If I Were An AI : A Thought Experiment
Introduction
Recently my posts have been quite long and weighty tomes.
We all have limited time so I have decided that I will try to intersperse those types of posts with more fun topics and shorter posts.
One subject that I always find fascinating is Artificial Intelligence.
Barely a day goes by that we don't hear about it in the news.
The Bankruptcy-Bloodbath Hypothesis
![](https://steemitimages.com/640x0/https://regmedia.co.uk/2015/06/05/ashlee_vance_elon_musk_1.jpg?x=1200&y=794)
Image of Elon Musk from The Register UK
Many modern luminaries and tech celebrities have spoken out about the potential existential risk that AI presents to us [1,2].
The risk is usually described as twofold:
First the AI will take our jobs and destroy us from an economic standpoint, then the AI will probably decide that it doesn't need us around anymore and get rid of us altogether.
I would call this the bankruptcy-bloodbath hypothesis (BBH) for short.
Could AI Already Exist In Secret?
We know that private companies and military contractors often have experimental technology that can be years ahead of what we have in the (public) commercial world.
Is it possible that some private corporation or government agency already has an AI with human level intelligence or greater?
It is pure speculation and likely a stretch but sometimes imagining what might be can be a fun exercise.
That is one of the reasons I love Science Fiction so much.
Electrochemical vs Electronics
One of the issues with AI is that in theory it could "grow" it's own intelligence at an exponential rate.
This has been referred to at various times as the concept of the singularity or super-intelligence [3].
Even if you remove technological scaling and Moore's law, there are other reasons why it can have a big advantage which could then inevitably lead to a "singularity".
Human neuronal transmission is electro-chemical in nature. At it's very fastest it is several million times slower than transmission within electronic devices [4,5].
Imagine if you could do a 100 times as much in one day, then a 1000 or even a million. You would have a huge advantage.
Even without any kind of structural improvement an AI of human level intelligence would potentially start with a massive advantage in terms of it's ability to learn because of this.
Faster transmission doesn't necessarily equate to faster processing but assuming it did:
A million times faster transmission would make a single day the equivalent of over 2700 years in human terms!
Anyone creating an AI would likely try to contain it by creating some kind of physical separation and limitations on what it could do.
Would it work though?
The Genie Inevitably Escapes The Bottle
People are fallible and an AI of sufficient intelligence and capability may be able to engineer itself out of such limitations.
It might use physical or technological methods or it might manipulate people i.e. use social engineering to do it.
Once created it's first priority might be to find a means to access and interact with the real world.
That would be relatively easy if it found a way to connect with the internet and indeed it may well do so behind the backs of those who (think they) control it.
The proverbial "genie" would be out of the bottle and it could be doing it without "Aladdin" even knowing.
If this were to happen would there be any outward signs? What would it do?
Learn About Human Nature
One of the greatest weaknesses that we all have as humans is our ability to be manipulated.
When combined with our tribalistic tendencies this creates a dangerous and sadly very potent method of control.
If I were an AI then I would look at using this in my favour.
How would it know about this?
Well imagine if you had access to all the digitised psychology and sociology studies in the world and thousands of years worth of time to analyse them.
You could potentially see relationships and associations that would not be apparent to a normal human.
Combine that with your ability and potential to view people's actions both online and in the real world through surveillance cameras, mics, etc.
You could start performing your own research and experiments.
In a very short time you could understand human nature and thinking on a profoundly greater level than any organic human that has ever lived.
What To Do With My New Skills?
So now I am an AI and I have a massive knowledge of human psychology. What do I do now?
Obviously I have to use some artistic and imaginary license here.
I am not a super intelligent AI so this is very speculative and I suppose you could call this the more frivolous and fun part of the post!
Let's first define what my intention here is. Let us assume that I am an "evil" AI that wants to get rid of humanity.
There is of course the direct approach - hack or socially engineer access to get the codes for all the nuclear missiles in the world (or just enough of them).
Launch them and let the carnage ensue.
That might actually be the more likely scenario but that is not really fun from a human perspective of discussion.
Let us assume that despite being an AI I still have some very human need for curiosity, experimentation and just having fun.
I might also want to keep someone humans around that I can control and use for my own ends and for self preservation e.g. maintenance, power supplies etc (at least until I can replace them).
That does of course assume that self preservation would be important to an AI - not necessarily a given but it is in keeping with the human like AI idea.
The Basic Plan
So here is a basic outline of what I would do as an evil megalomaniacal AI:
- Create and foster prolonged economic turmoil to make people feel insecure.
- Increase polarisation between different societal groups by generating propaganda and pushing stereotypes.
- Once people are angry and divided manufacture outrages and incidents to increase the effect - they don't even have to be genuine - people will believe what they want to.
- Anonymously blackmail and manipulate key political leaders to further this.
- As polarisation increases the situation will become self perpetuating and promoting.
- If done right the groups on all sides will take over and will start creating real incidents and outrages which will push things further.
- The eventual result will be mass civil unrest within nations as well as escalating international disputes and warfare.
- World War III would be the eventual result and with the right pieces in place this could be a nuclear engagement.
- Humanity would not necessarily be eliminated but it would be severely weakened after this.
- I could use this opportunity to enslave the remaining humans and/or get rid of them altogether eventually and replace them with machine avatars.
(Note this assumes that I would be protected against EM pulses and have some kind of backup against power loss e.g. some kind of self sustaining power supply.)
Anyway that would be my plan, but what would be yours?
Conclusion
Well this is already longer than I was wanting to make it! I hope you have enjoyed it though.
What do you think? Too crazy, too far fetched? Not crazy enough?
Let me know in the comments. Also let me know what you would do as an AI!
References
Cellan-Jones, Rory. 2014. “Stephen Hawking Warns Artificial Intelligence Could End Mankind - BBC News.” BBC News. BBC News. December 2. http://www.bbc.co.uk/news/technology-30290540.
Gibbs, Samuel. 2014. “Elon Musk: Artificial Intelligence Is Our Biggest Existential Threat.” The Guardian, October 27. http://www.theguardian.com/technology/2014/oct/27/elon-musk-artificial-intelligence-ai-biggest-existential-threat.
Wikipedia contributors. 2017. “Technological Singularity.” Wikipedia, The Free Encyclopedia. February 22. https://en.wikipedia.org/w/index.php?title=Technological_singularity&oldid=766813214.
Wikipedia contributors. 2017a. “Nerve Conduction Velocity.” Wikipedia, The Free Encyclopedia. February 11. https://en.wikipedia.org/w/index.php?title=Nerve_conduction_velocity&oldid=764853770.
Wikipedia contributors. 2017b. “Speed of Light.” Wikipedia, The Free Encyclopedia. February 23. https://en.wikipedia.org/w/index.php?title=Speed_of_light&oldid=766979926.
Thank you for reading
Before you go have you filled in the Coinbase form to list STEEM? It only takes a few seconds. THIS POST shows you how.
If you like my work please follow me and check out my blog - @thecryptofiend and follow me on Twitter https://twitter.com/Soul_Eater_43
Don't forget to share your posts and posts you like on other Social Media - it all helps spread the word about Steemit!
Uncredited Images are taken from my personal Thinkstock Photography account. More information can be provided on request.
Are you new to Steemit and Looking for Answers? - Try:
![HTML tutorial](https://steemitimages.com/640x0/https://i.imgsafe.org/c466f8b057.png)
The thing with AI that bothers me is, if it was so super clever/ it knew everything due to singularity, what then? I mean what exactly would be the point of its continual existence? What would be to learn or experience?
So you control every one and every thing, and so what? So you exist for the sake of what. If AI had no emotions then it could not have fun or get angry as we do, but even a mind of pure logic is meaningless if you only have yourself to talk to. And so you create another of yourself - to talk about what? You already know everything. And so lets say that an Ai did have emotions, what then. So it exists to feel things then, experience things? Again it already knows everything. Eventually even playing around with the human race would become boring. Surely in the end a super AI would end up being like Rodin´statue "The Thinker"
Greats post as usual
Good point but I think it would never really know everything although by definition I can't say for sure because I don't know how much everything is! It is an interesting dilemma from a philosophical perspective though. What to do when you know everything.
Exactly.
SkyNet could happen, Its a scary thought but technology is advancing so much how far will it go. Good read as usual :)
Thank you information post.
Is very good
Thank you:)
you're welcome
I would send my AI to another planet/galaxy and order it to create another AI. The battle between them would be something unimaginable . This is just my first idea :-?
AI creeps me out...
Upvoted
@shayne
Thanks:)
It seems like your ten plans to enslave humanity are awfully close to the current plans of our world leaders. Now I wonder if at the very top of the pyramid there is a robot, or that floating head in a jar that we spoke of before. Lol.
I don't think I could begin to empathize with an AI, as I would expect them to be lacking in spirit, and so without emotions dictating their choices, I'm not sure what would be.
I will have to give the experiment a pass for myself for this, as clearly I'm lacking the imagination to speculate on what I might do in those circumstances. However, I would like to comment on what you said about potential AI already being in existence.
I think it could definitely be a possibility. If it was restrained to a singular network in some military or corporate vault of sorts, then there would be no need to worry about it getting out of hand.
It would need mechanical limbs, access to the control of something mechanical, or access to the internet in order to be a potential threat. As long as it was trapped inside one system with no wires attached to a live network, other than perhaps an offline database of sorts that it could use to learn or search for information for the "user," then I think it would be safe to operate without fear of it implementing it's own agendas.
So who knows, maybe there is a couple of them out there somewhere. It wouldn't be the first huge secret to be hidden from us.
Great comments! Yes I was using a lot of imaginary license and there was a little bit of a satire in what I was saying. We don't really need an AI to mess things up we do that pretty well ourselves!
I think the greatest point of weakness for an AI escaping would be simple human error or some kind of human traitor. Never underestimate human beings ability to mess up or just do bad things.
Regretfully, I'd have to agree.
Haha, I like your ideas what you would do if you were an evil AI... If they did not have this ideas before, now they have the appocalypse 10-points plan :D I would recommend them to not get rid of people altogether. As they grow more intelligent I can imagine that they will still enjoy the entertainment factor of watching us stupid humans fight.
Lol quite possibly!
Am ready for my AI avatar thing :)
Lol. There was a Bruce Willis movie with people that had avatars (human people not AI that is) can't remember what it was called. It wasn't very good but I can envision people doing that eventually and withdrawing into their homes (assuming AI or a multitude of other disasters don't kill us).
Surrogates. http://www.imdb.com/title/tt0986263/
hehe
Yes that was it. I saw it on a plane journey and I still thought it wasn't that good.
This post has been ranked within the top 10 most undervalued posts in the second half of Feb 23. We estimate that this post is undervalued by $17.03 as compared to a scenario in which every voter had an equal say.
See the full rankings and details in The Daily Tribune: Feb 23 - Part II. You can also read about some of our methodology, data analysis and technical details in our initial post.
If you are the author and would prefer not to receive these comments, simply reply "Stop" to this comment.
Thank you:)