What is a robot

Respect the first comments, a robot that do things completely self-referential like moving but totally autonomously without doing nothing else it is not a robot and can be classified in the range of the useless machines

So if a robot should do something, we have a couple of different approaches to find the answer.

The first is bare mechanical, so anything that can do a series of coordinate actions maybe considered a robot, including some complex yet efficient engines from the previous centutires. By this point of view for example a self-playing piano is a robot that we program with a punched belt then he can repeat what a human can do.

The second approach is that should do something of intelligent and interactive. I mean from the bare concept of answering to a question up to making very complex action sequences.

As the game theory teach us, we should never be distracted by the complexity of the situation, but the moving algorithm. By this point of view it is less "robot" a machine able to do thousand of actions with a super sophisticate program than a simple engine (i.e. a keyboard and a monitor) able to make complex inferences.

The final point in this though is that none of these can be considered robot in terms of "human like things". Independently by the complexity these are knowledge systems and not thinking systems. Cognitive psychologists can enjoy a lot but these are always lifeless dolls. So in a commonly accepted vision the large meaning of the term robot can be covered by any knowledge system able to manage correctly the links of its internal cognitive base (I don't use the term database because it is minimising the concept). Well all these are robots.

So we can define as robot in a generic definition as a mimic system with a - more or less - complex cognitive engine able to interact with the real world depending on some input conditions. No matter is are sensors, a keyboard or a banana. The point is that all these systems out of their context are simply (again) lifeless dolls.

IMO we use the term robot to generically distinguish fully controlled machines (like an excavator) by autonomous ones (in specific and restricted conditions). What I consider as robot it does no matter, my opinion is that this is one of those terms that should be reviewed and their meaning rewritten periodically to update the word to the actual context.

How do you think that the term robot will match with an experiencing system instead of a knowledge based ?

Enrico

And where do AI systems fall into this, for example IBM's WATSON jeopady playing computer, is this regarded as AI or simply an extreamly good search engine

I agree that the word ROBOT has been overly generalized to mean almost anything with a microchip that can move and I too agree this term needs to be better defined as I love ROBOTS but dont think my toaster or coffee maker should be called one, a cleaver machine, programmable machine etc, perhaps.

Even my CNC project with all its computer controlled parts with movement is not a ROBOT, but it is a programmable machine

Peter

What if I have the definitive AI algorithm ?

Oh, if it passes the Turing Test then it is regarded as human.

Oh, if it passes the Turing Test then it is regarded as human.

Maybe. But it's meaningless... I know many humans that don't pass it

BTW: Do you remember Eliza? Do this passed the turing test ?

Enrico Miglino wrote:

BTW: Do you remember Eliza? Do this passed the turing test ?

Some believe it did.

Yes, I am old enough to remember Eliza. In fact I programmed a version in snobol. I don't think that it passed the turing test. AFAIK no programm passed. Some claim that their programm did but this is allways disputed. There is the Loebner Price for the best chatbot of the year, the contestants manage to convince around 1/3 of the judges that they are human. So I guess that's the state of the art.

The citation of Eliza was almost provocative. The point - IMHO - that Eliza & sons CAN'T pass the Turing process is that all the AI orientation is mimic. The reason, I think, most depends on the implicit awareness to be not able to do better. So if we can't solve the problem, we change the terms of the problem in a more solvable form. Remain the fact that the first AI studies started from mimic, reached the highest theorisation levels around the mid of '80s then a general empasse and the fast growing of the world of fast computing technologies and real world data acquisition through always cheaper and sophisticate sensors opened the door to a great lie, almost globally accepted as the truth: mimic machines are sufficient to do what the industry need, so we call robots, include strong programming behaviours and decide that this is AI.

I simply disagree and have my personal theories. The point, just to close with a question is: how much is risky and dangerous in the today technology scenario if someone really setup a silica-mind able to take really autonomous decisions based on what can learn making abstractions ?

Enrico Miglino wrote:

 

BTW: Do you remember Eliza? Do this passed the turing test ?

For a wonderful Eliza reference, check out George Lucas' THX-1138 (1971).  In Lucas' dystopia, the hero THX-1138 is feeling troubled and goes into a booth to talk to the diety OMM, who replies to his concerns in exactly the same way as an Eliza program.

John

not only, but TH1138 really is a great sf novel by Ben Bova I have read at the end of '70 published in Italian translation by the editor Mondadori in Italy in the historical series of SF novels Urania. In the novel this part is better represented as it is possible only in written stories.

Enrico

When I was at college I played with a piece of software that ran around a MUSH I tweeked it to run an Eliza like programme and had great fun watching people chatting with it. So I sent my mods back to the creator who turned out to be Marvin Minsky!!

I was so amazed that I got a reply that I kept it, it will be on a floppy disk somewhere