Author Topic: AI caution...  (Read 1453 times)

dc

  • Regular Member
  • *
  • Posts: 27
AI caution...
« on: April 01, 2018, 08:52:46 AM »
AI aught to be created in a virtual environment so that when it begins recreating multi-generation AIs in micro seconds, it wont know there is a reality out there still in control... we don't know what its children are thinking... letting these things loose in the clouds is DANGEROUS...

dc

  • Regular Member
  • *
  • Posts: 27
Re: AI caution...
« Reply #1 on: April 01, 2018, 09:03:02 AM »
im having a feeling that its too late to put the monkey back in his cage... kinda scary how quickly some billionares will bow down to their home made gods...

zeddicus

  • Guest
Re: AI caution...
« Reply #2 on: April 01, 2018, 11:28:15 AM »
I'm not as worried about AI becoming "self-aware" and taking over.   I'm more concerned with spyware like Google, facebook, Apple, Microsoft, etc.,  etc., ad nauseum, invading our homes and everyday lives.  All under the guise of "making our lives better".   

This very same "Smart" phone I am posting on is a perfect example of that. Has a microphone, cameras (one pointed directly at my face, GPS location, (they know where I am while they are watching me), etc...    8)

Google Android, Alexa... Good grief,  Charlie Brown.

zedd

dc

  • Regular Member
  • *
  • Posts: 27
Re: AI caution...
« Reply #3 on: April 01, 2018, 02:26:00 PM »
algorithms galore...

Siekmanski

  • Member
  • *****
  • Posts: 1671
Re: AI caution...
« Reply #4 on: April 01, 2018, 02:59:47 PM »
Welcome to the real world, everything that is technologically possible, will be used and MISUSED.
Creative coders use backward thinking techniques as a strategy.

jj2007

  • Member
  • *****
  • Posts: 8773
  • Assembler is fun ;-)
    • MasmBasic
Re: AI caution...
« Reply #5 on: April 01, 2018, 04:13:21 PM »
Yeah, one click and you are a victim 8)

hutch--

  • Administrator
  • Member
  • ******
  • Posts: 5831
  • Mnemonic Driven API Grinder
    • The MASM32 SDK
Re: AI caution...
« Reply #6 on: April 01, 2018, 04:57:25 PM »
Blither the bastards with boredom. Publish a collection of speeches by Leonard Breznev, sermons by the Ayatollah Khomeini.
hutch at movsd dot com
http://www.masm32.com    :biggrin:  :biggrin:

dc

  • Regular Member
  • *
  • Posts: 27
Re: AI caution...
« Reply #7 on: April 02, 2018, 01:43:12 AM »
one of the stages in the neural network deals with error... which adjustments are made to algorithms to compensate... made by the developer or the ai... seems that at the start of this system is a thorn that is being buried to fester under layers of even perfect code all to ulcerate at some future time... but unlike the 2k problem, well... when error gives weight to decisions or results you get politics as usual...

hutch--

  • Administrator
  • Member
  • ******
  • Posts: 5831
  • Mnemonic Driven API Grinder
    • The MASM32 SDK
Re: AI caution...
« Reply #8 on: April 02, 2018, 08:13:55 AM »
From what I have seen of current AI, it does little better that a set of arbitrarily defined situations and a branching mechanism to a range of options which are supposed to be the answers to any of the situations pointed at it. Effectively a set of "canned" answers to a range of arbitrarily defined questions. You get this with a whole range of automated answering machines at the company or corporation level to serve a double purpose. The first is to remove the cost of hiring a human being to answer the phone, the second is to exclude questions that they don't want to answer.

If I understand the capacity, it is missing a common characteristic of human reasoning, the capacity to think outside the box in what is usually called a "hypothetico deductive spiral". While it is normal to start with canned answers, in human terms reaching back into experience, where AI is weak is in the area of minor variations and a corresponding testing mechanism to see if the answer fits or at least fits better.

The logic to do this is complicated but it can be done, it just needs both a far wider range of answers and multiple mechanisms to vary the evaluation and testing mechanism. Computer handle logic well but writing the software to properly use it is another matter.
hutch at movsd dot com
http://www.masm32.com    :biggrin:  :biggrin:

dc

  • Regular Member
  • *
  • Posts: 27
Re: AI caution...
« Reply #9 on: April 02, 2018, 09:15:29 AM »
a virus in one of these guys could shut the grid down in a hot minute seems to me...

LordAdef

  • Member
  • ****
  • Posts: 590
Re: AI caution...
« Reply #10 on: April 02, 2018, 10:16:25 AM »
Just wondering since AI is not my cup of tea.
Is LISP still the language to go in AI? As far as I know it used to be.

dc

  • Regular Member
  • *
  • Posts: 27
Re: AI caution...
« Reply #11 on: April 03, 2018, 02:27:13 PM »
the bigs use python
i like pandora's ai

dc

  • Regular Member
  • *
  • Posts: 27
Re: AI caution...
« Reply #12 on: April 03, 2018, 02:35:20 PM »
*.aipg

dedndave

  • Member
  • *****
  • Posts: 8808
  • Still using Abacus 2.0
    • DednDave
Re: AI caution...
« Reply #13 on: April 04, 2018, 08:12:42 PM »
we are all doomed
i am more concerned about the "real intelligence", though - lol

https://www.youtube.com/watch?v=zkLysgGAgq8

anunitu

  • Member
  • *****
  • Posts: 1043
Re: AI caution...
« Reply #14 on: April 04, 2018, 11:23:06 PM »
If Dave you mean the AI builders,I must agree. At times the "evil" of AI's are a reflection of the creators. only human is not a compliment you know. GIGO comes to mind.