Ethical AI
APPLICATIONS PROPOSAL FOR AI - PATENT # 6,587,846
The dream of artificial intelligence has been a goal in the field of Computer Sciences since virtually the dawning of the discipline. This anticipated style of artificially intelligent computer would potentially assist in all aspects of human endeavor, with the intriguing prospects for ultimately transcending the fixed limitations of the human condition. The current dramatic growth in computing power finally enables economic feasibility to such a meaningful AI development. A number of key approaches, such as brain modeling through neural networks, has been attempted, although scarcely enough detailed information exists about the brain to warrant any such serious inroads. In actuality, the key solution to developing convincing artificial intelligence invokes an innate understanding of human language in general. Indeed, the preeminent test for AI devised by Alan Turing abstains from relying upon any direct measure of consciousness or perception for its determination, rather targeting only the communicative factors underlying human language. Consequently, assuming the symbolic attributes of human language could be convincingly simulated on the computer, then many decades of needless effort could potentially be cut from either the neural-net or consciousness/perceptual approaches. Along these lines, the U.S. patent (#6,587,846) - now expired- was granted for precisely such a technical approach based upon the symbolic attributes underlying affective language. Clear precedents already exist within the field with respect to chess-playing computers that prove particularly adequate for modeling the symbolisms underlying such an abstract gaming format (although scarcely capable of anything else). In a similar fashion, the symbolic attributes of the language tradition prove a similarly comprehensive goal, although several orders of magnitude more abstract and complex in this regard. Certainly the primary economic focus of society as a whole is mediated chiefly through the symbolisms of human communication, specifying language as the most rational focal point for ongoing research. This is not necessarily meant to imply that a sensory/motor enabled robot designed to make sense of its immediate environment is not a rational focus for directed research. Indeed, such an achievement could eventually be merged with the currently proposed language simulation model to permit a more conceptually complete computer avatar. As far as the most economically direct human applications are concerned, however, it proves entirely more cost effective to initially target the symbolic attributes of human language (in all of its various manifestations).
Fortunately, a convenient shortcut to the daunting complexity of this direct language simulation has recently been proposed (the technical basis for the aforementioned patent). This new approach directly focuses upon the motivational (or emotional/affective) aspects of language as its guiding principle, with the remaining bulk of value-neutral language filling-in in an accessory role. Indeed, as Robert Warren Penn once insightfully wrote: "What is man but his passions?" Along similar lines, most neuroscientists consider the mind/brain complex as a vast motivational analyzer that enables the individual to flourish in harmony with the environment through the principles of instrumental conditioning. The current patent establishes precisely such a foundation within conditioning theory; in this case, appetite in anticipation of rewards, or aversion in expectation of lenient treatment. Furthermore, when the more abstract forms of affective language are further viewed in the terms of an ascending interactive hierarchy of meta-perspectives, then the overall complement of the traditional groupings of virtues and values jumps neatly into focus.
Solicitous. . Rewards ..... Submissive . Leniency
Nostalgia . Worship ......... Guilt . Blame
Glory . Prudence ............. Honor . Justice
Providence . Faith ........... Liberty . Hope
Grace . Beauty ............. Free-will . Truth
Tranquility . Ecstasy ............ Equality . Bliss
Appetite . + Reinforce .... Aversion . Neg. Reinforcement
Desire . Approval ........... Worry . Concern
Dignity . Temperance ........... Integrity . Fortitude
Civility . Charity .................... Austerity . Decency
Magnanimity . Goodness ............. Equanimity . Wisdom
Love . Joy ................................ Peace . Harmony
In conclusion, by focusing primarily upon the affective aspects of human language, an economically feasible shortcut to the AI simulation of human communication finally appears within reach.
www.angelfire.com/rnb/fairhaven/patent.html
Much detailed programming remains to be done, perhaps necessitating a customized coding language (and supportive hardware) consistent with a project of this magnitude. With a starting staff roughly the size of a large encyclopedia work force, a first generation simulation could potentially be achieved within a fairly modest time frame. Subsequent design generations would further endeavor to achieve even greater clarity and versatility. This painstaking process might eventually be more dramatically accelerated if ultimately accorded the status of a national initiative, particularly in light of its outright commercial value (as well as military applications).
The systematic organization underlying the ethical hierarchy allows for extreme efficiency in programming, eliminating much of the associated redundancy, providing a precise determination of motivational parameters at issue during a given verbal interchange. This AI platform is organized as a tandem-nested expert system, composed of a primary affective-language analyzer overseen by a master control-unit (that coordinates the verbal interactions over real time). Through an elaborate matching procedure, the precise motivational parameters are accurately determined (defined as the passive-monitoring mode). This basic determination, in turn, serves as the basis for a response repertoire tailored to the computer (the true AI simulation mode). This innovation is completely novel in its ability to simulate emotionally charged language: an achievement that has previously eluded AI researchers due to the lack of an adequate model of motivation in general. As such, it represents a pure language simulation, effectively bypassing many of the limitations plaguing current robotic research. Affiliated applications extend to the roles of switchboard/receptionist and personal assistant/companion (in a time-share mode).
Although only a cursory outline of applications is possible at this juncture, a more detailed treatment is posted at:
www.angelfire.com/rnb/fairhaven/patent.html
BELOW IS APPENDED AN INTERVIEW THAT ALSO DESCRIBES PRACTICAL APPLICATIONS
2) The title of the patent implies that the program simulates artificial intelligence. Do you consider the program to actually be AI or a simulation of it?
Granted a simulation of something "artificial" is a contradiction in terms. In this sense, "simulation" refers to something akin to a flight simulator, except that affective language is now targeted as specified within such a true AI role (as documented through the formalities of Turing's Test).
3a) Can you give me a description of how and what the process looks like?
This process enables a computer to reason and speak in an ethical fashion, specializing in roles specifying sound human judgement. Accordingly, the most basic unit of input for the AI computer must necessarily be the sentence, for the schematic definitions for the virtues and values are similarly given in the form of a dual sentence structure. The AI computer employs parallel processing to determine the precise degree of correlation between the inputted (target) sentence and its corresponding definition template. This matching procedure directly scrutinizes each of the grammatical elements within a given sentence, attempting a statistical correlation with the specifics of a given virtue or value. For instance, the tense of the verb, the plurality or person of the noun/pronoun etc., are all scrutinized according to a pre-set formula. Each processor then determines the sum-total of correct matches, ultimately yielding the relative probability of a match with a particular schematic definition. The processor yielding the highest overall rating is uniquely singled-out as the best match/solution by the master control unit.
Nostalgia . . H-Worship ......... Guilt . Blame
Glory . . . . Prudence .......... Honor . Justice
Providence . . Faith .......... Liberty . Hope
Grace . . . . Beauty ........ Free-will . Truth
Tranquility . Ecstasy ........ Equality . Bliss
Desire . . Approval ............ Worry . Concern
Dignity . Temperance ...... Integrity . Fortitude
Civility . Charity ............... Austerity . Decency
Magnanim . Goodness .... Equanimity . Wisdom
Love . . Joy ....................... Peace . Harmony
The master control unit achieves such a result through the aid of a feedback loop, the priority of the individual microprocessors reciprocally weighted on the basis of previous determinations. Accordingly, each power pyramid definition is respectively composed of both past, as well as current, component segments: establishing context as yet a further consideration in the matching procedure. In fact, the suitably advanced AI program retains in a long-term storage the content of every relevant conversation within a given style of interaction. On this contextual basis, the master control unit selectively weights the individual processors according to a preset formula, taking full advantage of both past (as well as present) conversational patterns. In particular, the computer would be exquisitely sensitive to variations in human personality, just as humans are instinctively so, satisfying yet a further condition of Turing's Test.
3b) Can you explain how you arrived at the "power-pyramid definitions?"
Although the cohesive ethical hierarchy of virtues and vices proves convincing in an intuitive sense, the more stringent AI applications call for an even higher degree of precision, calling for the construction of what are termed the schematic definitions of the power pyramid hierarchy. This crucial innovation spells out in longhand the precise location of each virtue or vice within the linguistic hierarchy, while further preserving the correct orientation of authority and follower roles. Each such definition is formally constructed along the lines of a two-stage sequential format; namely (A) the formal recognition of the preliminary power maneuver, and (B) the countermaneuver currently being employed, and hence, labeled.
Take, for example, the sample power pyramid definition for justice, reproduced below from the comprehensive (320-part) collection of definitions
Previously, I (as your group
authority) have honorably acted
in a guilty fashion towards you:
countering your (as PF) blameful
treatment of me.
But now, you, (as group rep-
resentative) will justly-blame me:
overruling my (as GA) honorable
sense of guilt.
According to this specific "justice" example, the honorable sense of guilt expressed by the group authority represents the preliminary style of power maneuver, countered by the just-blaming tactic employed by the group representative. Note how the respective placement of authority and follower roles is effectively preserved. According to this formal two-stage format, the preliminary power perspective represents the one-down power maneuver, whereas the current power maneuver designates the one-up variety. Power leverage, accordingly, is achieved by rising to the "one-up" power status; namely, ascending to the next higher metaperspectival level. Indeed, this interrelated hierarchy of schematic definitions can further be viewed as a type of motivational calculus, specifying the formal rules of transformation governing how each level meshes with those above or below it. The instinctual terminology of operant conditioning initially dominates the preliminary levels, replaced in due fashion by the virtues, values, and ideals of the higher levels. At each succeeding level, a new term is introduced into format, representing the power maneuver currently under consideration. Indeed, beginning with the group level, the preliminary terms begin to drop out of the equation, necessarily freeing-up space for the current terms under consideration, wherein maintaining a stable buffer of terms.
3c) Is it a large computer and how would a person physically communicate with it?
(I keep imagining something similar to HAL from 2001: Space Odyssey.)
The hardware memory requirements necessary for encompassing the respective knowledge bases alone should approach that of a supercomputer, although built-in time-share capability should permit economic feasibility. Indeed, the number of parallel processors ideally equals the sum-total of ethical terms within the power hierarchy, for a grand total of 320: quite a modest number even by today's design standards. This integrated processor array would further be structured in a hierarchial fashion, effectively mirroring the systematic organization of the power hierarchy. This stratified architecture takes full advantage of the strict transformational logic governing the power hierarchy, eliminating much of the redundancy bound to occur in any convincing language simulation. Indeed, the greatest degree of complexity necessarily involves programming at the most basic personal level of the power hierarchy, the remaining group, spiritual, and humanitarian levels following naturally from this elementary foundation.
4a) I am curious about the incorporation of ethics/motivations into an expert system. How did you arrive at the "rules" of ethics?
The Ten Ethical Laws of Robotics represent expanded versions of the motivational dynamics implicit within the entire ten-level span of the power hierarchy. Each of these Ten Ethical Laws is written in a positive style of formal mandate, focusing on the virtues to the necessary expense of the corresponding vices: representing a general overview of the enduring conflict pitting virtue against vice in a human (and AI) sphere of influence.
4b) Explain your personal definition of "metaperspectives" as it relates to this system.
In his Pragmatics of Human Communication (1967) Watzlawick (and associates) examine the informational aspects of communication, specified as his "hierarchy of metaperspectives." Accordingly, communication between individuals is typically compounded by abstract "meta" messages specifying how the relationship is to be conducted. The metaperspective, from the Greek meta (above) represents a higher-order perspective on a viewpoint held by another, schematically defined as: "this is how I see you-seeing me." In addition to this most basic initial metaperspective, even more abstract perspectives are theoretically possible, resulting in what Communication Theorists term the meta-metaperspective. This more advanced concept is one "meta" level further removed from the more basic metaperspective format, schematically defined as: "this is how I see you-seeing me-seeing you." In fact, there does not appear to be any theoretical barrier limiting the degree to which reflection can serve as a basis for itself, culminating in an unprecedented 10th-order level of "meta" abstraction: as reflected in the ten-level hierarchy of virtues, values, and vices.
One might rightfully question the capacity of the human mind to entertain such a supreme multiplicity of metaperspectives, particularly at the most abstract "meta" order levels. The mind apparently is able to selectively focus-in on only the immediately relevant levels of the ascending power hierarchy, similar to the analogy of walking up a ten-level flight of stairs. The process of rising to a next higher level implies the primacy of the immediately adjoining steps, amounting to a span of three total levels (equivalent to the meta-metaperspective), quite a modest task for the versatile human mind.
5a) Although the system is programmed to be virtuous, you also intend to incorporate subsets of the vices. Why?
According to the formal schematic format, the parallel hierarchy of the vices can further be incorporated into its own unique complement of power pyramid definitions. In particular, this full forty-fold complement of definitions for the vices directly contrasts point-for-point with the respective listing of virtuous terms.
Laziness - Treachery Negligence - Vindictive.
Infamy - Insurgency Dishonor - Vengeance
Prodigality - Betrayal Slavery - Despair
Wrath - Ugliness Tyranny - Hypocrisy
Anger - Abomination Prejudice - Perdition
Apathy - Spite Indifference - Malice
Foolishness - Gluttony Caprice - Cowardice
Vulgarity - Avarice Cruelty - Antagonism
Oppression - Evil Persecution - Cunning
Hatred - Iniquity Belligerence - Turpitude
In fact, this darker complement of power pyramid definitions proves equally informative to the AI computer, as do the virtues. The programming of the virtues alone would scarcely be sufficient, for the modeling of the vices prove crucial for specifying what is to be avoided. This familiarity with vices actually amounts to a basic safeguard within the system, allowing negative transactions to be transformed into positive ones, while simultaneously preventing the reverse from occurring. This programming of the vices would strictly be allowed only in a diagnostic mode, permitting the computer to diagnose the troubling behaviors of its human counterparts. The computer would further be prevented from responding in kind, instead attempting remedial measures to rectify such a dysfunctional circumstance.
5b) Will having access to the virtues/values help the program identify how to respond to a given individual, and how would the program then be able to respond in a manner that would soothe the individual?
With such ethical safeguards firmly in place, the AI computer should consistently be able to make the right moral decisions, a virtual "saint" among men. Any direct familiarity with the vices necessarily occurs in a purely diagnostic mode, the prime AI directive dictating adherence to a strictly virtuous response repertoire. This unerring sense of ethical constancy would certainly prove to be an invaluable asset for such an AI machine, in direct contrast to the more questionable predilections sometimes seen for its human counterparts.
6a) Please give me a better idea of what the program can actually do and what will the user's experience be?
The parallel processing capabilities of the AI mode can theoretically process of a large number of conversations simultaneously, maximizing the available circuitry by staggering the timing to make use of lulls naturally occurring within a given conversation. These might further be managed through a time-share arrangement, where multiple accounts are accommodated, as rated in terms of urgency. Conversations requiring real-time parameters would be assigned the highest priority, whereas more leisurely response rates would be processed during free periods. This would further entail a centralized CPU complex, connecting end users through a user interface network or the Internet.
6b) How can the program be used as a security system and in data-mining?
Although the initial prototypes would be formally limited to the English language, it should ultimately prove feasible to translate the specifics in terms of other major language traditions, allowing for the IT replacement of scarce translator resources in both diplomatic and military scenarios: as well as "smart" data-mining over global communications channels. Whereas current systems (such as Carnivore) target only certain key words, the true AI system would understand such targets in context, eliminating much need for subsequent human scrutiny. This is necessarily in addition to the more routine applications that target commercial interactivity, such as switchboard/receptionist.
6c) How can the program be used as a personal companion?
This somewhat monolithic model of stand-alone computing could eventually be transformed by transferring the bulk of processing directly into the considerable resources of the Internet. Accordingly, the key knowledge bases of the AI computer would be distributed as open source code over an extensive network of broad-band servers. The end user's computer only need run a stripped-down version of the MCU program, the inference engine of which would interface directly with the web-based knowledge base on a real-time basis. Indeed, the initial ground-work for such a standardized database network is already in the works with respect to the newly proposed Semantic Web, the brainchild of Tim Berners-Lee (the original innovator of the World Wide Web). This newly conceived Semantic Web proposes to bypass the limitations of the current human-web interface, implementing a machine-to-machine version aimed at standardizing the wealth of network information. In conjunction with provisions for a further AI interface, this futuristic AI assistant might eventually become a reality for all those willing to entertain such aspirations.
7) According to your "Ten Ethical Laws of Robotics", the system has been programmed with ethical guidelines. Some of these include behaviors such as guilt, worry, and hero worship. How will that be presented, and will the program actually acknowledge a sense of guilt?
This AI entity is readily able to learn through experience employing verbal conversation (or via a controller interface). Technically it would not be (reflectively) aware of its own emotions, as this is a purely subjective determination. It would, however, be able to simulate this feature through language, wherein convincing others of the fact. Furthermore, the basic core-motivational-kernel is based upon the well-established behavioral field of conditioning theory, adding a crucial scientific dimension to this formal language simulation.
8) Please tell me a little more about yourself, your background, and your motivation for the project. How does your background in counseling figure into your invention? You have written extensively about ethics... how is the ethical AI an extension of these ideas?
I am a 50 year-old counselor and author native to the Southern California area. My credentials include a Baccalaureate Degree in Biological Sciences from University of California Irvine, followed by a Master's Degree in Counseling from California State University Fullerton: with an emphasis in Marriage, Family, and Child Counseling. Postgraduate career choices include social work with abused children and as an educator in the field of Psychology. I am currently engaged in Private Practice in Family Mediation Counseling in the San Bernardino County area. The patent applications mirror my several book titles currently in print concerning the character values, as described further at: www.charactervalues.com
9) Would you consider the application to be mostly theoretical?
It is theoretical in the sense that this system has never been thought up before, and hence, completely new on the world scene today. Its implementation, however, should be fairly straightforward given the proper research and development, particularly if granted the status of a crucial initiative.
10) Is there anything else that you'd like to mention?
This innovation is completely novel in its ability to simulate emotionally charged language: an achievement that has previously eluded AI researchers due to the lack of an adequate model of motivation in general. Only the most basic of details is permitted here, barely touching upon the full degree of content contained within the complete 90 page patent. The serious reader, accordingly, is encouraged to defer final judgment until conditions permit a fuller evaluation (the complete patent specification is posted at: www.angelfire.com/rnb/fairhaven/patent.html
Sincerely
John LaMuth