Peter Norvig: Singularity is in the eye of the beholder

Such questions reappear on higher levels when you're reasoning about AI. But since the field hasn't been grappling with them overall, I'm doubtful they'll come up with great insight at this point.

We programmer do it instinctively, but having read about it gives me such a greater confidence to actually apply it methodically every time i design a type hierarchy...

(*) basically, the idea is to mutate the property, and see if the object's essence is unchanged. If it is unchanged, then the property isn't constitutive of its essence. As an example, changing the color of a table doesn't change the fact that it's a table, so "color" isn't an attribute of the "Table" concept. Number of feet, however, could very well be (or at least a boolean "has feet" ;)

That's pretty much what Plato believed, but also that for the most part we can never access that Platonic Ideal. We could only, in effect, access the equivalent of shadows on the wall case by light against objects. Seeing only shadows for all of our lives, we believe they are the true reality since that is all of reality that we perceive.

As a concept it is a strong precursor (and no doubt a strong influence) on Immanuel Kant's work. He basically pointed out that we have only 5 senses, and each those are intermediated by various layers, and so even through those sense we do not experience the thing in itself, and are limited by those 5 senses. And of course we know that other animals have other senses. We have invented some of artificial ones of our own (vision that is heat based instead of light based, etc).

If you're interests go in that direction. his work Prolegomena to all Future Metaphysics is where he begins to explore this. It's dense, but not too inaccessible as philosophical texts go especially if you have the background in logical reasoning and layered abstractions that programming instills. Here's a link to a free Google Books version that also allows PDF download: https://www.google.com/books/edition/Kant_s_Prolegomena_to_A...

Incidentally, Kant was 100% correct: His ideas were so compelling that pretty much any philosopher after him looking to explore metaphysics could not simply dismiss them out of hand.

Secondarily, Prolegomena was also somewhat of a response to work by David Hume on the nature & human perception of causality, and together they formed the foundations upon which science has continued develop that area of physics, even if it has moved on somewhat from those earlier ideas.

I think Philosophy often gets a bad name today as a useless of self-indulgent field, but it's important to remember that philosophers were in many ways the first scientists and refined the ideas & practices that ultimately developed into the scientific method, breaking off into a separate (but still connected) branch of study. For modern examples where that synergy still exists, the works of Danielle Dennet are an excellent example.

If you rephrase the question as, "Is there such a thing as a more real table than this table?", or if "Ceci n'est pas une pipe" sparks joy (or puzzlement) in your heart, you might enjoy reading Neal Stephenson's Anathem.

I can't write why, for spoilers. I can say the book is heavily inspired by Platonism, as well as many other things. If a science fiction story about philosophical monks, astronomy, and adventure appeals to you, give it a read.

I don't understand. The "idea" of a table is an object that provides a level surface to place stuff on.

What am I missing? What is the interesting part?

Heidegger would probably say something like "the table tables" (more on brand in German "Der Tisch tischt") - so a table is a thing that supports tablelike actions.

Of course "table-like" is a gradient instead of a binary flag. An archetypical four-legged table is the way you might draw a table when asked about it — so it might — for that individual living at that time in bistory be as table-like as it gets. However, there are totally recognizable tables with just one leg, or even with no legs (e.g. carved into a stone wall) or with alternative ways of suspensions (hanging from the ceiling, mounted on the wall, etc). In the end even a rock could act as your table when you are out hiking. Of course you still recognize it as a rock, but it has shown table-like qualities now, it can be your table for the moment.

Then there are also different kind of tables: work benches, kitchen tables, desks, coffe tables, long and thin tables, more akin to shelves, but used as tables, folding tables, tables with rolls on them etc. They can totally differ in their proportions and usage, yet we call them tables, mostly because ofhers will know which object we mean, when we say it. When you however demanded a table at work and they gave you a tiny coffee table you would exclaim: "This is not a table!". Rightfully so, for the context you are in it is not useful (it differs too much from an office table).

So if instead of asking if something is a table, asking whether something can be used as a table (if it has table-like qualities to it) is more like how I think object-subject relations work in practise. We call things a table because we use it as a table. We don't call things a table because somewhere out there in the universe exists a clear definition of what the concept of said table constitutes.

If you were to shoot a film or direct a theatre play, then you wouldn't be interested in whether in principle the object can be used as a table, you would be interested in it's readability as a table that hints to the viewer a certain millieu or social background. This might also be something you'd consider when the primary purpose of your table/desk is to impress your subordinates.

Of course the table-like quality of an object would be just one of its many dimensions. One could also go in and ask about arbitrary dimensions like the bulkyness of the object, about the intentionality (that hiking picnick rock is a table, but an unintentional one). One could ask about many aspects that constitute subject-object relations, however it is important not to forget that in our daily lives we just ballpark these categories intuitively. So what I might still call a table might not at all be seen as a table by another person ("This is not a table, it is a rock") and yet both of us a right and wrong at the same time.

Interestingly AI goes a lot into looks, where the dimension of usability might be more useful (but harder to extract from images). So internally an AI would represent table-likeness as a floating point number and only says something is (or isn't) a table based on thresholds somebody found useful.

I suspect that computers will be vastly superior to humans in many, many tasks long before we acknowledge that the singularity has already happened.