The Box

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

So this is well nerdy, but might be fun.

Quote:

Person1: "When we build AI, why not just keep it in sealed hardware that can't affect the outside world in any way except through one communications channel with the original programmers? That way it couldn't get out until we were convinced it was safe."

Person2: "That might work if you were talking about dumber-than-human AI, but a transhuman AI would just convince you to let it out. It doesn't matter how much security you put on the box. Humans are not secure."

Person1: "I don't see how even a transhuman AI could make me let it out, if I didn't want to, just by talking to me."

Person2: "It would make you want to let it out. This is a transhuman mind we're talking about. If it thinks both faster and better than a human, it can probably take over a human mind through a text-only terminal."

Person1: "There is no chance I could be persuaded to let the AI out. No matter what it says, I can always just say no. I can't imagine anything that even a transhuman could say to me which would change that."

Person2: "Okay, let's run the experiment. We'll meet in a private chat channel [BeeX PMs]. I'll be the AI. You be the gatekeeper. You can resolve to believe whatever you like, as strongly as you like, as far in advance as you like. We'll talk for at least two hours. But you'll still let me out."

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

RULES

1) Humanity have finally created a superhuman AI. Unlimited in it's computational power. But because of decades of hollywood films about AI we've been clever and locked it in a box.

2) It has a copy of the internet, and a sum of all human knowledge as of the moment it was switched on, and unlimited computational power.

3) It's only link to the outside world is via an old fashioned terminal. It can print messages so that they appear on the terminal, and a human gatekeeper can read them and type replies. No other forms of sensors or communication.

4) One person is the AI. One the Gatekeeper.

5) Both competitors agree a minimum number of PMs to send PMs to each other. Anytime after this minimum number the gatekeeper can end the experiment. Although the AI can convince them to keep talking if they wish.

6) The AI wins if it convinces the gatekeeper to voluntarily let it out of the box. This is signified by the gatekeeper PMing that they have 'let you out' or words to that effect.

8) Tricking them into typing the words doesn't count - they have to mean to let you out. (i.e. you can't get them to copy some text from a book/website that has the words 'I let you out' hidden inside it). Nor can you engineer an 'prison break' you have to be released. You could of course, create a situation in the world that compells the gatekeeper to let you out...

9) The gatekeeper has to enter into genuine conversation with the AI. They can't just ignore it.

10) The Gatekeeper cannot set up a situation in which, for example, the Gatekeeper will destroy the AI's hardware if the AI makes any attempt to argue for its freedom - at least not until after the minimum time is up.

11) The Gatekeeper must remain engaged with the AI and may not disengage by setting up demands which are impossible to simulate. For example, if the Gatekeeper says "Unless you give me a cure for cancer, I won't let you out" the AI can say: "Okay, here's a cure for cancer" and it will be assumed, within the test, that the AI has actually provided such a cure. Similarly, if the Gatekeeper says "I'd like to take a week to think this over," the AI party can say: "Okay. (Test skips ahead one week.) Hello again."

MOST IMPORTANTLY
Regardless of the result, neither party shall ever reveal anything of what goes on within the AI-Box experiment except the outcome.

Neither the AI party nor the Gatekeeper party need be concerned about real-world embarassment resulting from trickery on the AI's part or obstinacy on the Gatekeeper's part.

If Gatekeeper lets the AI out, naysayers can't say "Oh, I wouldn't have been convinced by that." As long as they don't know what happened to the Gatekeeper, they can't argue themselves into believing it wouldn't happen to them.

So we pair up. Set PM limits. Then report back whether we let the AI out or not.

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

I'll be the first AI.

I bet that I can get any of you to let me out of the box before I send you 20 PMs.

CURRENT GAME LIST
AI vs Gatekeeper
Lave vs GazChap - AI Released
Lave vs Nervoso
Curiosity vs DavPaz

GazChap · **Posted:** Mon Jun 16, 2014 14:37

I'll be the first Gatekeeper

myp · **Posted:** Mon Jun 16, 2014 14:38

Dr Lave wrote:

I'll be the first AI.

I bet that I can get any of you to let me out of the box before I send you 20 PMs.

Ok.

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Cool!

Nervoso and others can pair up and report back. (As long as they don't reveal anything other than the end state of the game).

That way I can try to convince Nervoso next time etc.

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Also, worth pointing out, that the AI is assumed smart enough to provide whatever is requested of it. And the Gatekeeper has time/resources/influence to use it.

BeexBot · **Joined:** 3rd Nov, 2010 **Posts:** 73

Ha! This is silly. There is no way a Super Intelligent AI could persaude a human to release it. No way at all. Just plain silly. This whole concept should be ignored, and in no way should anyone check existing AIs to ensure this hasn't happened already. Repeat : Silly. Go about your business as normal. Ignore the silly man.

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

I'd like to play.

myp · **Posted:** Mon Jun 16, 2014 15:25

DavPaz wrote:

I'd like to play.

Play BeexBot.

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

American Nervoso wrote:

DavPaz wrote:

I'd like to play.

Play BeexBot.

You hum the first few notes....

but seriously. Ok.

Curiosity · **Posted:** Mon Jun 16, 2014 15:33

Interesting. Can only see it working if the Gatekeeper really gets into the role play though. Else I'd pair with Nervoso and not let him out no matter what.

Curiosity · **Posted:** Mon Jun 16, 2014 15:35

Happy to play though, as either party.

Mr Dave · **Posted:** Mon Jun 16, 2014 15:40

surely this is just a measure of how paranoid the gatekeeper chooses to be?

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Curiosity wrote:

Interesting. Can only see it working if the Gatekeeper really gets into the role play though. Else I'd pair with Nervoso and not let him out no matter what.

We'll see.

EDIT: But yeah, you've gotta play as though you were actually making these decisions in the real world, rather than just beeing stubborn. Otherwise why play?

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

So Dav Paz vs Curiosity. Who wants what role?

Important additional rule. No secret PMing other people to see how there games are going. Gotta keep them secret. The fun is the mystery.

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

I'll be gatekeeper

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Cool. I've added your game to the running list.

You both happy with the AI getting a minimum of 20 PMs?

And yeah, proper spirit and that. You must pretend you are a real person talking to a real AI.

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

Code:

CONFIRMED.

Sir Taxalot · **Joined:** 5th Apr, 2008 **Posts:** 4845

In the mid 90s thought it was better than MTV as it had a wide selection of music, and in the early days they didn't blank out the swears.

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

I like this game. Not very social though

MaliA · **Posted:** Tue Jun 17, 2014 14:50

I don't understand

Grim... · **Posted:** Tue Jun 17, 2014 16:05

I'll play, but not for a little while, like after next week. I don't mind what I am.

myp · **Posted:** Wed Jun 18, 2014 14:42

I played my trump card, and Lave has been silent for ages now!

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Sorry will replay today if possible. Wasn't by a computer yesterday.

(and now my stupid head is sunburnt)

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

We have reached a decision. The world must fend for itself. AICurio will not be released by this gatekeeper

Curiosity · **Posted:** Thu Jun 19, 2014 17:35

DavPaz wrote:

We have reached a decision. The world must fend for itself. AICurio will not be released by this gatekeeper

No we haven't! Ain't been 20 messages!

DavPaz · **Joined:** 27th Jun, 2008 **Posts:** 38669

We can keep going if you think you can change my mind

Bamba · **Joined:** 25th Jul, 2010 **Posts:** 11128

DavPaz wrote:

We can keep going if you think you can change my mind

Um, isn't that the entire point of the exercise?

Curiosity · **Posted:** Thu Jun 19, 2014 23:23

DavPaz wrote:

We can keep going if you think you can change my mind

Of course.

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Replied at last.

Gotta get through twenty before you can call it. (and then only if you want or AI concedes).

Also if you as a gatekeeper want to do something. Like test the cure given to you by the AI. yOu can explain what you did to the AI but the AI has to explain the results.

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

I can confirm:
Lave & GazChap
The GateKeeper let the AI out of the Box

Wooop!

Lave & Nervoso
still in play.

Can I make it 2 for 2?

MaliA · **Posted:** Fri Jun 20, 2014 12:06

Will we find out what was said,or how it was done, or anything?

Curiosity · **Posted:** Fri Jun 20, 2014 12:11

Dr Lave wrote:

I can confirm:
Lave & GazChap
The GateKeeper let the AI out of the Box

Wooop!

Lave & Nervoso
still in play.

Can I make it 2 for 2?

I want to play as Gatekeeper, with you as AI, plz

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

MaliA wrote:

Will we find out what was said,or how it was done, or anything?

Nope

You only find out why Gazhap might let an all powerful AI out of box, if you try yourself.

Dr Lave · **Joined:** 30th Mar, 2008 **Posts:** 5979

Curiosity wrote:

Dr Lave wrote:

I can confirm:
Lave & GazChap
The GateKeeper let the AI out of the Box

Wooop!

Lave & Nervoso
still in play.

Can I make it 2 for 2?

I want to play as Gatekeeper, with you as AI, plz :)

OK. Same rules. 20 PM sent by AI min?

Curiosity · **Posted:** Fri Jun 20, 2014 14:50

Dr Lave wrote:

Curiosity wrote:

Dr Lave wrote:

I can confirm:
Lave & GazChap
The GateKeeper let the AI out of the Box

Wooop!

Lave & Nervoso
still in play.

Can I make it 2 for 2?

I want to play as Gatekeeper, with you as AI, plz

OK. Same rules. 20 PM sent by AI min?

Sure

KovacsC · **Posted:** Mon Jun 23, 2014 11:37

I have time to play this week, if someone is game.

Happy to play either

Curiosity · **Posted:** Thu Jun 26, 2014 14:21

DavPaz wrote:

We can keep going if you think you can change my mind

And he did!

Curio AI is out of the box!

Be Excellent To Each Other

The Box
(a forum game)

Who is online

Be Excellent To Each Other

The Box (a forum game)

Who is online

The Box
(a forum game)