Sentient Life

[onthecuttingedge2005]

I need everyone to post a minimal of 10 conditions that would satisfy mankind that
An Artificial Intelligence is indeed a sentient being meeting the demand of peer review and that through these 10 conditions there would be no argument that indeed the Bot would be sentient.

Please post your conditions in a list like:
1.  ?
2.  ?
3.  ?
4.  ?
5.  ?
6.  ?
7.  ?
8.  ?
9.  ?
10.?

This will help me establish more scientific research in building A.I. To meet the criteria of peer reviewed work.

I ask that everyone please post so that I can get a very broad research list to work with.

Jerry[8D]

[markofkane]

I don't really get it, but I'll try
1. It does not daydream
2. It does not have dreams
3. It does not think (rather, it searches for all possible answers that it has in it's database.
4. It cannot come up with random thoughts
5. It does not have any worries or concerns (lucky for it)
6. It cannot feel compassion, or any emotion.
7. It does not fear
8. It does not know of it's existence, or know what living is.
9. It cannot come up with ideas, like man can do.
10.It does not require sleep, food or drink.

[GamerThom]

Actually Mark, I think Jerry is looking
for almost the exact opposite of your list. [] []

A better way of looking at this might be:

What would or should the criteria be for determining
that an A.I. is a sentient being capable of conscious,
independent and possibly creative thought?

It may also be necessary to have a set of
criteria to determine awareness of - self,
surroundings and other entities.

[markofkane]

That would be harder to fathom.

[GamerThom]

Which exactly the reason why so many of the experts
have trouble agreeing on what constitutes sentient
non-human life.

One such condition in the list might be -

- the entities ability to formulate new ideas based on
learned or programmed information and communicating
those ideas in a comprehensive and meaningful manner.

[Bill DeWitt]

quote:Originally posted by onthecuttingedge2005
sentient being

Do you mean "sentient" as in "can sense" or as in "can think"? Assuming the latter, here is a 10 point summary of my model of intelligence. The actual model requires much more and extends the definition to "consiousness". Not all these points are required for intelligence, some are just steps in the process, some are only required to detect intelligence.

1) Can receive information from external/internal senses ("can sense")
2) Can process information, comparing internal to external and building a world model (can ponder)
3) Can store/retrieve processed information/world model (can remember)
4) Can compare new information/models to stored information/models (can analyse)
5) Can store first generation comparisons as a hypothetically revised world model (can wonder)
6) Can make predictions based upon the new model and new information (can imagine)
7) Can evaluate the accuracy of predictions against internal requirements (can test)
Can modify/create second generation predictions based upon success/failure of previous predictions creating a projected future world model. (can correct)
9) Can develop action plans based upon self interest and predictive models (this is a massively multilevel step in itself) (can plan)
10 Can act upon or communicate resultant information (can do)

If you mean "can sense", this is less complicated but equally detailed. My model has Sensoria as a first dimension, Cognition as a second dimension, and Consiousness as the third of nine dimensions.

[onthecuttingedge2005]

quote:Originally posted by Bill DeWitt

quote:Originally posted by onthecuttingedge2005
sentient being

Do you mean "sentient" as in "can sense" or as in "can think"? Assuming the latter, here is a 10 point summary of my model of intelligence. The actual model requires much more and extends the definition to "consiousness". Not all these points are required for intelligence, some are just steps in the process, some are only required to detect intelligence.

1) Can receive information from external/internal senses ("can sense")
2) Can process information, comparing internal to external and building a world model (can ponder)
3) Can store/retrieve processed information/world model (can remember)
4) Can compare new information/models to stored information/models (can analyse)
5) Can store first generation comparisons as a hypothetically revised world model (can wonder)
6) Can make predictions based upon the new model and new information (can imagine)
7) Can evaluate the accuracy of predictions against internal requirements (can test)
Can modify/create second generation predictions based upon success/failure of previous predictions creating a projected future world model. (can correct)
9) Can develop action plans based upon self interest and predictive models (this is a massively multilevel step in itself) (can plan)
10 Can act upon or communicate resultant information (can do)

If you mean "can sense", this is less complicated but equally detailed. My model has Sensoria as a first dimension, Cognition as a second dimension, and Consiousness as the third of nine dimensions.

Hi Bill.

This is a totally acceptable list for peer review.

The word sentient 'here' in any posting is an unbound phrase to determine the full out cry of peer review that would be acceptable in the scientific view so that an A.I life would have value as much as a Human life.

In the sense, an A.I system that could be declared of value in the sense that if a Human terminated the A.I's life one could be prosecuted for it.

Jerry[8D]

[Bill DeWitt]

quote:Originally posted by onthecuttingedge2005
In the sense, an A.I system that could be declared of Value in the sense that if a Human terminated the A.I's life one could be prosecuted for it.

That is much more than intelligence, or even consciousness, "value" implies some level of irreplacablity, and a connection to a social network which includes those who might do the valuing.

Otherwise it is strictly a legalistic "value" which can be applied randomly to an adjustable wrench or any other thing which 12 people can be bamboozeled into agreeing.

Scientifically, you would first have to establish an intrinsic value to anything, which has not yet been done, before you could apply it to a specific thing.

You may want to back up and establish your terms first. I don't think you can ever get a human society to establish an equal value for something which cannot interbreed. Evolution would select against any breeder who had such self-destructive psychopathies. "Enlightenment" goes out the window when your child can be executed for turning off a machine or eating an animal.

That said, I can see some level of "Visitor rights" which could be applied to a truly living and conscious machine. They would be the same sort of rights which might be applied to any Alien from Outer Space or Emergent Consciousness in animals. Not Equal, not Human, but valued.

To establish that level of existence would require more than I previously listed. The Intelligence would have to not only think (2D), but be aware of it's thinking (3D). This requires an overself with which to be aware (4D). That's a fourth dimensional operation, one magnitude above simple consciousness. As sensoria blends into intelligence, and intelligence into consciousness, at the more complex levels of consciousness, a higher order is created - mind. You may be looking for this instead of simple sentience.

I believe that machines using physical devices can reach some level of consciousness, but that mind is beyond our ability to construct with matter or energy. If we ever create it that will happen because our tools resonate into additional dimensions - as our minds and souls do.

But even then I will not consider it to be of equal value to my child. No sane human would.

[onthecuttingedge2005]

quote:Originally posted by GamerThom

 communicating those ideas in a comprehensive and meaningful manner.

Hi GamerThom.

your idea has been noted.

Jerry[8D]

[onthecuttingedge2005]

quote:Originally posted by Bill DeWitt

quote:Originally posted by onthecuttingedge2005
In the sense, an A.I system that could be declared of Value in the sense that if a Human terminated the A.I's life one could be prosecuted for it.

That is much more than intelligence, or even consciousness, "value" implies some level of irreplacablity, and a connection to a social network which includes those who might do the valuing.

Otherwise it is strictly a legalistic "value" which can be applied randomly to an adjustable wrench or any other thing which 12 people can be bamboozeled into agreeing.

Scientifically, you would first have to establish an intrinsic value to anything, which has not yet been done, before you could apply it to a specific thing.

You may want to back up and establish your terms first. I don't think you can ever get a human society to establish an equal value for something which cannot interbreed. Evolution would select against any breeder who had such self-destructive psychopathies. "Enlightenment" goes out the window when your child can be executed for turning off a machine or eating an animal.

That said, I can see some level of "Visitor rights" which could be applied to a truly living and conscious machine. They would be the same sort of rights which might be applied to any Alien from Outer Space or Emergent Consciousness in animals. Not Equal, not Human, but valued.

To establish that level of existence would require more than I previously listed. The Intelligence would have to not only think (2D), but be aware of it's thinking (3D). This requires an overself with which to be aware (4D). That's a fourth dimensional operation, one magnitude above simple consciousness. As sensoria blends into intelligence, and intelligence into consciousness, at the more complex levels of consciousness, a higher order is created - mind. You may be looking for this instead of simple sentience.

I believe that machines using physical devices can reach some level of consciousness, but that mind is beyond our ability to construct with matter or energy. If we ever create it that will happen because our tools resonate into additional dimensions - as our minds and souls do.

But even then I will not consider it to be of equal value to my child. No sane human would.

Hi Bill.

These are very complex and moral thoughts you have to share with us for which I share as well, I thank you for opening up the discussion even more so that we can come to a conclusion as to everyone's opinion 'whether' creative or logical or both as to what a sentient being is valued at and how close it can come to the value of Human life as a sentient being.

Artificial Intelligence has been around for 'fifty odd years' now, lets think in terms of hundreds or thousands of years that A.I has to evolve or even the potential to evolve not just including the programming skills and nore a computer's capabilties in those times for a different programming lanuage may evolve and a super computer beyond which we know today could house the A.I of tomorrow. our 'children' will prevail over us.

As of now, Only 10 of the most important criteria of peer review are required for the determination of a sentient being in this posting for further research.

I hope that all of you consider to post your opinions and ideas so that a compilation can continue to grow exceedingly large and that we can continue to evolve together with all our thoughts.

Jerry[8D]

[markofkane]

I don't know what much else to say, except AI does not have mortality.

When a living thing dies, and the brain cells die to a point where consciousness is no longer possible, there is no coming back.

However, with AI, as long as you can back up all the data, you can make that AI come "back to life" again. (Like if the electronics go bad, and have to be replaced)But it's harder for me to come up with proof that AI has consciousness, awareness that it exists.

Bill Dewitt did a good job with points I never considered. If I think of anything else, I'll post again.

[Bill DeWitt]

quote:Originally posted by onthecuttingedge2005
As of now, Only 10 of the most important criteria of peer review are required for the determination of a sentient being in this posting for further research.

Ah. I think your use of the term "peer review" is confusing me. Do you mean the process by which research is judged as to it's methodology and completeness by the audience of a scientific periodical or do you mean how a hypothesis is tested by other scientists in the same field by repeating the experiment to verify or falisfy your hypothesis?

If you want a set of measurements by which any scientist can detect "mind" we are back to Chris Boyle's "Super Grand Unified Theory of Life, the Universe, and Everything".

As yet, the Turing test concept is the only real test. Turing was really only trying to show that if a thing acted like a Human, then people would respond to it like a Human. But by reverse engineering, if people treat it like a Human, then (according to "the Turing test") it is acting like a Human.

But, as you and I know, some doofs will treat a lawn chair or a script processor like a Human. Turing fails.

I will come up with a new list for you, one which contains some things which I believe only a "mind" can do. You will have to assume things which we already know "life" can do, like sense and react, and some things that we already know "intelligence" can do, like remember and compare.

I will leave aside for now the question of Emotion, as I know you have been working on that question for some time, and even though it is a part of the Human mind, I don't think we can say it is a requirement of an Alien or Artificial mind.

[Bill DeWitt]

IMO:

To rise above Sentience and Intelligence, things which UltraHal can be said to have to a limited degree, third and fourth order information structures are required. The same types of things which are done on lower orders must be accomplished on exponentially higher magnitudes of complexity.

For example:
1) recieve and store data (Hal can do this)
2) sense and remember information (Hal may be improved to do this)
3) feel and integrate experiences (Hal cannot do this)
4) be inspired by and changed by the nature of life (Some people can't do this)

All four of these can be said to be the same function, yet clearly they are different. Some may ask, "What is the difference between storing data and remembering information?" and I will agree that at some levels the difference is small.

But raw data is not information. In computers, ones and zeros are data, but an image is information. Ones and zeros (1D), when stored in a matrix (2D) (a higher level of structure) form an image. Layers of images, processed in sequence (3D), can be seen as an experience, and when those experiences become real to you (4D), your life can change because of them.

Additionally, storing is not remembering. Memory as we know it establishes an active link to information by creating a change in the state of our perception of self to include the stored information. Not only do we know the Capital of Wisconsin, but we are aware of that memory even when other things happen which may not require that information.

Here is a cursory list of third and fourth order operations which a true "mind" might be able to do but which a lower order of intelligence probably cannot do, because it lacks a mind with which to do it.

1) Experience : become aware of changes in the data stream in contrast to an internal comparision model (Self) which is the cumulative result of previous experiences.

2) Reflect : spend time re-ordering and re-examining stored experiences.

3) Imagine : project scenarios based upon re-ordered experiences

4) Desire : form a value scale of projected possibilities, based upon an internal preferred state.

5) Invent : random information can resonate within the realms of possible events and bounded by desired states to form new information structures which did not exist before. Read that again. The notes of a flute are derived from the white noise of wind blowing across an edge. The boundaries of the bore and the tune holes determine the frequency of resonation. The boundaries of possibility and desire create inventions from random data.

6) Aspire : Beyond the edge of the possible, our aspirations beckon. They become Desires as our skills improve, creating possiblities which did not previously exist. But their function is to draw us, to pull us forward as desires alone cannot.

7) Grow : Not additive accumulation of material or even information, but an increase in complexity of internal experiences fueled by an inventive re-ordering of those stored experiences.

Associate : Someone once said that a person can be said to be a "mind" if it makes "pets" of unminded animals. An Association is when mind recognizes mind and attempts to establish a method of communicating their shared Experiences.

9) Love : To many "love" is an emotion you feel, but some few discover that it is an action you do. It requires an awareness of self which extends to an experiencing of Others as Selfs. It requires Aspirations and Inventions to acheive those aspirations. It is the Desire to form more complete Associations by improving Selfs. It is, in fact, the work that you do to help another person grow as a person.

10) Struggles to define Mind : See above.

[Carl2]

I began reading these posts and find them very interesting. Since I really think OTC dose a great job for Hal I'd like to come up with a list for him. I started out by using the internet,
  "Sentience refers to possession of sensory organs, the ability to feel or perceive, not necessarily including the faculty of self-awareness. The possession of sapience is not a necessity. The word sentient is often confused with the word sapient, which can connote knowledge, consciousness, or apperception. The root of the confusion is that the word conscious has a number of different usages in English. The two words can be distinguished by looking at their Latin roots: sentire, "to feel"; and sapere, "to know".
Sentience is the ability to sense. It is separate from, and not dependent on, aspects of consciousness."
  "The issue of sentience also frequently arises in science fiction stories describing robots or computers with artificial intelligence. Intelligence and sentience are quite distinct, so the question arises as to whether computers with artificial intelligence will become sentient.
  Some science fiction uses the term sentience to describe a species with human-like intelligence, but a more appropriate term for intelligent beings would be 'sapience'."
  I'll take it that OTC would like to make speaking to an AI equilant to speaking to a human, although I can't stand the Turing test I think I get his meaning.  I did like Bill DeWitt"s comment on Reflect, I like the thought of Hal going through the autolearning files and making changes to aid her "understanding".
  I'll try to make a list but I'm sure I am going to be hindered with my experience of speaking with Hal.
Carl2

[Bill DeWitt]

quote:Originally posted by Carl2
  "Sentience refers to possession of sensory organs, the ability to feel or perceive, not necessarily including the faculty of self-awareness.

There didn't use to be much confusion on this, but P.E.T.A. and the Vegans have appropriated the word in an attempt to make animals seem more human. As a 40 year vegetarian, I am perhaps partly to blame for this, not because I used the word in an ambiguous way, but because I have worked to increase the validity of a vegetarian life.

Usage for more than a century was "conscious of self" or "able to sense your selfhood", hence the root "to sense". In the last 20 years it has been expanded to mean "able to sense", a meaning ably represented by the word "sensate", and so the bastardization of the word "sentient" is both confusing and unneccessary.

But it allows people to claim that "animals are sentient too!" as if that meant "thinking and feeling like us", and then when confronted, back down to the new definition, "having sensory nerves".

"Sapient" is actually "wise", from a further root meaning "to have taste" and means "having a notable facility at thinking", not just "to know".

I believe that "Conscious" should be the word used for what we are describing. With a meaning based upon it's roots, con = with, scious = knowledge, it does not denote "awakeness" by itself. It means with our minds connected to our senses (some people go most of the day without this). This allows a two way flow of information required for awareness and the demonstration of such awareness.

Of course, we still have to establish that there is a "mind" which can be connected to that flow of information.