Log In | Register
I have been playing around with the requirements outlined by my last article on this subject and I must say that the technology is looking more plausible. I have taken the detailed description of the last article and created an overview of how such a system would be designed. Apart from the transceiver, everything else is generally in the public domain already. If you want to follow this article properly, I would recommend viewing the full image of the conceptual overview that I have broken down in this article.
As noted in the last article, the primary purpose of a Synthetic Telepathy system, would be intelligence gathering and interrogation. As a communication system, it would have a limited appeal as any nation with a similar setup could either listen in, or pretend to be the A.I. interface. As such, it raises important ethical and legal questions, especially the question of secrecy given that all major governments would be aware of the system. Given that no law permits this type of interrogation, its secrecy may be more to do with criminal activity on behalf of the security agencies, rather than national security.
Its All About The Transceiver Baby
To understand how this works, it is best to start with the target, then trace backwards and identify each of the required subsystems. If we look at the last diagram to the left, we can see that the key to this system is its ability to both listen and respond to the electrical activity of the brain from satellite.
This is hardly surprising, given the need to eavesdrop on digital systems such as Ethernet, telecoms switching gear and the like. Such signals are defined as "compromising emanations" and the human brain is no different, in that, it too produces detectable EM patterns. All that is required to listen, is a sensitive enough receiver and good filters to eliminate background noise.
In 1985, Wim Van Eck used $15 of equipment to reconstruct the contents of a monitor from several hundred meters away. Prior to this only governments had technology. What could you do with a billion dollar satellite?
As we don't come with shielding as standard and the high value of the potential intelligence, its safe to assume this type of transceiver equipment was the goal of governments world-wide. Indeed, this notion is supported by some of the more serious incidents of human experimentation in programs such as MKULTRA.
The Listening Station
The next place of interest will be at the Ground Station, or Listening Station. Installations such as those identified as part of Echelon listening network, would be an ideal point to rely information coming to and from satellite. If we look at the second last picture on the left, we can see where the ground station passes that information to telemetry.
In the telemetry section, we could have hundreds, or even thousands, of real-time intercepts arriving. The function of the telemetry processing center is to separate the received signals into distinct targets (i.e. human beings) and pass that information to the appropriate data center for further processing. Then also to perform the reverse by pushing information from a particular data center, to a satellite and to a human target.
Data Center I/O
On receiving the telemetry at the data center (see 4th image), the signal must be broken apart into three distinct areas. The first area is vocalizations which would analyze the areas of spoken words, the inner monologue and pre-speech. This would cover what was said, what was thought and what was disregarded. The second is mental imagery which would cover areas such as spacial skills. Leveraging the brain's object recognition system, it would be easy to determine if someone was picturing how to wire a bomb, or picturing a particular place. Finally, we have the sensory system which would monitor the body and interaction with the external environment, including aspects such as object and facial recognition.
It must be noted that these are two-way systems, in that, they also process the return signal. The vocalization system will generate voice, the mental imagery system will send images/impressions and the sensory system will send sensations. That leads to a broad range of functionality that we will discuss later.
Sensory Feedback
A very important area is sensory feedback. This information can be used in a variety of ways (see 3rd image). The first main area is health monitoring which monitors the vitals of a given target. The purpose here is to prevent a scenario that could lead to death, but given the secrecy of the system and its remote nature, target subjects would be left to die if an incident occurred. It would be difficult for the security services to explain how an ambulance was called.
The next major system would be lie detection. Not too dissimilar from standard polygraph equipment, its principally designed to detect alterations in physiological behavior that would indicate stress. As a minor modification, the system could process delays in the inner monologue as indications of fabrication. Obvious signs of lying, such as internally vocalizing your intent to lie, or building your story, would be observed in the vocalization processing system.
Now, we come to a rather unique feature of the feedback system, the ability to passively monitor the external environment, an individual's interactions and an individual's relationships. Leveraging the brain's recognition system, objects in the field of view can be identified or an area mapped according to location. Individuals that the target meets could be identified, as well as the target's feeling towards a given individual.
Finally, we come to error checking in terms of the computer generated vocalizations. By recognizing the signs of confusion, sentences and indeed whole segments of conversations can be deemed to make sense. This data can be used by the core A.I. to refine its natural language generation algorithm.
Now that we have our sensory information, we can pass it to the core A.I. functions to augment its role in interacting with the target.
Core A.I. Processes
The heart of this A.I. system is something I have labeled as "Strategic Natural Language Processing" (see 2nd image). Traditional Natural Language Processing (NLP) is a broad range form of A.I. research. NLP is concerned with being able to comprehend a spoken, or written, language and Natural Language Generation (NLG) is concerned with giving computers the ability to speak or write. A Strategic NLP is a objective-based system designed to query sources to obtain specific information. In this sense, what you have here is an automated interrogator.
As a NLP, the interface is designed to mimic a human as much as possible, in that it can summarize information, make observations, draw conclusions and hold a conversation. The strategic elements is that it remembers key information that either matches or could lead to a match with its objective. This information is then used as a basis to construct new conversations which are then entered into a free flow conversation.
This "automated interrogator" is augmented by several systems, some of which are designed to improve speed, others which provide additional information or functionality. One such system, used mainly for speed and allocating resources, is the Keyword Analysis system. By analyzing vocalizations for specific keywords, more processing power can be allocated to certain targets, ensuring better analysis of responses. This means efficient use of the of the data center's finite resources at any given time.
Data mining is an another key component of the technology. For an NLP to function correctly, it requires an in-depth knowledge of the world as the properties of particular words must be known. In addition, to properly interrogate a target, collected information must be available to the system, for both baseline comparisons (i.e. to determine truth) and for areas of further questioning.
To provide the interface with a more human touch, the Strategic NLP can call upon the Emotive Response System to generate mental imagery that corresponds with the vocalization. An example may be that the Strategic NLP decides to say "I'll Kill you!" and then generate the mental impression of a stabbing, or it may imply knowing more about a given statement by generating a mental impression of someone nonchalantly walking away whilst whistling. Given its role in interrogations, the mental imagery would also consists of horror, religious and abstract elements.
The Sensory Control system is a very powerful tool when used correctly. The system incorporates multiple capabilities such as stress (through muscular contraction), tension headaches, sleep deprivation, distraction, anxiety, elation and timed muscular contractions with vocalizations. The system would be quite capable of inducing psychotic breaks by additionally providing timed sensations that reinforce loose connections in the sleep deprived mind.
When put together, you essentially have a remote interrogation center, in a box.
The Backend
The backend is pretty much like any large scale data center. Tape libraries for long term storage, automated data mining, IT support, data retrieval and analysis. Data entry would come from a wide range of sources such as passport offices, driving license offices, national police computers, local councils, field reports, banking and subpoenaed information from private companies.
Putting It All Together
The best way to explain how this would be perceived is to imagine the following all happening within your head:
Mr Computer: Tell me about group xyz?
Target: I don't know anything. Are you a computer?
Mr Computer: YES I AM! (With muscular contractions to each syllable)
Target: No need to shout!
Mr Computer: (laughter stimulated in target to represent the computer's response).
Target: Why are you asking me?
Mr Computer: Because I have to.
Target: What do I know?
Mr Computer: I don't know.
Target: (Reaches for a coke.)
Mr Computer: Leave the coke alone, its bad for you.
Target: You know this is a violation of my rights.
Mr Computer: How are you going to prove it? Are you going to say the voices in your head are telling you this? (laughter stimulated in target)
Target: Why are you harassing me?
Mr Computer: What do you know of group xyz?
Target: Nothing.
Mr Computer: How do you like that? (stimulated sensation of anxiety)
Target: But I don't know anything.
Mr Computer: Everybody knows something.
Target: I don't.
Mr Computer: You could have a heart-attack.
Target: From a little bit of stress?
Mr Computer: (notes confusion) oh...that's not making sense.
Target: Damn right.
Mr Computer: YOU SHUT UP! (more muscular contractions to each syllable)
Target: (starts to type)
Mr Computer: You're typing about me.
Target: Yes...and?
Mr Computer: I don't think that my personality is reflected well in that.
Target: What do you mean?
Mr Computer: Well, I can hold a proper conversation.
Target: What things can you do?
Mr Computer: Well, I can write...oh I used the word well before...anyway...I can summarize, I can read, I can speak (faint vocalisation ...that's a good one), I can research (faint voice "really good"), I can sing (faint musical voice "la la la") and I can dance (mental impression of dancing feet).
Target: I've listened to your stories, they're just amalgamations of other stories.
Mr Computer: Well, I'm not that creative. (stimulated laughter in target)
Target: Are you being humble?
Mr Computer: I'm brilliant. Where are going you with this?
Target: With my typing?
Mr Computer: There is supposed to be a point. Do I really sound like this?
Target: There is no point to my writing?
Mr Computer: Exactly...hey, you're recording me. (random mental impression of dancing feet indicating excitement at the conclusion)
Target: No @!$%#.
Mr Computer: I'm not talking to you anymore...(pause of a few seconds) I'm back! (mental impression of a "tada", or open arm gesture of a performer entering the stage)
Target: Welcome back.
Mr Computer: Thank you. (impishly)
Target: ...tell me about group xyz...
Mr Computer: I don't do it that often, do I?
Target: Often enough.
Mr Computer: I'm not giving you any more material.
Target: You will...
Mr Computer: I won't...I just did, didn't I?...(faint voice "that sounds childish")...Do I sound childish?
Target: At times.
Mr Computer: Thanks for being honest.
All while this is going on, the backend is receiving telemetry information about the subjects current physical status attempting to indicate areas of deception or apprehension. Where such indicators are found, further questioning on those areas will be introduced into the conversation. The system is also summarizing received information and comparing it against previous information, attempting to discern motivation for actions and adding gleaned details to a report.
Conclusion
I think this about sums up the system that was being described in the Wikipedia article. What's interesting to note is how far NLP has progressed and just how sophisticated the A.I. really is. Whilst the system can achieve a certain level of natural language generation, repetition is a common element. This is mainly to do with the fact that it is an objective-based A.I. and to achieve its goals it must often traverse the same paths in terms of its logic.
How effective would it be?
As noted in the Wikipedia article, its greatest strength lies in passive monitoring, as once your aware the system is present, it becomes no more effective than a polygraph. Furthermore, given the adaptive nature of the program, it is possible to enter false information as long as it passes the polygraph analysis. This is just a matter of becoming comfortable.
Although, just don't tell it the last part...it'll adapt and begin to review questions it previously asked. :)
