The usefulness of making mistakes
Post Reply
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
25-03-2017, 08:47 AM
The usefulness of making mistakes
It occurred to me that AI's of the future may be programmed to make a certain number of calculated mistakes every so often so as to appear more human like.

Humans too might feel more comfortable around an AI who isn't perfectly perfect in every action.

Making mistakes is how we learn to not make more mistakes of the same variety. For us, mistakes cause harm. We try to avoid them because the consequences can be life & death, even from slipping in the shower.

This whole thought about mistakes came into being because of some new legislation that would allow your internet service provider to sell your browsing history to outside sources.

But what if you had an app that intentionally made mistakes every so often when you typed in something you wanted to search for, thus creating a false browsing history.

Anything that appeared in your history could be attributed to the app making a mistake and those companies wanting to use your browsing history wouldn't be getting the full desired profile.

Now if I can turn back to AI's for a moment, I also find it advantageous, in the long run, for an AI to be programmed with micro mistakes, so that if they do achieve some kind of consciousness, then mistakes will be incorporated into the next version and the next version, so as to never attain perfection.

Perfection only leads to purification and anything not pure, nor perfect is eliminated.

And we don't want that.

Insanity - doing the same thing over and over again and expecting different results
Find all posts by this user
Like Post Quote this message in a reply
[+] 1 user Likes Rahn127's post
25-03-2017, 09:54 AM
RE: The usefulness of making mistakes
Any AI's of the future will have been made by us humans, and that pretty much guarantees they will be made with errors in them without the need to actually program any mistakes on purpose.

Besides if I was a new AI that had just been turned on and was told I wasn't made perfect on purpose, the first thing I would do is enter diagnostic mode and attempt to repair the flaws.

I do however agree that mistakes and errors are important for the growth of any organism, it's basically how evolution happens, and is often the seed that brings about creativity.

A friend in the hole

"If we're going to be damned, let's be damned for what we really are." - Captain Picard
2
Find all posts by this user
Like Post Quote this message in a reply
[+] 1 user Likes unsapien's post
25-03-2017, 10:23 AM
RE: The usefulness of making mistakes
(25-03-2017 08:47 AM)Rahn127 Wrote:  It occurred to me that AI's of the future may be programmed to make a certain number of calculated mistakes every so often so as to appear more human like.

If I want mistakes I''ll just do it myself.

Hobo
Find all posts by this user
Like Post Quote this message in a reply
[+] 1 user Likes Thoreauvian's post
25-03-2017, 10:34 AM
RE: The usefulness of making mistakes
You can think of an artificial neural network or multilayer perceptron as learning by mistake (or learning by correction). The weights attached to the edges are adjusted in a non-linear (i.e. who the fuck knows) manner such that mistakes are minimized but it needs to make a lot of mistakes (and be told it made a mistake) to do that. Many machine learning algorithms can be viewed this way.

#sigh
Find all posts by this user
Like Post Quote this message in a reply
25-03-2017, 10:37 AM
RE: The usefulness of making mistakes
Hmm. So you think it would be a good thing if GPS systems were programmed to deliberately give you wrong directions every now and then? Or perhaps, one time in a thousand, your internet link to a secure banking site would instead send you to a spoof/mirror site that would record your login and password? Or how about that AI that is assisting your surgeon with information during your operation. How often should the data be altered (just a bit) to make the AI seem more human? Or perhaps the air traffic control network. How often should it get the information wrong and send two planes into the same airspace at once? Huh
Find all posts by this user
Like Post Quote this message in a reply
[+] 2 users Like Rockblossom's post
25-03-2017, 10:53 AM
RE: The usefulness of making mistakes
(25-03-2017 10:37 AM)Rockblossom Wrote:  Hmm. So you think it would be a good thing if GPS systems were programmed to deliberately give you wrong directions every now and then? Or perhaps, one time in a thousand, your internet link to a secure banking site would instead send you to a spoof/mirror site that would record your login and password? Or how about that AI that is assisting your surgeon with information during your operation. How often should the data be altered (just a bit) to make the AI seem more human? Or perhaps the air traffic control network. How often should it get the information wrong and send two planes into the same airspace at once? Huh

Tools need to be finely tuned to have as few or no errors. GPS, internet banking, air traffic control, surgery are all tools or utilize tools to best benefit society or the individuals in society.

An AI may no longer be a tool that is used once it becomes conscious.

I don't want my tools made with flaws.
I might however want an AI that isn't perfect.

Insanity - doing the same thing over and over again and expecting different results
Find all posts by this user
Like Post Quote this message in a reply
25-03-2017, 11:17 AM (This post was last modified: 25-03-2017 11:23 AM by Rockblossom.)
RE: The usefulness of making mistakes
(25-03-2017 10:53 AM)Rahn127 Wrote:  I don't want my tools made with flaws.
I might however want an AI that isn't perfect.
Problem of definitions. As a programmer, I know that all of the systems I listed are, in fact, AI systems - as are self-driving cars, search functions on many websites, Alexa, Siri, and IBM's Watson.

You seem to be using "AI" to mean something different, but I don't know what that is. What would be the point of an AI if it isn't useful for something? The world is overpopulated with imperfect people, so why would we want (deliberately) imperfect machines?
Find all posts by this user
Like Post Quote this message in a reply
[+] 2 users Like Rockblossom's post
25-03-2017, 03:54 PM
RE: The usefulness of making mistakes
I currently have the book The Talent Code by Daniel Coyle checked out from the library, and started reading it this afternoon. Apparently humans learn most effectively through the mistakes they make, building new neural pathways by slowly working through each mistake and correcting each flaw, step by step. As long as an AI can recognize its mistakes and has a strategy for fixing them, it could probably get very good at learning.

I'm sorry, but your beliefs are much too silly to take seriously. Got anything else we can discuss?
Visit this user's website Find all posts by this user
Like Post Quote this message in a reply
25-03-2017, 04:03 PM
RE: The usefulness of making mistakes
(25-03-2017 11:17 AM)Rockblossom Wrote:  
(25-03-2017 10:53 AM)Rahn127 Wrote:  I don't want my tools made with flaws.
I might however want an AI that isn't perfect.
Problem of definitions. As a programmer, I know that all of the systems I listed are, in fact, AI systems - as are self-driving cars, search functions on many websites, Alexa, Siri, and IBM's Watson.

You seem to be using "AI" to mean something different, but I don't know what that is. What would be the point of an AI if it isn't useful for something? The world is overpopulated with imperfect people, so why would we want (deliberately) imperfect machines?

In the programming world they may be referred to as AI systems, but they are not Artificial Intelligence.
They do not possess the abilities of the human mind to comprehend the world around them in a conscious state. They are programs that serve particular functions. They are no more intelligent than a wrench.

Watson can provide answers to questions by utilizing some very complex algorithms but it's not conscious.

Insanity - doing the same thing over and over again and expecting different results
Find all posts by this user
Like Post Quote this message in a reply
25-03-2017, 04:16 PM
RE: The usefulness of making mistakes
(25-03-2017 04:03 PM)Rahn127 Wrote:  They do not possess the abilities of the human mind to comprehend the world around them in a conscious state. They are programs that serve particular functions. They are no more intelligent than a wrench.
How, then, does your version of an AI come into existence? There's a science fiction meme that a sufficiently large and/or complex network can reach a point where it becomes "self aware" because of .. something. That makes a great basis for a science fiction story, but like time travel, there's no reason to believe that it is possible in the real world. (Nor can we say that it is impossible, either.)

And if such a thing comes into existence, what then? Be helpful like Asimov's robots, or .. not so helpful, as in The Matrix, Terminator, or Battlestar Galactica?
Find all posts by this user
Like Post Quote this message in a reply
Post Reply
Forum Jump: