 2015
  • Loc: 35° N

Posted 15 March 2023 - 09:51 AM

Message from the Moderators

 

This is a current topic of great interest and potentially wide-sweeping impacts to our lives and future. Ideally we should be able to calmly discuss the topic raised, and let us try to do that (for those who choose to participate). However, if the subject cannot be discussed calmly, without attacking or being dismissive of or derisive towards fellow members, the thread will ultimately have to be closed.

 

After review and a cleanup by the Moderators of some unpleasant posts, this thread is reopened for discussion. 



#29 Tony Flanders

Tony Flanders

    ISS

  • *****
  • Posts: 21,951
  • Joined: 18 May 2006
  • Loc: New Lebanon, NY and Cambridge, MA, USA

Posted 17 March 2023 - 12:58 PM

Getting back to the initial question, the first post started like this:
 

We have made machines that can play chess better than we can.  We are close to making machines that can write novels better than we can.


I think the second statement simply isn't true. Mind you, being better at chess is easy to define rigorously, whereas whether a novel is good, bad, or indifferent is a matter of opinion. But I think I am far more likely to be killed by a lightning bolt than that I will read a machine-produced novel that enriches my life more than War and Peace or Pride and Prejudice (to name two of my faves).


  • Michael Covington likes this

#30 Sketcher

Sketcher

    Vanguard

  • *****
  • Posts: 2,181
  • Joined: 29 Jun 2017
  • Loc: Under Earth's Sky

Posted 18 March 2023 - 01:30 PM

For starters, is there a limit?  Any reason we shouldn't go there?

In response to the existence of a limit:

 

Unless some form of hardwired, fail-safe, limit is incorporated into the design; I see no limits -- only a continuous increase in power and capability.

 

Will enforceable limits even be possible?  Will limits eventually be put into place?  I can see this occurring in the short-term; but sooner or later I believe that it's inevitable that a push will be successful in going beyond those limits.

 

In response to:  "Any reason we shouldn't go there?":

 

(Yes, No, Maybe)  I don't believe that any answer can be clearly accepted as being more correct than the other two.  There are those who can (or will) have legitimate reasons to support each of the three options.

 

In my opinion, the genie is out of the box.  The capabilities of AI will continue to grow.  And eventually, as the technology and functioning evolve, the term "Artificial Intelligence" will need to be replaced by a more accurate, a more appropriate descriptor -- after (or perhaps even before) a new form of life is given birth and recognized as such.

 

The technology itself will see to it that the question: "Any reason we shouldn't go there?" becomes irrelevant.  It's going to "go there" regardless of the desires and wishes of those who believe it shouldn't.



#31 EJN

EJN

    Fly Me to the Moon

  • *****
  • Posts: 5,495
  • Joined: 01 Nov 2005

Posted 18 March 2023 - 01:52 PM

What can't artificial intelligence do? 

 

Drink beer.


  • Rickycardo likes this

#32 yuzameh

yuzameh

    Messenger

  • -----
  • Posts: 429
  • Joined: 13 Dec 2022

Posted 18 March 2023 - 02:16 PM

Limits are mathematically defined by Church and Turing and some other USA fellow I always forget the name of.  Not sure if they've ever been applied to the hominid brain, though, for although we barely know how that functions to give this "consciouness" thing, computability is mathematically capable of being deilmited.

 

Then again, as 'consciousness' itself is a term emotive, with it and terms like subconscious and unconscious in the technical sense being pooh-poohed by those against the Freudian pseudoscientic perspective, we'd have to define consciousness first.

 

Does an idiot savant who cannot socially interact or care for themselves but has innate abilities at doing things considered intelligent, if at times almost of a genius level capability, but no ability to engage in original thought, conscious?  In effect then intelligence is being seen, but not deemed necessarily self aware.  That is, if true idiot savants have ever really existed and not nothing more than a myth.

 

And this is why this topic can go on and on for there are issues of semantics and outlook akin to those that exist such that there will never be one universal view of politics or one universal view of religions.  The indefinability of it all is one of the reasons it can end up becoming part of their bailiwick.

 

When the machine, without prompting, spits out the request like "Can you charge me up using a USB C port next time?  This ilghtning thing makes me TTL's tingle", THEN I will creep out.  But it still won't creep me out as the misuse of the ai concept in daily life.  Remember 'expert systems', they came in very useful, and help for instance medics diagnose quicker and better, but are no more intelligent than the shelves of books they replaced.  But there, there was good data input.  Nowadays there is a lot of models based on models based on models with the empirical data it is all based on well buried, thus decision making machines will not be able to function properly unless they can learn by themselves by objectively fetching their own data for which to intercompare.

 

PS

 

I got fed up about hearing about this chatgpt thing, which I don't believe claims to be artificial intelligence, which has caught on as a fad, with lots of people trying to trick it into being racist, sexist, rude, obscene, and just plain wrong.

 

If something having 'artificial intelligence' does ever achieve 'consciousness' then thanks to being messed with by humans (especially their primary contacts who will strong agendas in AI and related mindviews likely not similar to the rest of us) it will no doubt have a nervous breakdown, become neurotic, schizoid and paranoid, crawl into an electronic corner, curl into an electronic ball and softly shed electronic tears.

 

Meanwhile, for the human's paranoid about AI, put some funding into converting tasers into hand held EMP projectors, then we won't need Sarah Connor at all.



#33 garret

garret

    Surveyor 1

  • *****
  • Posts: 1,979
  • Joined: 26 Jul 2009
  • Loc: Netherlands

Posted 18 March 2023 - 02:43 PM

Is there a limit what the human brain can come up with with or without AI?

I say yes there is a limit.
But of course I can't tell or fantasize what would be beyond that limit, that's also my limit what my brain can do.



#34 bobzeq25

bobzeq25

    ISS

  • *****
  • topic starter
  • Posts: 31,755
  • Joined: 27 Oct 2014

Posted 18 March 2023 - 03:44 PM

Is there a limit what the human brain can come up with with or without AI?

I say yes there is a limit.
But of course I can't tell or fantasize what would be beyond that limit, that's also my limit what my brain can do.

"The universe is not only stranger than we imagine.  It's stranger than we can imagine."   Arthur C. Clark

 

I'm guessing a self constructing self programming AI could imagine it.  I think many of the posts above are looking at the present state of AI, or imputing some magical characteristic to our complicated brains.  If an AI has 100X or 1000X the number of cells and connections I think it would have more capability.

 

Again, we'd have nothing to fear from such an advanced AI.  We have nothing it would want.



#35 StarBurger

StarBurger

    Apollo

  • -----
  • Posts: 1,120
  • Joined: 06 Mar 2018
  • Loc: North Country NY

Posted 18 March 2023 - 03:46 PM

What Descartes might say if alive today:  "I don't think. Therefore I am not"


  • Epick Crom likes this

#36 ButterFly

ButterFly

    Aurora

  • *****
  • Posts: 4,759
  • Joined: 07 Sep 2018

Posted 18 March 2023 - 04:32 PM

"The universe is not only stranger than we imagine.  It's stranger than we can imagine."   Arthur C. Clark

 

I'm guessing a self constructing self programming AI could imagine it.  I think many of the posts above are looking at the present state of AI, or imputing some magical characteristic to our complicated brains.  If an AI has 100X or 1000X the number of cells and connections I think it would have more capability.

 

Again, we'd have nothing to fear from such an advanced AI.  We have nothing it would want.

Looking at orca and elephant brains is rather helpful in that regard.  Forget language, there's indication of culture among groups of these animals.  Humans still have a larger encephalization ratio (brain to body ratio), so it's not entirely clear whether number alone is important for animals.  Taking that to things that don't have "bodies" in the conventional sense is more difficult.  Is it the number of cameras?

 

On the needs and wants, "what if orca had opposable thumbs" beg the question "what would they do with it?"  There is no evidence that present AI wants anything.  There really is no basis for saying it will want anything in the future.  Perhaps more lugnuts?  Unfortunately, there is also no basis for saying that some AI won't want anything in the future.  There is, however, plenty of evidence that humans who want thing are using AI to get that which they want.  It's the humans we should be weary of.


  • bobzeq25 likes this

#37 Rickycardo

Rickycardo

    Desdenova

  • *****
  • Posts: 2,215
  • Joined: 29 Mar 2009
  • Loc: 3rd Rock

Posted 18 March 2023 - 05:08 PM

This line of thought reminds me of this:

https://media.giphy....wlbtS/giphy.gif


  • ButterFly likes this

#38 ButterFly

ButterFly

    Aurora

  • *****
  • Posts: 4,759
  • Joined: 07 Sep 2018

Posted 18 March 2023 - 05:57 PM

This line of thought reminds me of this:

https://media.giphy....wlbtS/giphy.gif

I'm entirely content with having no purpose and doing my job poorly, so why shouldn't AI?


  • Rickycardo likes this

#39 Inkswitch

Inkswitch

    Messenger

  • *****
  • Posts: 499
  • Joined: 09 Nov 2003
  • Loc: Harmony, NY USA

Posted 20 March 2023 - 10:24 AM

Regarding limits.  There is only so much "computing power" that can be enclosed within a space.  Once that space gets big enough then the speed of light will become the limit.  Processing nodes on one side of the space will be unable to communicate quickly with nodes on the other side of the space.  For some computable problems this issue will increase the time to compute unreasonably.

 

As you attempt to increase the "computing power" in a given space by squeezing in more stuff (transistors or whatever) you will eventually run up against fundamental size restrictions that cannot be overcome.


  • Michael Covington likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics