Book Quotes: Human or AI? Using Law to tell them Apart
Search in this book only
Source: Otherland: City of Golden Shadow, Page: 36
Our Thoughts on this Quote
How do you tell a really advanced artificial general intelligence, controlled by a neural network, with learning, pattern recognition and adaptive behaviours coded into it, along with embodiment with emotional expressions and body language – apart from a human embodied through an avatar in the same way?
If the AGI is advanced enough – or even if its not, but is using simple psychological cues, as the old Eliza bot did – then the answer may very well be, “you cannot”. A detailed investigation may reveal the truth, as the Turing test sets out to do, but even that is not a guarantee.
In 1950, Alan Turing posted the question “Can a machine think?” He devised a test to answer this question, where one user sits at a computer and interacts with two other users, one of whom is an AI, the other is human. The goal? To tell the human from the AI.
The Turing Test has held up so far. For a number of years the Loebner prize, established in 1990, has strived to put Turing's test to the well, to the test. Every year the most advanced artificial embodied minds compete to prove to a panel of judges that they are indistinguishable from the human users.
Every year so far, they have failed to do so, failed to hold intelligent conversations crossing informally into any number of topics under the judges' discretion, lasting as long as the judges deem necessary, to tell the human from the machine.
Yet, the very existence of the prize fund, together with the steady improvements in such AIs, hints at the strong likelihood that one day the Turing test will be beaten, and it won't be possible by and large, to tell an AI from a human, when interaction is solely through computer mediation in whatever form.
When this day comes, we will need another method of telling human from bot, at least for as long as the AIs cannot be considered independent, sentient organisms in their own legal right. Otherland, as quoted here, acknowledges this, and uses the somewhat unique and functional stance of working the law in favour of this determination.
In the book, a 'puppet' is an AI. Not a sentient AI, but an artificial intelligence designed to mimic the functions of a human in a particular job. Able to hold a conversation, prattle about subjects, hold full knowledge of their chosen area and essentially be the perfect (unpaid) employee in an online venue.
So, as they have no mind to call their own, these puppets dangle from the strings of whichever corporation wrote their code and hosts their program. In order to tell them apart, the law of Otherland's society demands that this question be encoded into all of them. Asked if they are a puppet or a user, a bot is legally required to answer “puppet”, along with version information. If they don't, their company is looking at serious legal repercussions.
In the same vein, it is the law that a human user, if asked the question, must answer human. Only one answer is needed, obviously, to fully answer the question, no matter how many times it is asked.
If they do not, it is again a criminal offence with very tangible punishments. It is of course also quite rude to ask a human if they are a puppet, for to do so implies that their behaviour or mannerisms seem forced, artificial somehow. So, such a question would usually only be asked, when it is imperative the person asking, knows if they are talking to another intelligent, receptive mind, or simply to a glorified office machine.
The legality of sentient AIs, whilst it is discussed in Otherland, is not tackled in relation to this law, so it is unknown at this time as to how the legal route would approach the situation when the AI has the same cognitive faculties as the human user – and presumably the same legal status. Quite possibly, it would simply involve tacking a third acceptable answer to the question.
No resources have yet been matched to this quote.
About the Book 'Otherland, Volume 1: City of Golden Shadow'
By Tad Williams
Produced By DAW Books
The first book in the Otherland Saga, City of Golden Shadow serves as an introduction and ground setting book that enriches and deepens the later works. Among the many aspects of life in this near-future world it depicts in detail, is an advanced virtual reality technology.
City of Golden Shadow introduces the basics of total immersion VR technologies, slightly further ahead than those of William Gibson?s Neuromancer. It delves into all the possible uses for such technology in stunning depth and detail, yet it does so in such a way that the pace of the plot never slackens.
Click here for full review of Otherland, Volume 1: City of Golden Shadow
Other Books by this Author