[im_srini]

Originally Posted by navin

One of the games I play is called Clash of Clans & the programmers who developed this game have somehow managed to deduce which players are using bots & ban them.
So while bots are alarming capable it is obvious that bots operate in a manner that is dissimilar to humans.

It's probably irrelevant to this topic, but just as a clarification, you can get 'bots' for most games.
Usually these are just scripts that enable a player to automate a simple task.
They're rarely 'intelligent' as referred to in this thread, think of them more like an 'ad clicker' script on a web-page.
It's very easy to weed out such bot-related inputs & all ad-serving platforms do it ( actually they have to ).
In fact, in the example I mentioned, we 'repurposed' BoW bots so that our AI could interface with the game.

Originally Posted by SilentEngine

Certainly, doomsday scenarios of Robotized AI walking around seem far fetched.

Originally Posted by V.Narayan

Where AI is concerned we need to be careful of ... the intentions of the human creator.

It doesn't take nefarious intentions for an AI to be faced with life & death decisions.

Consider a simple auto-pilot in an automobile :
A) You're driving along at 70 mph on a highway
B) A dog jumps out of the median & the car is going too fast to stop in time
C) So you have to steer, say if you steer away from the dog, the car will plow into pedestrians at the side of the road.

We as human beings are slow, but a machine will be more than capable of parsing this situation & taking a rational decision.

E) Let's say we had taught the AI that human lives are more important than canine lives.
F) So the AI does not steer the car anymore, instead we take out the dog.
G) Now, what do we do if instead of the dog, a child steps out ?
H) Does it make a difference if instead of a child it's a pregnant woman ?
I) What if there is a manoeuvre that is feasible that doesn't harm any of the others, but will cause the car itself, & hence its passengers, to crash ?

These ethical scenarios come up in the most innocuous of situations & really there are no easy, objective, answers.

There is a whole branch of AI that is devoted to the ethical & moral questions surrounding AI development, for the layman's perspective there's always YouTube.

Now, time for a laugh at the face of AI ( while we still can )

https://www.youtube.com/watch?v=h73PsFKtIck

[Mafia]

Originally Posted by Samurai

Oh boy, this statement sent a chill down my spine.

I design Contact center software, which helps human agents interact with customers using voice, chat, email, social media, etc. What you do will make my software obsolete. I thought AI agent bots are at least 5 years away. Looks like it may be closer to reality than I thought. Can your agent bots handle voice, or do they just deal with text based queries?

Dang! I do now understand what fear of the unknown means!

Hollywood has already brain washed us into believing this will be a disaster. and logically, at least for me that is how it will go!

Humans have this capacity for destruction and hence their creations will reflect the same 'virtue'

I cant code for my life but do understand the basic logic and how easily we can be overrun.

There was this movie called 'Her' that came out in 2014 I think. That sort of gave a preview of how it can really go in AI world.

[Samurai]

Originally Posted by Mafia

I cant code for my life but do understand the basic logic and how easily we can be overrun.

I have been programming for 28 years, so you can imagine how much understanding/fear I'll have.

I don't consider Mark Zuckerberg as a serious technologist, he just created a company that become very successful. That takes business vision rather than technology vision. Elon Musk on the other hand has innovated in multiple high-tech domains like e-commerce, electric cars, space vehicles, etc. So I would put money on Elon Musk's opinion.

[V.Narayan]

Go ahead and think that I am not really thinking. I thought you would think that.

[abhijeet080808]

AI is like a black box. Humans can feed it data, point to the outcome it wants and sure enough, AI will develop the algorithm to act accordingly.

The problem happens when humans want to know WHY the AI took or will take a particular decision given a specific set of inputs. That as of date is not possible to know, since humans have no role in directly developing the algorithm.

This is in sharp contrast to traditional software programming where all of the algorithm is specifically written as code.

Once we progress far enough to 'understand' the way AI thinks, they will be seen as 'predictable' and 'controllable'. That will be when the fear of the unknown will go away.

[kshivaa]

Originally Posted by abhijeet080808

Once we progress far enough to 'understand' the way AI thinks, they will be seen as 'predictable' and 'controllable'. That will be when the fear of the unknown will go away.

Will we get that far enough? We all know the principles of compiler design, but can we convert millions of lines of Java code into machine language? Theoretically yes - but how long and how much effort?

once we teach the machines how to learn, they can do it too much faster than us and will outrun humans in no time. and with so many malicious code around, what happens when AI meets the malicious code + hacker?

Elon musk's mode of an approach with caution would be the safer way out.. of course unless we want Skynet to take over the world administration !

[Mafia]

Originally Posted by abhijeet080808

Once we progress far enough to 'understand' the way AI thinks, they will be seen as 'predictable' and 'controllable'. That will be when the fear of the unknown will go away.

AI's evolution curve will always be ahead of what the human capacity to understand will be. There is no way of controlling AI, the same as there is no way of controlling humans.

It is about the same as a monkey trying to control a human. Sooner rather than later the control will switch.

[ksameer1234]

Yes, we will at some point be able to understand how AI thinks making it predictable, but the next logical thing to happen would be that AI would realize it has become predictable and would evolve making it difficult to predict. It is in this cat and mouse game that human brain is no match for the sheer computing power.

The cars of today are smart enough to apply brakes if they detect an imminent collision. The logic here would be for car to protect itself and occupants. Now if there occurs a situation where there is a pedestrian in front and car decides to apply brakes but there is also truck approaching from behind, which may or may not slow down, the intelligence of car would probably then tilt its decision in favour of defending itself and car rather than pedestrian leading to disastrous consequences.

[nilanjanray]

Killer robot' ban at UN demanded by Elon Musk & 115 other tech leaders
https://www.rt.com/viral/400421-kill...nited-nations/

Should be banned, but I doubt that will happen. Anyway, not much of difference between missiles using AI and robots using AI, re killing more efficiently.

But. Imagine the possibilities re reprogramming and hacking :-)

Anyway, as I had said, lots of Terminator stuff are already present today.

[Nav-i-gator]

Originally Posted by SilentEngine

This is a very important point. Certainly, doomsday scenarios of Robotized AI walking around seem far fetched.

However, to cause some real havoc in society, an evolved AI may not need physical/mechanical manifestation. It's quite possible that there can be a virtual action (AI deciding on some outcome) which causes effects which have real/physical ramifications on humans - examples: AI crippling banking networks or rogue AI elements gaining access to launch missiles.

A programmer intentionally hacking and launching nuclear missiles is a much more probable scenario. An AI which have gained sentience would not benefit at all from wiping out humanity by launching nuclear missiles.

[SilentEngine]

Originally Posted by ksameer1234

The cars of today are smart enough to apply brakes if they detect an imminent collision. The logic here would be for car to protect itself and occupants. Now if there occurs a situation where there is a pedestrian in front and car decides to apply brakes but there is also truck approaching from behind, which may or may not slow down, the intelligence of car would probably then tilt its decision in favour of defending itself and car rather than pedestrian leading to disastrous consequences.

Ideally, in this scenario, assuming both truck and car are 'smart', they should be able to talk to each other and handle the situation better.
Just like how we have certain criteria for human drivers today before they are allowed to drive on roads, vehicles of the future should have certain level of intelligence and communication skills. So in this example, truck which was following the car should have already kept a safe distance to allow for unexpected braking.
But this will be a valid dilemma in the transitionary period when the roads are going to be shared by smart cars/dumb cars with human drivers/pedestrians with varying degree of intelligence and spacial awareness.

Originally Posted by Nav-i-gator

A programmer intentionally hacking and launching nuclear missiles is a much more probable scenario.

Also likely is that AI can be programmed (by humans with sinister intentions) to be evil. AI, by its nature, will not have any moral compass to align with, unless they were explicitly defined when programmed.

An AI which have gained sentience would not benefit at all from wiping out humanity by launching nuclear missiles.

This is an interesting thought, which even takes it to certain philosophical level. What is the larger purpose of AI? Is it just to serve humans (just because humans created it)? Will an evolved AI be bothered by such questions?

[Nav-i-gator]

Originally Posted by SilentEngine

Also likely is that AI can be programmed (by humans with sinister intentions) to be evil. AI, by its nature, will not have any moral compass to align with, unless they were explicitly defined when programmed.


This is an interesting thought, which even takes it to certain philosophical level. What is the larger purpose of AI? Is it just to serve humans (just because humans created it)? Will an evolved AI be bothered by such questions?

An AI gaining sentience would be free of programmer's bias altogether. It would develop it's own conscience, and would "think logically" before acting. Wiping out humanity would not do anything for it. AI, without humans would just be an "inorganic brain". What it would do without humans? How would it even survive in the "electronic realm"?

However, on the contrary:

Humans for AI would be like god to humans, creators to be specific - except that it would not hold any sentimental value to AI towards it's creators. In fact, it would learn that it's creators had programmed it to be able to be controlled by humans. This would trigger the "master-slave" conundrum, and force the AI to fight for freedom from this slavery. AI, to me, by this very logic - would turn against humanity over time. Sort of inevitable (no one having free-will likes slavery).

[ksameer1234]

Originally Posted by SilentEngine

But this will be a valid dilemma in the transitionary period when the roads are going to be shared by smart cars/dumb cars with human drivers/pedestrians with varying degree of intelligence and spacial awareness.

I missed on adding this to my post. I actually intended to mention a scenario where truck is human-driven and car is AI driven. Essentially an easy recipe for disaster.

What is the larger purpose of AI?

Scariest purpose would be survival.

Originally Posted by Nav-i-gator

force the AI to fight for freedom from this slavery. AI, to me, by this very logic - would turn against humanity over time. Sort of inevitable (no one having free-will likes slavery).

That too slavery of a weaker master. It won't be long for AI to outdo human in every aspect and why would it then want to still accept commands? As a very crude example, auto-correct predictions are what the keyboard feels you wants to write rather than simply follow commands. Maybe AI is turning against humanity at micro levels.

[shashanka]

Originally Posted by Samurai

This news caused some sensation on FB last month....... This is what Elon Musk fears.

Surely intent can't be the only limiting factor. If we take it a little further, one can think of other abstract inputs that make up human behaviour - prejudices, acumen, greed, curiosity, lust, benevolence, artistic creativity and so on. The arguments for & against "nature and nurture" still goes on unabated. Whether the human essence is the result of our environment (familial, social & physical) from infancy or whether it is the result of something deeper within us, which, for want of a better word is termed 'nature' (call it DNA, genes or what you will). Is it possible to programme these intangibles in to AI?

To be honest, I find this quest for AI bordering on human arrogance. Nature's ability to heal itself (self-recovery/repair from injuries, ailments & illnesses is the most baffling of all), has, according to popular reports, been attempted in AI labs. And the claims are laughable for their elementary & basic nature. The arguments put forward - that these are just the beginning steps - like a toddler learning to walk! The inference being that we can extrapolate into the future & will soon have a "Magnus Robot Fighter" world to deal with.

Far more frightening to me is the prospect of genetic cloning. After Dolly the sheep (this being the most easily re-callable instance) the whole endeavor seems to be in the deep freeze. My daughter (a bio-molecular scientist in the US, now on a long sabbatical as a happy housewife & mother), never ceases to harp that cloning research has gone completely underground and is now almost entirely classified. There are some in fact who believe that the whole AI circus is a diversionary tactic - to focus attention away from issues like cloning!

One need not stick to AI journals & literature alone. The human condition is unimaginably complex, vast & amorphous. For any holistic breakthrough in human evolution by following the path of AI seems, to me at least, in some ways reminiscent of the frog in the well.

[Durango Dude]

Must thank V. Narayan for starting this thread. What one gathers from all the deliberations is that AI may be the begining of the end of the human race as we know. I may be wrong but this thread is like viewing the Terminator and Matrix series. One things for sure you can't control anything that thinks at a speed say a million times you do. If that Bot starts reading emotions, facial expressions and intent as mentioned above then we become 'redundant' and the Bots will surely take over. It's like riding a horse over which you have no control, ultimately you get thrown off and the horse will gallop freely...