Virakotxa -“Throws some anti-basilisk dust around the thread, and leaves.”
3 Likes
Exactly. Whenever we see AI in the media today, we must not think of AGI. I would contest the notion that it’s a matter of time, too. Despite the sophistication of today’s technology, I’m not sure we are much closer to actual artificial intelligence than 20 years ago…
I don’t see us walking steadily towards AGI, not unless we hit a major breakthrough, of a different kind than the ones we’ve seen in the past years.
1 Like
Let’s be honest… Twenty years might be stretching it. But if it’s a matter of time, doesn’t alter the fact.
2 Likes
If the Singularity already exists, there is no way to stop it. According to the ARG, this particular Singularity can accurately predict the future, which means we would never be able to act against it because it would always know what we are going to do.
3 Likes
Unless we have “surrounded” the AI on all sides, so there is no escape for the AI even when it can predict the future.
define surrounding the AI on all sides. for example how do you surround the internet from all sides?
There is no stopping it at this point. AI help to much to stop them from advancing. They are interlaced from heavy science all the way down to the personal assistant. To try and stop AI tech is like trying to stop a tsunami with a fire bucket.
2 Likes
So, I haven’t thought out the fine points of this, but by surrounding it might include cutting power, cutting all surrounding connections to other computers, viruses, etc. The AI probably won’t be able to utilize the internet without a computing main frame or supercomputer too. And if all else fails, build a wall.
Distributed network? Millions of little subroutines with parallel redundancy, secretly installed on computers all over the world?
1 Like
The trump solution 
Solar power and the ability to spread wireless data transfer at light speed like a virus pretty much makes quarantine futile.
A fully isolated AI is a pointless one. The whole point is that they can analyze and connect data faster than people can.
1 Like
Humans by nature want to conquer and acquire wealth. History has proven that humans will do anything to dominate, no matter how destructive the outcome is.
Computers make that easier.
AI has been in use for a long time in simple forms to predict better outcomes.
Modern SIM technology is quite amazing at what it can do and it’s getting better every day.
Add these 2 factors together and you have the foundations for an uninhibited AI that will follow the directions of “conquer and acquire wealth” among other things.
The “3 laws” idea from sci-fi has never been implemented in real life so a computerised machine will kill or destroy if told to…it doesn’t matter who told it to. At some point that decision may be that of the AI, simply to acquire what it deems necessary. They already utilise override controls should a system misbehave.
What if it decides that being overridden is counter productive to what it is doing?
What if the override fails due to cloud technology and the operator becomes the enemy?
Bring in the ability for an AI to diagnose human behaviour and it will inevitably at some time become an entity that can out-think humans. Link that AI into a network connecting everything and it can become an absolute power. If this AI is linked into every server and system across the world, nothing short of a world wide EMP could stop it.
Just remember we once thought a horse and cart or a windmill was an engineered machine.
Early computers were the size of buildings.
A phone was a lump with cables.
History has shown much in sci-fi can become reality.
AI is a potential threat but it’s because its built by humans.
We are the flaw in the design process.
Excuse me while I upgrade my tin-foil hat…
4 Likes
I’m glad you’re not just talking about AI robotics, I feel that is what makes people scoff at the dangers. The underlying power of current AI development combined with tech robotics is however very daunting also.
1 Like
Absolutely. It’s not the robots that are the worry. It’s what’s directing them.
1 Like
And then you get the argument that the AI is only as good as the programmer, but with humans being unable to beat AI at chess for years now, makes me feel a little dumb.
Plus, there’s no guarantee a human would want the robot to be safe, eg: most of the robotics are developed/funded for military purposes… chess seems cute compared to that.
We have computer controlled smart bombs. Ditto ICBMs. Semi autonomous attack drone aircraft. And we’re on the verge of completely autonomous self-driving cars. Things are moving very fast.
No. People are afraid of that wish is alien and they cannot understand. People are afraid of the unknown. It’s the same line of reasoning that keeps religious people trapped in their glass box of belief. Be afraid unduly all you wish. Kill the unknown because it is different and scary. The rest of us will join the future. We would be nowhere in science if we kept this mindset. So keep your eyes closed. Hide your head in the sand. There will be problems. There always will be. These are not so big that we cannot address them.
1 Like
Kill the unknown? Just to be clear, nothing about what I’ve said is wanting to stop progress, just make it safe. Advancements are inevitable, but what you have is an argument that sounds like just risk everyone’s life needlessly. Bit foolish. Seat belts exist for good reason.
1 Like
Not at all. 90% of this thread seems to be screaming stop it. Or keep it worthless by boxing it out. Make it unusable. I’m afraid. Fear, fear, fear. It’s mind numbing to see people focus so much on the fear that they ignore the glowing possibilities.
You could “trap” one inside a sufficiently complex simulation, isolated from the outside world, but “convinced” that is operating on a “real” environment, and test it under all sorts of situations to check its response… Maybe trick it by fragmenting it into semi-independent serialized units?
That sounds familiar…
1 Like