LOPAPRa7T6 108181569PST, 72037000000000000
Mall, from MITM@MC Pevd at S8=APR=76 LOL8=PST
Date! 18 APR 1976 1129"EST

Froms JOSEPH at MITeMC -

Teo: Lederberg at SUMEX#AIM
Dear Joshue!

. You muat know that John MeCarthy has placed your Neots . review
of my book {fn hie "Eleetrie Magazine”, There exiats. a. standing
invitation te anyone, whe gen gein access to the ARPA net to browse
through thet magazines f found yeur review there, (CAftee ~ discovered
we Meld of people at MalaTe told me that. they had seen. tt weeks
eariler,

-o . Below Va: my eesponses at is whee J Ineend to send to the NeY,Ts
after your review appears there, J think you should be the one to dee{de
whether jt ja te be inserted in JMC!s Magazine before then,. I would ike
#0 see it there, I. fear the Times may not tike te. to be debated before
{t appears im thelr paces, however,. °

_ I did noty in my. response, touch on @ point about whieh I fee!
atrongiy: this {8 your twice stated observation that I am debating
positions which zealots held 20 years age, You say I qyete from thelr
weitings of 20 years ager for example... An analysis: of the references
that appear at the end af the book refutes that, | The mjor works I quote
are Simonian "Selences of the Areidiadat® (1969), Newell and Simons
"Hyman Problem Solving” (i972), Colby and Sehank's "Computer Models: of
Thoyaht and Language" (1973), and other tuehs Farrestor!s. peatlnony
before the U,8, Congress was given in 1970 -and is. stil! being, etreuleted
by him today, I could go on, . Remember alno that I was writing ip. 1973
and early 1974, You eanit expect many quotes from 1976, As for the
Bealots of 20 yeers egot Colby te a professor of peyehiatry et UCLA,
Mineky js Donner Professor of Sefance at MylaTps oefaeee is Professor of
CS and head of the AI lab at Stanford, Newell ¢s University Prafessor: at
CMUp and so om, _The impression you create that these "zealots: of fe
years pac! heve folded thelr tents and stientiy sifpped away. is
eontradiated by the fact that these very people censtitute Cfor better: op
worse) the prinaiple teedershin of AI today, (My respese to your review
does not mention this, is mes very senaitive point, becavee I dearly hope
and believe you wil! alide it from the final version of your review.

On the whole, I am flattered that you took the trouble to read
the book and comment on it,

Thanks for your help in all things,
Joe

A REPLY TO JOSHUA LEDERBERG
. by,
Joseph wel zenbaum

The last chapter of my book {8 entitled "AGAINST THE SMPEREANEOM:
OF INSTRUMENTAL REASON", A philosephye a way of thinking, of, for that
matter, @ nations ie imperialistic whan: lt sttenpta: to ominate a domain
lerger then thet to which te {fs ‘leg tinetely eneitied, . I. do not
"aatecorically reject Inetrumenteal reason in jes application. to: human
affaira® as Joatue Lederberg would haye es. | 1 relect the svetematiaally
euelusive use of instrumental reason tn, affaira that call.elee for ceason
{mn other forms, JT reject, ae f know Joshua pederbers dees: the kind -of
rationality thet views individual human. beings: and. human societies as
mere instruments te be used aa means: toward the user's ends,

. I try to say in my beeok that. the faituce to. dietinoul ah between
imperia}iatic. Fretrumental reason and sretcementel feason . properly
eenatrained jeads to an equation of pationaliey with ogteal (sys . Onee
that dlatinetion has been erased, ie 48: no lenger possible to fault the
thinking of those statesmen who saw in their nation a new} eh Preblem"
and quite rat Oey ee ae etianes e logical way to. "selve" (t weuld. be
to murder the Jewish popujation. of Europe,.. Nor could one. then: fault
those other atateamen who, in order to free their ally's count rvatde: of
guerrilles, gaye theim alpmen ifteense to kil! every living thing in ite
A Perfectly rational solution,

| Joshua Lederberg, believes thet. "those wha deify the machines
deserve the human sacr ‘lees that may reault from neglegting the: Auman
reeponsibility for moral decisians,” But he fears: power (nm tha hands of
"¢prational” man, I sey that. the Inperal atte use. of inserumentel reasen
¢e: a ge| ication of mechaniagic thinking and ¢@ 9 nagieet of human
reaponaibility for mora) deeisions, No ene gan. say that Ghose whe
practice it are “drrational", To the contrary, thelr madness conatere ir
that they carry rationality to absurd extremes, . And such madness {ts
perhars even mote to be feared than. 18 prational levy Pechape here {¢.4@
perverse morel galculus within which Hitler and his statesmen did deserve
the sacrifice of the Sixemilii{on, But did the Sixemi}ifen deserve te be
slaughtered?

— I toughed on the seope of the rescensibitity oF modern man in. my
heok, I wrote! |

"In blolicel times few peonie could do anything that was likely to
affect others beyond the boundertes.of thelr own Iiving spaces,
Man's esefence and technology have altered. this efreumstance
drastically, Not only gan madern manis eetians affegt the: whole:
Planet that is his habitat, but’ they ean determine the future of the

PAGE 2

entire human species, It follows, eee or eittetes es particularly
man the scjentiet and enaineer, has peaponelbilities that trenscend
hia immediate eltuationy thag. in fact. extend directty ‘ta future
generations, These responelb lithes ere especially grave since
future generations cannot advocate their own cause AoW, We are ai}
their ¢ruatees,"

It is simety net enough thet we here: and now predicate what we do on the-
knowledge that wer the actors, deserve the consequences: of our eetiona,.
a Finallys, Joshua Lederberg agrees with me that even the most
intelligent computer programa do not necessarily lead te. understandings
that they conatitute “experiments” to be "validated only by experionae,"
Raaders may share my unease over the prospect.e= and in many cases tt To
already mere than mere prospect e= that computer programs Cand selentists:
and. statesmen) wil). indedate large seale. "experiments" from whese:
experience we ean learn only too later and then only that we ought fever
have abdicated aur responsibilities te them, The vietims ef hydregen
bembs launehed by a computer controlled antdeballistie missite system
(ABM) woulds of courses have AO regrets, .

: « .
Sseenseuan