Constitutional Law and Artificial Intelligence: The Potential Legal Recognition of Computers as "Persons"
Proceedings of the Ninth International Joint Conference on Artificial Intelligence. General. Pages 1271-1272.
© 1985 International Joint Conferences on Artificial Intelligence
Abstract
American Cor.stituti.on<ü law grants the status of "persoa • to the rnembers of certain groups, but denies that status to other groups. Various legal analogies could be used to determine whether such status should be extended to conputer systems and, 1r so , what limitations should be placed upon thaL re~ognition.* The concept of legal "personali ty" in United States constitutional law has changed considerably over the years. An ever-increasing nurober of groups of ascertainable entities have been recognized as persans under the law, and the rights and Obligations accruing to the members of those groups have changed more Lhan they have remained constant. The general question is th us presented of whether computers might be recogni~ed as persons; it is submitted that the correct answer is a qua lified "yes".
Full text extracted automatically from the PDF; figures, tables and equations may differ. The PDF is the version of record.
CONSTITUTIONAL LAW AND ARTIFICIAL INTELLIGENCE: THE POTENTIAL LEGAL RECOGNITION OF COMPUTERS AS "PERSONS"
Marshal s. Willick Attorney at Law Thorndal, Backus, Maupin & hanouki an 333 North Rancho Drive, Ste. 333 Las Vegas, NV b9l06
ABSTRACT
American Cor.stituti.on<ü law grants the status of "persoa · to the rnembers of certain groups, but denies that status to other groups. Various legal analogies could be used to determine whether such status should be extended to conputer systems and, 1r so , what limitations should be placed upon thaL re~ognition.*
The concept of legal "personali ty" in United States constitutional law has changed considerably over the years. An ever-increasing nurober of groups of ascer tainable entities have been recognized as persans under the law, and the rights and Obligations accruing to the members of those groups have changed more Lhan they have remained constant. The general question is thus presented of whether computers might be recogni~ed as persons; it is submitted that the correct answer is a qua lified "yes.·
The legal histories of several groupG, such as blacks and wonen, have followed the patter~ of their initial "recognition" as legal persons, follm1ed by a slow accretion of rights and obliga tions. Viewed another way, the ini tial recognition of the members of these groups started a lengthy period during which the legal gap between them and previous ly recognized persons narrowed.
Decision-makers have used various rationales over the years in extending legal recognition to new groups. The individual decisions tend to reflect the values of their times and do not shed much light on the essence of legal personality. Bach such extension, however, constituted an acknowledgment that the individual entities being considered were more like the persons doing the considering than like the property belanging to those
*This position paper is largely abstracted fr01:1 \T'. J l i c:k 1 Art ificial Intel! igence: Some Legal Approaches and Implications, AI !lag., Summer, 1983, at 5, which contains
authorities and background information.
persons. As a group, those extensions mark an expanding societal defini t ion of "person."
It may fairly be said that, as time has passed, the test of personal i ty has focused more on behavior than on appear ance , and more on mental traits than on physical ones. Western think ing has come to disfavor tests of persondlity based on "status" (ownership of property, religious affiliation) or "structure" (gender, race); those tests were invoked, '"''d defeated, almost every time a ne\11 group was added to the roster of persons. Accordingly , recogni t i on has generally been extended to ~roups whose members have demonstrated t.he capaci ty t.o behave in a manner indicating that they think Dore like legal persons than like--äDything eJse .
The initial question is therefore whether there could exist circum~tances such that a decision-maker could exarnine the behavior of a computer system and decide that the machine had crossed the thresho ld of legal personality. It seems nearly inevitable that the issue will arise in our increasingly computerized society; science fiction literature abounds with propoued fact ual seenarios in which that legal issue could be present ed. The real que::;tion is whether current or foreseeable law provides a plc.u.sible foundati on for a determination of computer personality .
It i s possible that a decision concerning computer personality could come from the executive or legislative branches of government. Popular opin ion can find expression through those chanriels when the judiciary is unwilling or incapable of treating ascertainable entities as persons even though society as a whole per cei ves them as such. An example of such a legislacive determination is the Twen ty-sixth Amendment (forbidding discrimina
tion against 18-21 year olds).
Perhaps example of mandate was the most obvious historical recognition by executive t.he freeing of black slaves
who had previously been considered lawfully recoverable items of propertJ. Congress at one time declared itself incapa~le of ewancipation, and the courts rnainly addressed thc irnpacL of local recaption statutes and Lhe Fugitive Slave Letws upon local k idnapping laws. An essentially Executi ve act ion ( u rari ty in the field of extensions of legal person ulity) , swiftly followed by a series of Const i tut ional amendments, presumed to change the slaves' legal &Latus in one step from that of ascertainable individual items of properLy LO that of fully fran chised persons.
In common law systems, however, most "law· comes frorn the decisions of courts. Disputes arising from novel circurnst~nces are rf~conciled by courts that att.empt to draw ~omparisons between t.he facts of the cases before t:bem c:nd the facts of cases previ ou.sly dec i ded . ':!,ouched on below are a few of the pos sib.i.e an aloyies provided by histor y that could ue seized upon by cuurts seeking to resolve the question of ~omputer personality. Which analogy is uti 1 i:.:ed in a gi ven case could well determine the result of the legal dispute.
It is submitted that the first. computers to atta in "personhood · will do su indi vidually, if at all, because they will be forced to prove that they are more than the "mere machines" they will be presumed to be. Courts are likely to hes itate before extending to comput ers the legal precept that "all men are created equal ," because even today they can be purposefully designed tu have any of a broacl range of operating characteristics and capabilities.
Given t.he enormaus variety of the machir . .-,s called "computers ," even propo nents of Computer personality will probably concede that some will possess the nec~~sary characteristics for that statc.s while others vlill not. liany problems , such as the evaluat.ion of computers of the same model, the impact of a capacity for significant machine learn ing, etc. , remain to be addressed. such questions, however, will most probably be decided through use of the terminology and tests developed in other legal disputes.
Courts seeki ng a definition of "per son " m.ight 1 ook to the abortion decisions, which draw distinctions based on the degree of individual development ( tr imestPrs); by aualogy, any individual computer exceeding a minimum behavioral capaci ty r oughly equating fetus wviabil ity" would be presumed to be a person . hS with thP abort ion decisions, a singl~ such decision concerning computer perconality cuuld affect many more persans than the parties before the court.
When a human person dies, he loses all of his rights. The law in. this area tends to set an over -inclusive minimum, so t.hat any human but one who can be show11 to have died t:ends t:o be def .ined as "al i ve." Given the recent emergence of "brain death" as a critical factor, and since rnany computers today can exhibit far more "intelligent" behavior than that of comatose human bPiny s (who do enjoy legal recognition), a legal rninimum standard test of personality could probably be satisfied by a comput.er system in the proper circumstances .
The emergence of the r.toder:, cor iJO rC:ttion provides the most subtle means by which computer systems might achieve legal recognition. Cerparations have names, can buy and sell property, and can commit crimes, but they cannot be drafted, be married, or vote. They are persons, but they are owned, constituting a recognized class of non-human persans that has legal right:s and Obligations peculiarly tailored t:o the unusual attributes of its members.
An analogy between such "a rt.ificial" persans and cor.tpuLer :Jystems will appear Jes s strained than camparisans with human beings. Additionally, to the degree that t:hc operations of a corporation can be computerü.ed , the corporation and the computer would effectiveJy be t:h ~ sar.te entit:y: no legal change would be required for such de facto recognition of computer personality. -----
Corporutions provide an example of the concept of partial personality, whereby an ascertainable entity may be recognized i:tS o legal person for one purpese but not another. The concept has many applicat.ions and is not limited to non-humans; certain laws treat fetuses as "persons" for the purpese of inheritance, while others provide that the abortion of such fetuses is not generally to be con sidered murder.
The concept is applied in many ways in modern society . Hinors, for example, slowly accrete rights and obligcn:ions as they grow older because of their .eresurneci capacities, while rights are removed tror.: the r etarded and the .insane when thei r behavio r proves to be too far below or ÖutsTde- the societal mi nimuw. The les;al system is thu& equipped wit.h a variety of approaches with which to decide the extent and variety of rights that should be given to computers that are recognized as persons.
Computer systems that perform in creasingly complicated tasks in an in creasingly competent manner will be thrust onto these shifting sands of constitution al presumptions, tests, and Standards.
Since there does not seem to be an analyticall:z' sound test of ·personality" that will exclude computer systems which behave intelligently, the question of legal recognition will remain one of "when" and not "if" until and unless some absolute limitations on the abilities of such machin es can be demonstrated. Once computer systems can satisfy establ ished legal tests of personality, either a valid ground of distinction between them and humans will have to be found, or the distinction will have to be abandoned as mere prejudice.
Current artificial inr.elligence reseu.rch increases t.he need for a prompt examination of these problems. Courts have already begun to in1pose on certain professional:.; the rec..ui.rement of use of certain computer systems.* As applied artificial int.elli.gence techniques cause computer systems to behave in ways tradi tionally associated with hurnan intelli gence, the likelihood of legal scrutiny of the status of those systems increases.
No uniformly recognized definit.ion exists for intelligence, so it is not sur prising that there are at least four different (and largely contradictory) definitions of "artificial n intelligence. Hhile there appear to be many instances in which mea~urable intelligence is unneces sary to recognition of legal per;.;onalit}· (corporations, comatose humans, etc.), s uch recognition appears to be mandated under modern tests whenever such tntelli gence io l!S_e� 'O~lt .
A traditional legal test asks whether "r easonable men could differ" as to a proposed question. If they could, the test allows the question to be subrnitted to a judge or )ury as a suestion of fact rat. her than one of law. Given the many legal tests of personality, it is submit ted that there is (or soon will be) a quese ion of fact as to. whether a computer which appears to be exhibiting intelligent behavior is a "person" under the law. Given the appropriate facts and a sympa thetic jury, that question will at some point. be answered in the affirmative.
Developments in biotechnology could lead to recognition of cercain computer systems even if society prove.:; unwilling
*For a ci tat ion-saturated di scuss .ion of the rami ficat ions of professional compu terization, see Willick, Professional llalpractice and the Unauthorized Practice of Professions: Some Legal and Ethical Aspects ot the Use of Computers as lJeci sion-Aids, which is due to be published this Fall in Rutgers Computer and Tech nology Law Journal.
to recognize Computers per se . humans do not endanger their legal recognition by using devices t.o enhance or replace parts of thernselves; the legal tesc is subtractive, and it presumes continued recognition. Metaphysical considerations aside, only technologica l (as opposed co scientiric) barriers appear to e:dst to the eventual direct integrat ion of hm.wr. brains and computers. No reccgnition enddngering event would occur by such int.egration; a "computer-enhanced" person would retain recognition.
Presuming chat the computer could be made able to perform various tasks as the human lost the ability to do so, no behavioral differences would appear as the human parts failed. Because the tradi tional legal test looks to behavi or , might the mechanical remnant of such a person retain legal recognit.ion ?
Those arguing otherwi ~e would face the difficult task or couvincing a court that the coml..>i!Jation had lost its right to recognition at some tim~f biologtca l failure, despite its continuati on of it:.; normal activities. The law abhors the removal of rights absent. behavio r outside of or below certüin roinimum requirements, goinc; so far as co rule that permanently comatose humans remain persons. Gi ven t.he foreseeable behavioral capacity of comput er systems, i t is submi.tted that mechan ical remnarocs of human/computer combina tions will retain legal recognition, at least where the takeover of once-human functions is gradual.
such developments would present the difficult legal question of how to distin guish between two computer systems, one of which slowly took over the function s of a human br~in, and the other of which simply rolled off of an assembly line. Presuming equal behavioral ca~bci~ies (or even close ones; the Jaw recogn i zes bot.h geniuses and idiots), no valid ground appear s t.o r-:;d.s t for ihe denial of legal recognition to tbe syst em that was never connected to a human braiu.
CüllCLUSiüN
Compucers today are increasingly behaving in ways traditionally identifi ed as exhibiting consciousness, unde rstand ing, and leatnins. It may prove impossi ble in t.he future to draw a valid legal distinction between humans and computers, eit.her because of the increased behavioral capacity of the latter group, or because the t.wo groups will be literally, phys ically, inseparable. At that time, constitutional law will recognize at leas t some computer system~ as "persons."
@inproceedings{ijcai1985p254,
title = {Constitutional Law and Artificial Intelligence: The Potential Legal Recognition of Computers as "Persons"},
author = {Willick, Marshal S.},
booktitle = {Proceedings of the Ninth International Joint Conference on
Artificial Intelligence, {IJCAI-85}},
publisher = {International Joint Conferences on Artificial Intelligence},
editor = {Aravind Joshi},
pages = {1271--1272},
year = {1985},
note = {General},
url = {https://ijcai.org/proceedings/1985/254},
}
TY - CPAPER TI - Constitutional Law and Artificial Intelligence: The Potential Legal Recognition of Computers as "Persons" AU - Willick, Marshal S. T2 - Proceedings of the Ninth International Joint Conference on Artificial Intelligence, IJCAI-85 PB - International Joint Conferences on Artificial Intelligence PY - 1985 SP - 1271 EP - 1272 UR - https://ijcai.org/proceedings/1985/254 ER -Copied
Machine-readable versions of this paper, for tools and AI agents: