Can a machine exercise judgment when faced with circumstances its maker did not anticipate?
Yes4
- 1864Babbage, Passages from the Life of a Philosopher
I endeavoured to ascertain the opinions of persons in every class of life and of all ages, whether they thought it required human reason to play games of skill. The almost constant {466} answer was in the affirmative. Some supported this view of the case by observing, that if it were otherwise, then an automaton could play such games. A few of those who had considerable acquaintance with mathematical science allowed the possibility of machinery being capable of such work; but they most stoutly denied the possibility of contriving such machinery on account of the myriads of combinations which even the simplest games included. On the first part of my inquiry I soon arrived at a demonstration that every game of skill is susceptible of being played by an automaton. Further consideration showed that if _any position_ of the men upon the board were assumed (whether that position were possible or impossible), then if the automaton could make the first move rightly, he must be able to win the game, always supposing that, under the given position of the men, that conclusion were possible. Whatever move the automaton made, another move would be made by his adversary. Now this altered state of the board is _one_ amongst the _many positions_ of the men in which, by the previous paragraph, the automaton was supposed capable of acting. Hence the question is reduced to that of making the best move under any possible combinations of positions of the men.
SummaryEvery game of skill can be played by an automaton, because a machine that can choose the right move from any position of the pieces can answer whatever its opponent does.
Chapter XXXIV, 'The Author's further Contributions to Human Knowledge' (section 'Games of skill can be played by an automaton')
- 1914Quevedo, Essays on Automatics
No ocurre lo mismo cuando se trata de si será posible construir un autómata que, para determinar sus actos, pese las circunstancias que le rodean. Se piensa, generalmente, que sólo en algunos casos muy sencillos puede conseguirse esto; se cree que es posible automatizar las operaciones mecánicas, puramente manuales de un obrero, y que, por el contrario, las operaciones que exigen la intervención de las facultades mentales nunca se podrán ejecutar mecánicamente. Esta distinción carece de valor, pues, exceptuando el caso de los movimientos reflejos, de los cuales no hemos de ocuparnos aquí, en todas las acciones humanas intervienen las facultades mentales. Intentaré demostrar en esta nota — desde un punto de vista puramente teórico — que siempre es posible construir un autómata cuyos actos, todos, dependan de ciertas circunstancias más ó menos numerosas, obedeciendo á reglas que se pueden imponer arbitrariamente en el momento de la construcción. Evidentemente, estas reglas deberán ser tales que basten para determinar en cualquier momento, sin ninguna incertidumbre, la conducta del autómata.
The same is not true when the question is whether an automaton can be constructed that weighs its surrounding circumstances in order to determine its acts. It is generally thought that this can be achieved only in a few very simple cases; that a worker’s purely manual, mechanical operations can be automated, but that operations requiring mental faculties can never be performed mechanically. This distinction has no value, since, except for reflex movements, which do not concern us here, mental faculties intervene in all human actions. I shall try to demonstrate in this note—from a purely theoretical standpoint—that it is always possible to construct an automaton all of whose acts depend on certain more or less numerous circumstances, obeying rules that can be imposed arbitrarily at the time of construction. Evidently these rules must suffice to determine the automaton’s conduct at any moment without uncertainty.
machine translation
SummaryTorres rejects the mental/manual distinction and proposes automata whose actions are determined by circumstances under rules fixed at construction. Arbitrariamente means rules freely specified by the constructor, not random conduct. This passage does not establish learning beyond anticipated cases.
Revista de la Real Academia de Ciencias 12 (1914), p.394
- 1943Rosenblueth, Behavior, Purpose and Teleology
Machines governed by negative feedback correct their course toward a goal from information about their own errors, so purposeful behavior is not unique to organisms.
Philosophy of Science 10, no. 1, pp. 18–24
- 1950Turing, Computing Machinery and Intelligence
An important feature of a learning machine is that its teacher will often be very largely ignorant of quite what is going on inside […]
SummaryTuring distinguishes a trained machine from a computation whose internal state the operator aims to follow. The omitted continuation allows some prediction of its behaviour; opacity alone does not establish judgment. Brief extract: the full pp.458–459 discussion should be reviewed before making a sorting pool.
Summary disputed by reviewer: Brief quotation checked against the original scan; full passage and inherited draft category need review.
Mind 59 (1950), section 7, p.458; paragraph continues on p.459
No3
- 1637Descartes, Discourse on Method
The second test is, that although such machines might execute many things with equal or perhaps greater perfection than any of us, they would, without doubt, fail in certain others from which it could be discovered that they did not act from knowledge, but solely from the disposition of their organs: for while reason is an universal instrument that is alike available on every occasion, these organs, on the contrary, need a particular arrangement for each particular action; whence it must be morally impossible that there should exist in any machine a diversity of organs sufficient to enable it to act in all the occurrences of life, in the way in which our reason enables us to act.
SummaryMachines act only from the disposition of their organs, which must be arranged for each particular action, while reason is a universal instrument usable in any circumstance.
Part V
- 1836Poe, Maelzel's Chess-Player
But the case is widely different with the Chess-Player. With him there is no determinate progression. No one move in chess necessarily follows upon any one other. From no particular disposition of the men at one period of a game can we predicate their disposition at a different period. Let us place the first move in a game of chess, in juxta-position with the data of an algebraical question, and their great difference will be immediately perceived. From the latter — from the data — the second step of the question, dependent thereupon, inevitably follows. It is modeled by the data. It must be thus and not otherwise. … But in proportion to the progress made in a game of chess, is the uncertainty of each ensuing move. A few moves having been made, no step is certain. Different spectators of the game would advise different moves. All is then dependent upon the variable judgment of the players. Now even granting (what should not be granted) that the movements of the Automaton Chess-Player were in themselves determinate, they would be necessarily interrupted and disarranged by the indeterminate will of his antagonist.
SummaryCalculation is determinate and can be mechanized, but in chess no move follows necessarily from the last and the opponent's will interrupts any fixed sequence, so a pure machine could not play.
Examinons encore quelques-unes des difficultés que la machine doit surmonter pour que son assimilation à l’analyse soit plus complète. Il y a certaines fonctions qui doivent changer de nature lorsqu’elles passent par zéro ou l'infini, ou dont les valeurs ne sauraient être admises au delà de ces limites. Lorsque ces cas se présentent, la machine au moyen d’une sonnerie peut avertir que le passage par zéro ou l'infini a lieu , et alors elle s’arrête , jusqu’à ce que le gardien l'ait remise en action pour telle autre opération qu’on voudra lui ordonner, Si cette opération est prévue, la machine, au lieu de sonner, se disposera de manière à présenter les nouveaux cartons relatifs à l'opération qui doit succéder au passage par zéro et l'infini: Ces nouveaux cartons peuvent faire suité aux premiers et n’entrer en jeu que lorsque les deux'cireonstances indiquées précédemment auront lieu.
Let us examine further some of the difficulties which the machine must overcome for its assimilation to analysis to be more complete. There are certain functions which must change their nature when they pass through zero or infinity, or whose values cannot be admitted beyond these limits. When these cases arise, the machine can give warning by means of a bell that the passage through zero or infinity is taking place, and then it stops until the attendant has set it going again for whatever other operation one wishes to order it to perform. If this operation has been foreseen, the machine, instead of ringing, will arrange itself so as to present the new cards relating to the operation which is to follow the passage through zero and infinity. These new cards may follow on from the first ones and come into play only when the two circumstances indicated above occur.
OCR correctedmachine translation
SummaryThe engine cannot interpret a result that is not self-evident, such as a value becoming infinite; it can handle the case only if the programmer foresaw it and punched a card for it.