• 247
    MEMBERS
  • 1429
    MESSAGES
  • 0
    POSTS TODAY

Discussions

Artificial Intelligence

Started May-15 by HWPeeler (HPeeler); 327 views.
HWPeeler (HPeeler)

From: HWPeeler (HPeeler)

May-15

We might be able to have computers that simulate intelligent behavior but this is far from simulating humanity. People think in terms of pictures (generally). We then translate that picture into the language we prefer. Computers can only process the text, but this does not form a picture in their imagination. They have no imagination. They are not conscious.

What are memories? We remember in pictures (again only generally) . Memories of a song can bring back associated memories of a concert we went to decades ago, smells at that concert, people we met there. Computers can't do this. A human can pick one sound out of many and isolate just that one. A computer cannot take a waveform of an orchestra playing Swan Lake and pick out just the oboe portion and play it.

RGoss99

From: RGoss99

May-16

While I take your point, I find fault with your examples.

One example, as an oboist, and with a degree in comp sci, yes, I can program a computer to isolate an oboe, much easier then I can train an untrained musician to do so. With each instrument playing the same note, there is a measurable profile, and it only takes programming to analyze and determine it because a computer so programmed will not be distracted as would be the average human, by the other instrument profiles.

My thesis was in database management, the basic work my project performed was if we say x is true, then identify x as opposed to other non x´s. My thesis had 10,000 present examples, to compare with an equal number of historic examples, and discovered that only one of the present, matched with its claimed source. Having done that, it provided a ranked order of the near misses. Something that could never be done by a human in a timely manner. In my final project(1986) the process took 3 minutes, printing out the results took 8 hours. Over time and place language and images change. In the case of words, definitions not only change, but are equivocal when it comes to time and geography. The test was quite simple, if x1 is true, is x2 equally true - note above, of 10,000 examples only 1 passed the test as true. Humans have memory filters that distract one from a given stimulus.

HWPeeler (HPeeler)

From: HWPeeler (HPeeler)

May-16

Re: the oboe

I can see isolating a single note by its frequency, but the oboe must be the only instrument playing that note and this only works for a single note, not bars of music. A human can pick out the oboe by its tonal quality. The human does not have to analyze the waveform to do this but the computer only has the waveform to work with. The computer is not conscious.

Your thesis only dealt with text? Text is simple to analyze. Humans are not limited to text. We can synthesize a mental image of what the text describes. The computer can not form a mental image.

Even a cockroach can form a mental image motivated by a smell. A computer can't. With proper sophistication a computer might be able to identify the smell and assign a text string to it, but this is not a mental image.

A computer can not analyze "pretty" or "pleasant". In your thesis did you do much more than "match the text". The effectiveness depends on interpreting the observation into text.

Can a computer determine if a poem is beautiful? Beauty can not be determined by parsing text. Beauty is in mental image the text implies. Will artificial intelligence ever get to the point of resolving this mental image? It is the magic of biology.

RGoss99

From: RGoss99

May-16

I made no reference to frequency, but was referring to tone quality, which is much more complicated then frequency, just as a human can learn from another that this sound is an oboe, a computer and learn from an algorhythm to do the same thing. Given an algorhythm a computer con do so more efficiently. No a human does not have to scientificly analyze the sound any more then a human has to analyze color that way, in both cases through some language both can learn to discriminate.

No my thesis did not only deal with text, also sound, and visual. So your human point  is invalid.

Cockroach image by smell is taught, just as a computer can be taught.

Your concept of pretty or pleasant  is unrelated to my post which only stated that your oboe example is invalid. I stated at the outset that I agreed with your point, just not all of your examples. So the rest of your post is pointless based on an implied misquote on your part.

But moving on, concept of beauty, pretty, pleasant, are culturally subjective. But even here while artificial intelligence can not creat these it can be more efficient then humans in analyzing lack of beauty, pretty and unpleasant, taught in the same way theat people are.

A lot of artistic values we have can be measured by computers.

Heraldry has rules, that determine good and bad, for most of them it would not take much for a computer to determine this.

Example, basicly heraldry has 7 colors (term used broadly) divided into tinctures (red, green, blue, purple), metals (gold-yellow, silver-white) plus black. For functional artistic reasons there is a rule that a shield can not have two tinctures togather or two colors. So if you look at most flags - U.S. example it is described as red with 6 white stripes. and a conton as blue with n white stars, which follows the rules. However at a distance the flag could not be "read" if the stars were green, red, purple, nor the stripes of blue, green, purple.  While a computer could "read" such a flag, it could be taught a rule to make this not the case. However scientificly the human eye, would not have to be taught, because the issue is beyond its powers of discrimination.

A Scientificly minded person, nows that a rainbow does not have n number of colored bands, but an infinate number of shades. So when a person draws a rainbow, the selection of colored bands is a function of the cultural determined adjectives in their langauge. So if a language has 7 color words the rainbow has 7 corresponding bands. However in Africa are tribes with only 3 or 5 primary color words, so their rainbows, as drawn and percieved only have 3 or 5 bands.

The Greeks had a formula for beauty in human images based on ratios of measurements. So to a Greek or Roman, or Renaissance artist, violations of thse rules, equates to ugly, bad, etc.

So while I agree that computers can´t think, you over estimate what humans can do in this department unless they are programmed by their culture, and its values.

In the West, white stands for purity, black for death, and red for slut (dated around 300 a.d. when it was first determined that Mary Magdaline was a hooker). So brides are beautiful in white in the West, but Red in the East, where white represents death, as opposed to black. There was a time when women with huge breasts were considered more attractive then those less endowed. However under Disney rules, bad women have large breasts, and good women tend toward the post pubertal breasts. So giving computers rules based on this, depending on time and place, is a way of ranking beauty.

My basic point is that artificial intelligence can´t produce greatness or champions, but it is not that much different then the average human, for determining less great or ranking it.

HWPeeler (HPeeler)

From: HWPeeler (HPeeler)

May-16

Re: tonal quality

I can see a computer analyzing that if it is the only sound present. But when mixed with an orchestra how do you break the waveform down to isolate just the oboe?

RGoss99 said:

My basic point is that artificial intelligence can´t produce greatness or champions, but it is not that much different then the average human, for determining less great or ranking it.

Point made.

RGoss99

From: RGoss99

May-17

How can a computer differentiate w/i an orquestra. I am not a sound engineer, but they do it all the time, police with gunshots separating guns, explosions, ask any band techie, when setting up a system, no longer with cables, just a hand held computer, if they can mix sound, they can un mix it.

HWPeeler (HPeeler)

From: HWPeeler (HPeeler)

May-17

Mixing sound is a lot easier than unmixing it. On TV the gun shot is the only sound being analyzed. "Edited for television" applies here.

The tonal quality of a given musical instrument or sound depends on the "attack" (the rising of the wave form), "sustain" (how long the pulse is) and "decay" (falling time if the pulse) as well as the frequency )how many pulses per second) and magnitude. When dozens of musical instruments are added together it makes a mess of a waveform.

,Mt musical talents are minimal but even I can pick out just one instrument and make a good guess at what the music it is playing might look like. With multiple instruments playing the same notes you can't isolate just that instrument electronically.

RGoss99

From: RGoss99

May-17

Already responded to, simply repeating your opinion without any backup is a waste of both our time.

G=G+1

HWPeeler (HPeeler)

From: HWPeeler (HPeeler)

May-17

Just giving a description of what tonal quality of a wave form is in terms of frequency, wave shape and magnitude. A child can hear the difference without such an analysis and tell the difference between a violin and a oboe.

Unmixing??? Can you unmix paint? Once the oboe is mixed into a single waveform it is difficult to extract just that oboe sound.

RGoss99

From: RGoss99

May-17

Difficult is not necessarily impossible.

Yes, a child can tell the difference, but unless he is taught as a computer protrammed, he will not be able to define the difference.

TOP