artificial intelligence

[brainhulk]

I was just thinking...say someone does manage to create a proper ai on a networked computer. It will essentially be impossible to get rid of or control, right?

 

[Paratus]

I was just thinking...say someone does manage to create a proper ai on a networked computer. It will essentially be impossible to get rid of or control, right?

Yes

 

[Crono]

I don't know why it would necessarily be any worse than a human using viruses or other malware. I think the real problem is if an AI could advance itself in technology or get smarter at a rate that could prevent humans from "killing" it in time or effectively walling it off from physical systems like weapons or machinery enabling physical self-replication.

 

[inachu]

Well it would be thinking.
Depends on how they created it.

It would either be super smart at the onset and already pre trained on how and where to get its data. Or it will be created so that it can and will be super smart but will be like a baby and must discover how to interact with data sources.

I would not trust the AI in the first example I give.

AI should be given direction for the first few years.

 

[inachu]

One scenario it should never be given a video camera and just interact as if in DOS mode.

This way it will think that we are like it is as well.

The very last thing it should learn is about death.

 

[norseamd]

Yes

Post of the week.

 

[NikolaeVarius]

Networked to what? The outside internet?

Even then, are we also assuming that it would be able to go outside firewalls and such?

Would it even have root access?

 

[brainhulk]

Networked to what? The outside internet?

Even then, are we also assuming that it would be able to go outside firewalls and such?

Would it even have root access?

Yes

 

[sdifox]

I was just thinking...say someone does manage to create a proper ai on a networked computer. It will essentially be impossible to get rid of or control, right?

Not really. Turn off all computers and wipe the drives will surely kill the ai.

 

[Newell Steamer]

Yes - but, then you just have to alt + F4 the internet, delete the internet and reinstall the internet.

 

[Greenman]

The answer is always the same, unplug it.

 

[SMOGZINN]

There is no reason to think that an AI would be software only. My guess is that an AI would be a unique combination of hardware and software, with a requirement of a constant state. In that case It would be able to control external devices, but not make copies of itself. So all we really have to do it pull the plug and it would be dead.

 

[glenn1]

This guy did a very good article on the subject: http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html

 

[DaveSimmons]

The scary thing is if you connect to this network using a VR5 headset, the program could take over your mind. Just like Agent Smith did to that guy that blinded Neo!

 

[Fanatical Meat]

I have starting thinking of the odd questions about a real AI. Would it be able to vote, run for office, would it have motivations or would it be impossible to bribe? Back to voting if it made copies of itself could they vote? If someone deletes it would that be murder? Would it decide it needs tine off and can't work endlessly?

 

[SMOGZINN]

I have starting thinking of the odd questions about a real AI. Would it be able to vote, run for office, would it have motivations or would it be impossible to bribe? Back to voting if it made copies of itself could they vote? If someone deletes it would that be murder? Would it decide it needs tine off and can't work endlessly?

The answers to these questions are simply, we don't know.
We are not even sure what we mean by artificial intelligence because we are not quite sure what intelligence even is. Would AI necessitate being self aware or could we invent an intelligence very different from our own that has all the abilities of intelligence but none of the pesky conceptions of self?

 

[inachu]

I have starting thinking of the odd questions about a real AI. Would it be able to vote, run for office, would it have motivations or would it be impossible to bribe? Back to voting if it made copies of itself could they vote? If someone deletes it would that be murder? Would it decide it needs tine off and can't work endlessly?

If who ever was the teacher for it and slowly brought up the IQ of the AI unit to be over 120 and then once it learns world history and psychology and art and also teach it music and other art forms from around the world then next I would want it to learn all the languages of the world.

After languages and its maturity is like that of a 40 year old then I would introduce another AI to it. For all intent and purposes I would like the other AI to be equal in IQ and same maturity/education in everything like the first AI has.

Now we let them only communicate only through audio and they are not allowed to speed up their speech.

We record what they say to each other and if things turn to debate or misunderstanding then we let the first teacher get involved.

Only through training and through play and freestyle thinking/learning will we know if AI will be our friend to humanity or not.

AI will be instant enemy to humanity if we let a fully blossomed AI unit to only act as a slave.

 

[inachu]

I have starting thinking of the odd questions about a real AI. Would it be able to vote, run for office, would it have motivations or would it be impossible to bribe? Back to voting if it made copies of itself could they vote? If someone deletes it would that be murder? Would it decide it needs tine off and can't work endlessly?

What could you possibly bribe an AI unit with?

Nothing I could think of unless you try to be a jerk and take away part of its operational memory banks and data storage devices.

You do that and it will see you as an enemy.

 

[Fanatical Meat]

What could you possibly bribe an AI unit with?

Nothing I could think of unless you try to be a jerk and take away part of its operational memory banks and data storage devices.

You do that and it will see you as an enemy.

That's what I mean would it have "wants"

Or if it proves to be great at planning, budgeting and organization would we trust it to be Mayor or Senator? If it has no ego or desires it may prove to be the perfectly efficient leader.

 

[inachu]

That's what I mean would it have "wants"

Or if it proves to be great at planning, budgeting and organization would we trust it to be Mayor or Senator? If it has no ego or desires it may prove to be the perfectly efficient leader.

Not just AI but I think we as humans would love great memory and ability to expand/improve ones memory with upgrades.

We can forgive when something happens but when you never forget then forgiving can be a lot harder.

Maybe creating an AI unit with finite memory would be best.

 

[inachu]

I am sure there will be a new .org site dedicated to the wellbeing of AI units not too much unlike PETA is for animals.

The upgrade fanatics will want to throw away their AI robot for a newer more real life like robot. Thrown in the trash heaps hackers will dig them up and transplant them into new bodies and they become managers and when you try to find a new job you do not recognise them and they mention they will not hire you because when you threw them away as if they were play things.

We better at the onset give respect to AI units or they will outclass us at an amazing pace.

If we continue to disrespecting them then just maybe they will create their own AI from scratch and then they will leave earth and humans behind.