A robot goes into a cake shop...

[sknox]

>what form does that thought take?
I'm just guessing here, but
110010101110000101000101011111000101010101010101111010100010100000010100010101010100001010101
Which is odd, when you think about it.

 

[Dave]

Isn't the problem here, not doing the actually coding, but the measurement and comparison of two subjective thoughts, based upon emotion and personal preference rather than rational logic. The robot can easily be designed to make choices, however, that choice is not an objective one, nor is it a yes or a no choice.

The choice here is the robot's need and desire for an un-affordable cake versus the detrimental effects of depriving someone else of the said cake. You might also add in the detrimental effects of being caught, along with the probabilities of being caught and being successfully prosecuted. When a threshold is reached where the need and desire are greater than the perceived detrimental effects, then the theft can occur. The level at which that threshold is set is purely subjective and will be different for every robot.

The other factor is just a time delay, however, it is possible that as the variables change over time, the threshold could rise and fall. If the need and desire is very close to the detrimental effects, then given enough time elapsed, the threshold could fall to a level at which the cake would be stolen. If the threshold then rises again later, the question is, would the robot then feel regret?

 

[Kylara]

My partner and I chatted about this last night and this is what he said: my thoughts on the matter would go something like; if its an ai with a neural net then depending on the outcome, positive (got away with it) or negative (got caught) then it would be visualised at a change in the weightings of the nodes connections. As its a decision with a measurable outcome, it will either reinforce that path, making the decision likelier in the future or degrade the path, making it less likely.
it depends also on how far back you go. Has it been told 'want cake. Spend no money' or just 'want cake' or 'steal cake' - third party involvement or is it autonomous in which case it will have learned (or been taught) that stealing cake is most efficient way of getting the cake.
And he says yes you could see how it 'thought' its way to the decision. It would be messy, but you would be able to see everything. And he also said that the more it did the better it would get at stealing the cake as more data is acquired.

 

[Penny]

I would say that, If your robot was outputting a plain text log and not just running on compiled code. the log might look something like.

Running Cake Search Routine...
Running Cake Search Routine...
Running Cake Search Routine.. Cake Found!

Run Cake Aquisition v1.0...
Navigating to cake: Running Navigational Waypointer... Initiating Rayscan...
Rayscan took 3.4seconds
Path found..
Initiating move command

WARNING IMPEDIMENT DETECTED: Class 1 Human Interaction
Initiating Asimov Saftey System
Cake Aquisition: Run Deterrance Protocol
Voice Response: "I am just browsing"

Cake Aquisition: Run Deception Protocol
Voice Response: "Sir I am detecting a fire in the building you should leave while I secure the premises and call the fire brigade"

Cake Aquisition: Run Aquire cake
Cake Aquisition: Cake Aquired! Mission Successful!
Cake Aquisition: Return to base Initiating Return Home Protocol

Robot returns to base.

thats how I might do a log

If we assume the robot is running some kind of ROBO OS, a dumb AI that essentially chooses what is the best instruction or command to run based on its current set of user inputted instructions.
You can simply output what commands it might be running in a similar way to how DOS display's user commands.
so it runs various commands and they output various results to the console, all the nitty gritty coding is hidden just the important parts are written down.
Its understandable by your standard reader because its the same kind of text that scrolls up alot of computers on boot.

 

[Venusian Broon]

Of course, I doubt he'd get very far. If he's anything like a lot of fictional robots, his pincers will squash the cake and he'll announce "Commencing cake theft!" before starting.

Only a fifties or 'fallout' robot of course. Surely a 2020's robot would have the sense to order the cake over the internet, get it home delivered (hell, even get the human lackey to pack it away for it) and then steal it using credit card fraud.

Then it could mash it up with it's pincers in an attempt to open the packaging.

 

[Ursa major]

The robot is very particular about the cake it wants, which has to conform to the OSI model (so must have seven layers).


What the robot has forgotten is that it really needs at least one more cake to make it all worthwhile.

 

[paranoid marvin]

Can a robot forget? Perhaps it's chips are fried

The robot is very particular about the cake it wants, which has to conform to the OSI model (so must have seven layers).


What the robot has forgotten is that it really needs at least one more cake to make it all worthwhile.

Ah, so it will receive it's just desserts?

 

[mosaix]

has to conform to the OSI model (so must have seven layers).

Very good, U.M.

 

[Kylara]

Are you sure the cake is not a lie?

 

[farntfar]

The cake is said to be in a cardboard box.
No-one knows whether it exists or not until the box is opened.

You can take the money, or you can open the box.