Artificial Intelligence 2020 - is there progress? - page 41

New comment
 
Реter Konow:
That is why I do not believe in mathematical formulation of human logic. It is non-linear and "warped" by a mass of unpredictable nuances.

oh

Here's the simplest example.

they used to add, multiply, divide, etc. on paper, in their heads, and it was OK.

now who wants to live in the past?

They use calculators, programs

 
Dmitry Fedoseev:

Another property is missing. Some dead humans don't have heads. So some dead mammals don't have heads.

...

So, in order to get a correct logical conclusion, you have to add ALL the details to each fact?

1. Some dead humans don't have heads.
2. Dead people, biologically, are mammals.

3. Some dead mammals don't have heads.

And what is the value of such a conclusion? The reasoning about dead mammals seems absurd. This logic will become even more ridiculous as the private properties of an instance continue to accumulate and be transferred with all the details to a niche of its category, which, is not adapted to accept all this informational "junk".
 
Dmitry Fedoseev:

...

It's not about logic, logic is as ironclad as a machine and everything is clear and unambiguous with it. The point is that one does not live by logical thinking alone.

Right. He distorts it right and left. In fact, he has so many logics that fight each other.
 
The properties of a biological species must be collected from its instances by "aggressive" generalisation and discarding all low-priority material, through selection and filtration. In this case, we will get a pure category with a set of correct properties. Otherwise, mathematical logic will cram a bunch of uncollected rubbish in there, mixed with the right parameters.
 
Another mistake of cat-tail logic:

1. Some cats do not have tails.
2. Cats are mammals.
3. Some mammals don't have tails.

4. Humans are mammals.
5. Some humans don't have tails.
6. The rest, therefore, do.

P.S. Bottom line. "Straightforward" logic is not good for building AI.
 
Реter Konow:
5. Some people don't have tails.

Not necessarily, Peter. Some mammals don't, others do. The words "some mammals" don't say anything about individual species within.

 
Aleksei Stepanenko:

Not necessarily, Peter. Some mammals don't, others do. The words 'some mammals' say nothing about the individual species within.

It's about creating a logic engine based on mathematical formulas (if I understand the daughter of that inventor correctly). He simply showed that the task of constructing the field of human knowledge by the engine from formulas scouring book volumes is not solved and instead of the correct data piles of rubbish will be collected, which then have to be "manually" cleaned up).

Pure categories and correct knowledge will not emerge.
 
I get it. It's complicated. It's often hard for a person to make sense of themselves. How can one create a mind?
 
Aleksei Stepanenko:

Not necessarily, Peter. Some mammals don't, others do. The words 'some mammals' don't say anything about the individual species within.

If we collect information about the category "mammals" from many specific species, not all of their material can be used. After all, we further extrapolate the properties of the category to other species to fill in the gaps in knowledge. But if the category has rubbish (such as data on the presence/absence of tails), then we will project it onto unknown species, making gross errors.
 
ORACLE11111:

The AI is a data conversion tool.

Here's a one-minute EUR/USD chart forecast for example
Dynamics and price seems to be guessing.

Have you managed to get any real earnings?
1...343536373839404142434445464748...74
New comment