Original topic -

"Of brain and the usefulness of studying brain in designing/building AI"

My opinion on that original topic -

The word “brain” is mostly marketing when people talk about AI. I think in the near future, “neural networks” could be replaced by “quantum networks” as it would sell better.

https://en.wikipedia.org/wiki/Brain#History

If 380 years BC, Aristotle was given a computer, would he be able to program a good chat bot? A chat bot that is at least as good as our current chat bots using neuronal networks? Teach Aristotle only how to program, don't reveal to him what brain does. Keep telling Aristotle that the only thing brain does is cooling blood. Would Aristotle be able to program a good chat bot?
I think yes.

Take heart for example - it could be replaced by various types of mechanical pumps. Each of these pumps using various principles of liquid propulsions. Does it matter how it pumps blood? It moves blood, that's what matters. It does not need to perfectly repeat the priciple of a real heart.

"Brain", “quantum”, “Mars” are just buzz words used nowdays to sell.

In my own opinion.

Not to mention that, repeating something and inventing something are two different things that deserve different merits.

Hypothetical case 1 - “Scientists managed to create an exact copy of a human brain made up of synthetic materials. And it thinks like a human.”

Hypotetical case 2 - “Scientists INVENTED a new approach to intelligence different than human brain and it is as intelligent as humans.”

Which of the two cases is more awesome?

In my own opinion it is the second.

Still, nowadays, some people try to sell inventions like that - “Our flying drone imitates the movements of a bird/bee. It is awesome because of that!”
What is awesome is how humans invented blade propellers rotating around a fixed axis. There is not such a thing in nature and we humans should be proud of our genuine invention as a species. But instead of that, people still try to copy and repeat. To plagiarize.

It is all related to the original topic - learning how brain works in order to be able to imitate it, or just completely ignore how brain works and create something completely new.

In my own opinion, second is better.

https://www.forbes.com/sites/fernandezelizabeth/2019/11/30/ai-is-not-similar-to-human-intelligence-thinking-so-could-be-dangerous/#1636a7926c22

Zero intelligence inside the code, just a mere tool -

https://link.springer.com/article/10.1007/s11023-019-09506-6

This is what our modern NNs and ML actually are -

A mere tool without its own intelligence, that we can tune to fit a task better than other.

In my own opinion.

Maybe the problem is people try to imitate a brain. Maybe this is the reason why humans are stuck with NNs. On YT i chatted with various users who preffer “a Logic Unit plus a Database” approach to AI. So, I am glad that not all of humanity is obsessed with NNs. Only 99.99% of humanity.

Note - I am not banning NNs, and ML. I am just stating the truth the way i see it. Maybe in the future somebody makes a slight change in NNs that changes it all and creates real intelligence. Who knows…

In my own opinion.

NikiTo said:
and creates real intelligence

This is a core issue brought up with every beginner in the field and re-iterated as you learn more.

Nobody can answer that. Not even experts on the fields can agree what that is, even in general terms.

As one example, if I train up some computer software to emulate a creature, and the emulation is so complete that an observer cannot tell the difference between which is the real one, is that intelligence? Avoiding the entire humanity thing, if we made a computer-simulated cat, and the virtual cat behaved indistinguishable from a real life cat, does the computer-simulated cat have actual cat-level intelligence, or simulated cat-level intelligence, or merely rules run by an automaton? When the behavior is indistinguishable, how is an automaton's behavior unintelligent when the actual creature's behavior is intelligent?

The question touches a bunch of fields, from sciences like mathematics and computer science and biology to psychology and even to debatable subjects like philosophy and ethics. There are plenty of discussions and topics, but no solid answers.

Go back to the 1950's and you get Alan Turing's famous test: he sidestepped the question of intelligence and went with the task of convincing an interviewer that the respondent was human. Another is the “Chinese Room” experiment, where even if a machine can convince a human that they are an actual Chinese-speaking human being that satisfies Turing's test, does that mean it literally understands Chinese, or merely simulates the ability to understand Chinese?

Nobody has answers to them yet, nor do we collectively fully understand the questions.

The topic of the thread is AI so Computer learning is not far off the subject.

Nobody can answer that. Not even experts on the fields can agree what that is, even in general terms.

I`m not absolutely sure this is what he means but you may call real intelligence intelligence that `feels` real, the type of feeling that would deceive a child.

Calin said:
but you may call real intelligence intelligence that `feels` real,

If an AI tells you to feel real and being alive, and you believe it, then the AI has only passed the Turing test, but the questions remain.

Hint: Ignore those questions. A researcher working on AI has more relevant questions to answer during work. so it's just distraction, more likely topic for externals like talk shows or law makers.
The more we focus on discussion of such philosophical questions, the more we expose our self as AI noobs who find it an interesting topic, but won't contribute to get it done most likely.

JoeJ said:
but won't contribute to get it done most likely

I don`t think I`m going to be directly involved in developing a terminator however I do think there is a place for a new level of intelligence (real intelligence if you want) inside video games

frob said:
This is a core issue brought up with every beginner in the field and re-iterated as you learn more.

If a beginner brings this to AI, it is a good thing for the beginner. Making definitions clear is one of the core problems in AI. Defining, what intelligence is, what a goal is, what a recompense is etc. It is something that needs to be done in the beginnig.

For example - recompense. With the case of a living organism we can easily define and understand recompense. We train a dog with recompense - food. It is easy to understand by any living thing - recompense of food.

But recompense in software - is it real recompense the computer can not live without - food, or it is a codded recompense. A dog can not live without food. But an AI theoretically can overwrite itself to stop respecting your recompenses.

Every concept needs a clear definition in the field of AI. And we don't really need to repeat biology here. I think we need to accept once and for all, that computers are different “beings” than humans. We need to accept computers are not motivated by food.

If somebody wants to precisely repeat a human in software, i think he should better quit programming and move to microbiology. There he can create, while following and using the biological laws.

Lot of people are trying hard to repeat a human. And this is what moves modern marketing.

I am very confortable with elf ears and cat tails in humans.

Having cat ears, in my opinion makes it better as it can help the robot/hologram/NPC express its "feelings" beter -

Why limiting ourselves to human ears?

Time in the link-

It uses its ears to express itself better -

Why any living thing loses its mind when it comes to dating. Everybody loses its reasoning when in love. Hormones overwrite intelligence. Maybe because knowing how complicated is it to grow a child, maybe the female knowing how painful is to give to birth, males and females would never date. And the human race will go extinguished. But nature though about this - it gave us hormones and alchohol. Do we really need hormones? Dr. House said we need pain in order to accidentally not poke out our own eyes. So, why we try to recreate feelings and senses in AI that a program/robot naturally does not need at all.

Why we need to program a complicated “fear of dying feeling” into an AI in order to prevent it from killing itself. When we can just put the OFF Switch on the back -

Why should we hit our heads into the wall trying to figure out how to program something AI would never need. Faling in love? Why an AI needs that? Falling in love means to put a person over another. This leads to injustice and wars. Why an AI needs to fall in love? It can simply clone itself, no need to gift flowers on a dinner.

And why human brain is considered to be the best example of intelligence. Maybe somewhere in a far far galaxy, brains of Aliens work in completely different way.

A brain naturally thinking in multiple dimesions - https://www.youtube.com/watch?v=_Ok3LOXvOYg

We can not know how alien brains work, but we can consider that a combination of a quantum and classical computing could be better creation than human brain.

A single instance of AI can use the whole information ever stored by humans and it can use al lthe computation power of humans. So a single instance of AI EFFECTIVELY can solve problems we humans were trying to solve for decades. All of our scientists tried to find a cure for ageing for centuries. AI can take into consideration ALL the human knowledge and put it in one single thought in order to discover the cure in no time. No human can see all the knowledge of humanity as a big picture. AI should be able to do it.
Who cares if it is an artificial cat if it is 1mln times more effective at solving problems than a real cat.

Beyong the Turing Test, we should start talking about soul. And soul is not proven either.
"If it jumps like…" versus "Humans have souls".

(What about a single brain with wired access to all the information at once? Who knows…)

We as a species often picture aliens as us, we picture AI as us, we even picture god as us. Humans are narcissistic.

There are other kinds of intelligences we often skip over.

Calin said:
I don`t think I`m going to be directly involved in developing a terminator however I do think there is a place for a new level of intelligence (real intelligence if you want) inside video games

So do i. There is always a place for progress in every aspect of video games. No matter if it's gfx, physics, sound, game design or AI.

And i have clear visions about the former things in the list, but i lack so for the latter things. I wish we could at lest define some goals. Without a problem, no hope for a solution.

It's clear AI is a goal, but we lack definition about what intelligence exactly means. So we need to narrow things down.
I remember, in older games characters may run in circles, or even against walls. I have not seen such issues for a very long time. They are probably solved.
But are NPCs still dumb? Do they miss something? Is it only about decorative behavior, or about limited gamepaly as well?

I noticed some people are concerned about path finding limitations e.g. to be restricted to planar maps. But i can not agree because graphs are independent of spatial dimension.

Another topic is computer vision for NPCs, so they have a improved sensing. I can imagine this would help with more complex behavior, e.g. considering cover and target, without a need to flag a desk as ‘i could serve as a cover, dude!’.
But i sit worth the computational complexity? How exactly, and in what situations would this improve gameplay?
There are some experiments with AI playing Super Mario or Doom. Do those AI players feel more natural, more interesting, more intelligent than regular game AI bots?

If so, when does this happen, and how could we improve our bots to get such advantages within practical performance budgets?

What's the research results in this field? Can we list some papers as reference? Can we show examples of potential improvement, to build a up a vision or goal?

We surely could, but we do not even try to do so.

I hope you get what i mean. We will not learn anything from discussing Terminators and Robocops again and again, while some other folks actually work on real problems, with realistic expectations, achieving results.
While we hang out here, just talking, we probably even miss the results and ideas of such folks. Because instead looking up for them we just talk about philosophy and far fetched fantasy.
Think about it. May there be some truth in what i try to say?
If so, it is about time to get rid of some bad habits. And what i mean precisely is your saying: ‘I can not learn from people i do not know personally’ (or something similar).

If you are interested in AI, the way forward is to adopt common knowledge in the field of ML or game AI. Only after understanding how it works exactly, you can see the exact limitations and open problems.
Even if your goal would be to come up with something completely different - as along as you do not know what this is, spending time on learning state of the art is always a better investment than ignoring it because it's ‘not good enough’, or because papers are hard to understand.
I don't know what's the reason for your quoted saying, but the reason can only be a bad one, leading to unnecessary obstacles.

JoeJ said:
And i have clear visions about the former things in the list, but i lack so for the latter things. I wish we could at lest define some goals. Without a problem, no hope for a solution.

well for starters you could emulate the multiplayer experience in games. In FPS games bots are pretty good at imitating a human player (but even here there are limitations, half life went some way in creating an assistant that would follow you around the level but it wasn`t that sophisticated ) however RTS games don`t have anything close to a realistic human like bot. Same thing goes for RPG no AI to match a human player. Once you get to a new level of AI you realize that the RTS gameplay can be evolved as well.