Accelerat ing t he world's research.

Artificial intelligence and the

ideology of capitalist reconstruction
Bruce Berman
AI & Society

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

T he Social Reconfigurat ion of Art ificial Int elligence: Ut ilit y and Feasibilit y
James St einhoff

[Kevin_ Robins,_ Frank_ Webst er_ (aut h.)]_ T he_ Technica(book4you.org).pdf

kevin robins

Cont emporary Technology Discourse and t he Legit imat ion of Capit alism
Eran Fisher
AI & Soc (1992) 6:103-114
1992 Springer-Verlag London Limited ALl ~ S O C I E T Y

Artificial Intelligence and the Ideology of Capitalist

Reconstruction
Bruce J. Berman
Queen's University, Kingston, Ontario, Canada

Abstract. The growing interest in AI in advance capitalist societies can be understood not
just in relation to its practial achievements, which remain modest, but also in its ideological
role as a technological paradign for the reconstruction of capitalism. This is similar to the
role played by scientific management during the second industrial revolution, circa 1880-
1930, and involves the extension of the rationalization and routinization of labour to
mental work. The conception of human intelligence and the emphasis on command and
control systems of much contemporary AI research reflects its close relationship with the
US military and corporate capital, which are the sources of many of AI's key metaphors
and anolgies.

Keywords: Artificial Intelligence; Ideology of; Politics of; Capitalism; Crisis; Technologi-
cal change; Management.

Technological Paradigms, Hegemony and the Labour Process

After thirty years as a marginal scientific discipline dogged by a 'freaky' image, as

one of its few historians put it (Fleck, 1982: 175, 209), AI emerged during the
1980s as a major focus of military and corporate interest. By 1986, according to
the Industrial Relations Center at Cal Tech, the market for expert systems (at the
moment, the most commercially attractive of AI 'products') was "expected to
erupt from an estimated $35 million this year [1986] to nearly $2 billion by 1990"
and " m o r e than half of the Fortune 500 (including Hughes, Litton, General
Dynamics, Westinghouse, IBM, GM, G E and Lockheed) have AI projects under
way" (Cal Tech, 1986). While the commercial market failed to achieve expected
levels, the slack was taken up by the US Department of Defence, which
quadrupled annual spending on AI research and development contracts and
allocated a total of almost $1 billion by the end of the decade. Almost three
quarters of computer science research, including AI, is now funded by the
Pentagon (Athanasiou, 1989: 116-17).
104 B . J . Berman

Why did this happen to a seemingly abstruse field such as AI and why during
the 1980s? The basis for an answer to these questions lies in the search for a means
of lifting the global capitalist system out of the recurrent crises and recessions of
what appears to be another period of major structural change. The intercon-
nected technologies of micro-electronics, computers and telecommunications
play a central role in this reconstruction of capitalism both as new dominant poles
of capital formation and acctimulation, and as the means for the radical change of
the production processes and organizational structures of existing industries. In
this context two areas of research - robotics, which employs some aspects of AI in
the pursuit of more conventionally understood types of automation, and expert
systems, which are more directly a part of AI and attempt to replicate key areas of
human technical expertise - are being rapidly taken up as potentially profitable
investments and as means for sustaining the competitiveness and profitability of
large corporations.
In this essay I would like to focus on a less widely analyzed aspect of AI: its
character as an ideological discourse and its emergence as a potentially key
element in the reformulation of the ideological hegemony of capitalism that will
accompany its structural transformations. As Christopher Knee (1985: 123) has
pointed out, the debate over AI is really a forum for political arguments, and AI
specialists have constructed a powerful ideological discourse which, as we shall
see, is likely to retain its importance even if the actual practical achievements of
AI turn out to be considerably less than expected.
AI as an ideology is beginning to reshape certain central conceptions we have
of the capabilities of humans and machines, and of how the two can and ought to
be fitted together in social institutions, which may become a conventional wisdom
of unspoken and untested assumptions that make the very conception of
alternatives impossible. AI can thus be seen as a potential hegemonic principle
within the sphere of formal organizations which facilitates the 'fit' of human
beings into the revised structures of a capitalism based on micro-electronic and
information technology, and ideologically contains, and significantly mutes,
resistance and social conflict.
AI can be understood as a key component of a new 'technolgical paradigm' for
the Third Industrial Revolution that is now in progress. According to Christopher
Freeman, "a technological revolution represents a major change of paradigm,
affecting almost all managerial decisions in many branches of the economy", and
the new 'techno-economic paradigm' is "a new set of guiding principles, which
become the managerial and engineering 'common-sense' for each major phase of
development. ''l (1984: 499). Technological paradigms have provided the tacitly
accepted assumptions for the design of technology and organization, and for the
appropriate use of living human labour, in the historical phases of the develop-
ment of capitalism. In so far as the major elements of the technological paradigm
become, despite resistance, diffused through the popular culture of a society, it
becomes a major component of hegemonic ideology and helps to contain labour
struggles within limits that do not challange the essential structures of the system
and make coherent alternatives increasingly difficult to conceive. 2
The ideological importance of AI can best be understood as analogous to the
role played by scientific management in the second industrial revolution around
Artificial Intelligence and the Ideology of Capitalist Reconstruction 105

the turn of the century. AI seeks to achieve the same control over mental
processes that scientific management sought to achieve over physical labour
through a process of rationalization, fragmentation, mechanization and routini-
zation.

Scientific Management and the Second Industrial Revolution

The two generations on either side of the turn of the 20th century, roughly
between the great capitalist crashes of 1873 and 1929, were an epoch of profound
change in almost every sphere of Western capitalist society. This was the period
of the Second Industrial Revolution, marked by the shift from steam to electrical
power and the start of the linkage between science and technological innovations
in a range of new industries - electricity, chemicals, telecommunications,
pharmaceuticals, petro-chemicals - dominated almost from the start by large
oligopolistic corporations. These are also the years of intensified class conflict,
which saw the emergence of modern trade unions and mass political organiz-
ations, the consolidation of liberal democratic states and the beginnings of social
reform and welfarism. It was also the period that saw the development in Western
universities of all of the modern disciplines of engineering and applied science
and of the social sciences.
In his influential study, Labor and Monopoly Capital, Harry Braverman
identified scientific management as articulated by Frederick Taylor as "the
bedrock of all work design" and the dominant force shaping the labour process
and corporations of twentieth century capitalism. Taylorism, according to
Braverman, was grounded on an assumption of the stupidity and laziness of
workers and focused on the control of a refractory labour force in a setting of
antagonistic social relations. It attempted to achieve control over the most minute
details of work by placing in the hands of management every decision made in the
course of work, and the knowledge on which it was based. This separation of
mental and manual labour, conception and execution, and the monopolization of
the former by management, led to the degradation of the skills of the workers and
the loss of control over their own labour. Work was fragmented and routinized
into the rapid repetition of simple, abstract physical movements designed by
management. The achievement of such control over the factory floor required the
separation of the sphere of production from a growing sphere of pfanning,
organizing and calculating staffed by rapidly increasing numbers of managers,
engineers, designers and clerical record keepers (Braverman, 1974: Chs. 4-5).
As an ideology, Taylorism played a crucial role in the development of scientific
management as a set of concepts and principles regarding the rationalization of
the labour process and the design and linkage of machinery and living labour. It
did not, as such, represent a major intellectual innovation, but rather a
technically sophisticated refinement and application of the capitalist task division
of labour. Scientific management, whether in the versions of Taylor and his
imitators or rivals, was the ideology that emerged within the second industrial
revolution to guide and legitimate a widespread process of organizational and
106 B.J. Berman

technological restructuring and innovation in the new core industries (Littler,

1982; Wood, 1982).
The reconstruction of capitalism in the 1880-1930 epoch was marked by fierce
but uneven worker resistance in the US and Europe, and the success of capital in
imposing the new methods and machines varied widely in different contexts.
Nevertheless, scientific management provided a conventional wisdom or
common sense for managefs and engineers that assured them that there was
always a better, more efficient and more profitable design of work and machine
possible and brought them back to try again and again where they had initially
failed (Littler, 1982: 180-181). The invocation of the authority and method of
'science' and constant insistence on the essential neutrality and objectivity of its
results made scientific management into a powerful ideological weapon to
bludgeon the 'irrational' and self-serving resistance of workers. If nothing else,
scientific management gave capital a sense of mission and a belief in the necessity
and inevitability of its achievement. This permitted repeated assaults by capital
on previously established patterns of the forces and relations of production,
themselves the historical product of earlier confrontations and struggles, through
the imposition of new forms of work organization and technology.
Scientific management, however, also had a direct relationship with technolo-
gical development and a central rote in the technological paradigm of the second
industrial revolution. Charles Babbage noted in the 1830s how the division of
labour, by reducing the process of production to an ordered sequence of simple
steps, provides the organizational and conceptual basis for mechanization:
"When each process has been reduced to the use of some simple tool, the union of
these tools, actuated by one moving power, constitutes a machine" (Babbage,
1832: 174). ~ Furthermore, the machinery, once in place, facilitates managerial
control of the labour process by generating an apparently objective and neutral
technological imperative for particular forms of behaviour. As Marx noted in a
widely known passage, capital "possesses in the machine system an entirely
objective organization of production, which confronts the worker as a pre-
existing material condition of production" (Marx, 1977: 508).
Whatever its practical limitations, scientific management facilitated the ideolo-
gical depoliticization of the labour process and its depiction as the site of apolitical
technical relations ordered by rational calculation and the maximization of
efficiency, and encouraged the massive extension of mechanization through the
fragmentation and routinization of work. It also expedited the systematic
integration of science into the service of capital as a source of profitable
innovation of products and processes, while isolating workers from knowledge of
scientific development and virtually eliminating them as a source of technological
change. Finally, scientific management persisted as the technological paradigm of
capitalism even after the Keynesian compromise of the post-1945 era, in so far as
the commitment to full employment and the regularization of bargaining between
capital and labour organizations, on the basis of increases in real wages in return
for increases in productivity, left the design of work and technological innovation
in the hands of management.
Artificial Intelligenceand the Ideologyof CapitalistReconstruction 107

Computers and the Third Industrial Revolution

Telecommunications, microelectronics and computers comprise the technologi-

cal foundation for the contemporary restructuring of the global capitalist system.
They are being used, in current management jargon, to 'downsize' bloated
corporate structures and make them 'taut', 'flexible', and 'competitive'; to reduce
the reliance on human labour through increasingly automated design and
production; and to increase the centralized control and coordination of corporate
operations at a number of different levels: First, the synthesis of computers and
telecommunications systems is being employed to enable international corpor-
ations to increase the centralized and simultaneous monitoring, integration and
control of widely dispersed and differentiated organizational units. Second, on
the factory floor microelectronics and computers are being employed to dramati-
cally increase the automation of production through the development of CAD/
CAM systems and robotics. Third, within the office the technology is being
applied to increase the productivity and reduce the numbers of clerical and
professional employees (Kaplinsky, 1984; Shaiken, 1985; Schiller, 1986; Mosco,
1989).
This restructuring has been aggressively supported and encouraged through
extensive public subsidization of R&D in microelectronics and computing by
national governments increasingly uneasy over the anticipated social and political
consequences of falling behind in the intensifying competition among capitalist
economies (Flamm, 1987). In the United States, the government has used
defense expenditures, especially through the Strategic Defense Initiative (SDI)
or 'Star Wars' program, to provide funding for corporate research and develop-
ment in information technology and robotics in a way that helps sustain corporate
profits and shields the major defense contractors from the effects of recession and
competition (Mosco, 1989; Athanasiou, 1989).
All of these attempts to transform the existing forces and relations of
production constitute an assault on the established patterns of employment,
income, skill and status that characterized capitalism after World War Two for
professional and clerical employees as well as production workers. Not surpris-
ingly, as Christopher Freeman rather delicately puts it, "there may be long delays
in social acceptance of revolutionary new technologies, especially in areas of
application far removed from the initial introduction. Legislative, educational
and regulatory changes may be involved, as welll as fundamental changes in
management and labour attitudes and procedures" (Freeman, 1984: 498). In
other words, resistance to the organizational and technological reconstruction of
capital engenders ideological struggle as the old technological paradigm loses its
hegemonic force in the new circumstances of confrontation and a new rationale is
sought to legitimate the imposed changes. Analysis of this ideological struggle
has, for the most part, focused on 'neo-conservatism', particularly in Britain and
the US, with its free-market rhetoric and attacks on the welfare state and labour
unions, as an attempt to free the hand of capital to carry out the needed
restructuring. But neo-conservatism cannot adequately serve as a technological
paradigm shaping ideology and practice in the sphere of production itself. It is
there that the unique potential of AI has emerged.
108 B.J. Berman

The Epiphany of Artificial Intelligence: Deus Made Machina

From its earliest years in the 1950s AI has been characterized by extravagant and
as yet largely unfulfilled claims about the imminent development of an 'intelligent
machine' capable of matching and exceeding the capabilities of the human mind.
Repeated failures to reach these goals has never deterred AI specialists from
continually renewed pronouncements about the great achievements to come
(.Roszak, 1986: 122). Such hyperbole has, from the start, received extensive and
often breathlessly naive coverage in the press and in popular science writing, with
little follow-up about actual achievements. "If we succeed in taking the next step
forward", intones one typical example of this media hype, "we will very likely
face a future of inconceivable wonders, a golden gateway opening out into the
cosmos" (Ritchie, 1984: 10). 3
The immodest claims of AI practitioners have been matched by their
imperialistic attitudes towards non-computational forms of knowledge, with
attempts to colonize these fields with their own theories. Armed with a zealot's
belief in AI as a new universal paradigm for understanding the world, they tend to
dismiss all previous modes of knowledge as invalid if they cannot be readily
assimilated to AI forms. What AI cannot replicate in its programs, is simply no
longer worth knowing (Turkle, 1984: 260).
This behaviour probably derives from some of AI's distinctive characteristics.
First, for most of its history, it has been an extremely small and marginal specialty
involving a relatively ingrown community of a few hundred researchers at a
handful of universities located primarily in Britain and the United States. Even
some of its practitioners have recognized its image as a "bad child of the science
9. . flaky, crackpot and futuristic" (Schank, 1984: 213-14). This has led to the
active search for and dependence upon powerful patrons to both supply reseach
funds and implicftly confirm AI's scientific legitimacy. This patronage has come in
the US almost entirely from the military, primarily the Advanced Research
Projects Agency (ARPA) of the Department of Defense. In the last few years the
funding of AI research has been increased by hundreds of millions of dollars from
the Strategic Computing Initiative, intimately linked with the SDI (Athanasiou,
1989; Mosco, 1989; ch. 6). Second, the increase in government-funded research
and the recent recognition by major corporations of the commercial potential of
AI has introduced the possibilities of enormous wealth and greed into the game.
It is the rare AI practitioner today who does not wear the cap of the eager
promoter and entrepreneur as well as that of the university researcher. 4
The dependence of AI research on military funding continues a long estab-
lished pattern in the economic and technological development of the United
States, and has served as the basis for its linkage with capitalism and conservative
politics. From the early 19th century the military agencies of the United States
have directly encouraged major innovations in technology and production
management through research and development contracts, design specifications,
testing facilities and the support of particular firms as 'chosen instruments' for the
development of technologies from interchangeable parts for muskets, to com-
puters and automated machine tools (Noble, 1984; Smith, 1985; Flamm, 1988).
The provision of research and development funds through the Strategic Com-
Artificial Intelligence and the Ideology of Capitalist Reconstruction 109

puter Initiative has also brokered the relationship between AI and capital by
stimulating the interest of major defence contractors. While the Star Wars
program may never provide the claimed shield against nuclear attack, it is, as
Mosco points out, already working to the benefit of major American economic
interests by providing public funding of research in robotics, CAD/CAM and
expert systems that can modernise and reconstruct American industry (Mosco,
1989: 153-61). Furthermore, through its connection with Star Wars, a fundamen-
tal component of the political strategy of technological omnipotence and 'techno-
patriotism' of the American Right, AI was connected with and has become an
ideological adjunct of a conservative political agenda (Mosco, 1989: 165-70;
Athanasiou, 1989: 123). The linkage was recently demonstrated in a lavish new
volume celebrating the present and future achievements of AI, which contains
numerous brief essays by AI luminaries and a concluding selection by George
Gilder, one of the most strident conservative apologists for American capitalism
during the 1980s, who extols AI as a 'technology of liberation' (R. Kurzweil,
1990: 454-57).
Human liberation does not seem to be one of the motivations of either those
who fund or those who carry out AI research, however. While AI researchers
believe that they are engaged in something new and revolutionary in the history
of human thought, they are actually increasingly part of the latest wave of the old
phenomenon described above: the application of the division of labour in
bureaucratic organizations in order to routinize and control the labour process
(Berman, 1989). What is novel, however, is the effort to extend the logic of
rationalization to human mental processes by analyzing human cognition into a
sequence of simple computational steps that can be ordered in a program of an
'effective procedure' or algorithm.
The types of research that are now the center of activity are robotics and expert
systems. While the thrust of robotics to extend even further the elimination of
human labour from the sphere of production is familiar, expert systems and more
sophisticated forms of AI research on the computer simulation of mind represent
a crucial extension of this goal into the bureaucratic apparatus created by
scientific management during the second industrial revolution. It has been
claimed that "the number of working expert systems swelled from 700 at the end
of 1986 to 1,900 by the end of 1987. There were 7,000 systems under development
at the end of 1986 and 13,000 at the end of 1987. This has swelled to tens of
thousands of systems in 1990" (R. Kurzweil, 1990:301).5 Ironically, the principal
objects of this effort to rationalize and control mental work are the professional
arid clerical personnel, who were the instruments of scientific management's
efforts to mechanize and control the process of production.
A major focus of military-funded AI research has always been the develop-
ment of command and control programs capable of making decisions and
holding out the promise of the automated soldierless battlefield. This model has
proved increasingly congenial to private capital seeking to centralize managerial
control in the complex and dispersed structures of modern corporations. It is in
this context that expert systems, which replicate a linear, calculating rationality
focused on the solution of narrowly defined 'technical' problems, give man-
agement the opportunity to capture the professional knowledge and skills
110 B . J . Berman

of lower and middle management and technical personnel (Athanasiou, 1985:

15).
Implementing the computer-based rationalization of mental and manual work
is likly to run into increasingly serious resistance, particularly from professional
and managerial staff who possess extensive organizational skills and will bitterly
resent the technological proletarianization of their positions. It is here that the
ideological role of AI as tile core of a new technological paradigm becomes
crucial both to disarm and undermine resistance and to assure corporate and
government elites that what they are doing is 'objectively necessary' and requires
perseverance in the face of 'irrational' opposition. Indeed. the head of the
Japanese Fifth Generation AI project, Professor Tohru Moto-oka claimed that
with intelligent computers "totally new applied fields will be developed, social
productivity will be increased, and distortions in values will be eliminated"
(Weizenbaum, 1985:88 ]emphasis added]). Moreover, because it receives a
disproportionate share of media hype, AI has an increasing influence on the
broader ideological and cultural effects of computerization. AI can thus play the
role of a legitimating principle for capital in the hegemonic struggle of the current
crisis by 'naturalizing' the restructuring and automation of work in organizations,
making the new arrangements for living and working appear necessary and
inevitable and, thus, shaping common sense expectations about the nature of
work and the relationship between human and machine capabilities.
AI as an ideology has its effect primarily through a series of images and
metaphors that derive from its fundamental theoretical premises. The very core
of AI ideology is the computational or information-processing model of mind that
is expressed in the linked metaphors, 'the brain is like a computer' and 'the
computer is like a brain', which mechanize the brain and anthropomorphize the
computer. 6 Human thinking is equated with information processing and becomes
in principle reproducible in a program. The construction of an intelligent machine
or the understanding of human cognition thus become problems of uniting
engineering and psychology, decipherable through metaphors of program, code,
information, script, etc. (Pylyshyn, 1984). This has several crucial consequences.
First, the computational model of mind is a narrow and reductionist one,
defining intelligence solely in terms of a linear, calculating and essentially
mathematical rationality, i.e., the same as the conceptual basis for the design of
computers in the first place. Intelligence thus tends to be defined by AI in
machine rather than human terms. While such reductionism may be justifiable for
analytic purposes, the constant use and spread of the computer metaphors in the
wider society tends to reify the computational model of mind into common sense
knowledge. Furthermore, the reductionism of the model is reinforced by the
persistent failure of AI to deal with a broader range of human mental processes
such as creativity and imagination, or with natural language processing, which
deals with essentially ambiguous and context-specific meaning. This leads AI
researchers to "insist that what an effective procedure leaves out really doesn't
matter or can be fed in later as part of some yet-to-be-developed program"
(Roszak, 1986: 126).
Second, the computer metaphor's anthropomorphization of the machine tends
to exalt its abilities above that of humans, while the mechanization of the human
Artificial Intelligenceand the Ideologyof Capitalist Reconstruction 111

brain degrades it below the machine. In other words, the computer is a better
brain, while the brain is an inferior computer. An 'intelligent' computer can do
anything a human mind can do (as defined by the computational model) and do it
better, if not today then very soon. This denigration of human intelligence runs
through AI literature, especially that directed at the general public (Ritchie,
1984: 70, 93, 188).
The tendency of the AI information processing model of mind to denigrate
human intellectual abilities results in what Roszak terms a 'technological idolatry'
that reifies the computer metaphor, generating "a haunting sense of human
inadequacy and existential failure" and propagating a deference to computers
"which human beings have never assumed with respect to any other technology of
the past" (Roszak, 1986:44 45). This reveals the ideological importance of AI in
both legitimating and restructuring of capitalist society and generating a techno-
logical imperative requiring the installation and subordination of human labour
to 'intelligent' computers. The AI specialist is the contemporary avatar of an
engineering ideology represented 80 years ago by Taylorist 'efficiency experts'
and grounded in the antagonist social relations of capitalist production. They
share a similar distrust and hostility toward human labour that relegates "human
abilities to the status of system static, an unfortunate variability that plagues
efficient operation" (Shaiken, 1985: 269). AI represents an instrumental reason
that seeks to abolish all uncertainty and ambiguity, placing control and order over
all other values.
Lest we think that this connection of AI with the interests of capital is purely
adventitious, a reading of some of the most important figures in AI reveals a
considerable baggage of capitalist ideology. For Herbert Simon, the object of AI
is an intelligent machine that embodies the idealized rationality of the capitalist
entrepreneur "enshrined in modern economic theories, particularly those called
neo-classical" (Simon, 1982: 29). Marvin Minsky, meanwhile, "uses corporate
hierarchy as descriptive of the various levels of control and operation in human
thought." (J. Kurzweil, 1985: 9). The intelligence being simulated in a computer
is thus not that of an individual human mind, but rather the collective and
inherently social 'mind' of a bureaucratic organization seeking to devise an
explicit decision procedure or program based on the shared assumptions, rules,
and knowledge of trained expertise. Nevertheless, AI specialists seem unable to
distinguish individual cognition from social processes of 'knowing' and organi-
zational 'intelligence'. As Jack Kurzweil points out, "it is never qui.te clear
whether any fundamental distinction is being made between a human being, the
entrepreneurial abstraction of a human being, and a corporate structure. One
wonders, within this paradigm, how human beings thought before there were
corporations or whether corporations are the quintessential realization of the
core of the human spirit" (J. Kurzweil, 1985: 10). Moreover, the artificial
intelligentsia show little sense of the social basis of knowledge. In fact, knowledge
and understanding are conceived largely in asocial terms which neglect the degree
to which what is accepted as 'knowledge' is embedded in a context of socially-
shaped and shared measings that emerge out of a complex process of conflict and
consensus, i.e., the very same process out of which emerge the hegemonic ideas
of a society (Roszak, 1986: 132-33; Schank, 1984: 42-3).
112 B.J. Berman

The narrow and reductionist model of mind in AI is thus matched by an equally

impoverished notion of the social. The unspoken assumptions and unacknow-
ledged biases on which they are based reveal AI as a potential technological
paradigm that remains in the historical confines of capitalist ideology. That AI
researchers appear, in their naivete, to be largely unaware of this ideological
baggage (and largely incapable of conceiving the social consequences of AI
except in gushing descriptions of the great benefits it will bring to humanity)
makes the influence of what they do all the more insidious and demanding of
critical scrutiny.

Conclusion

AI encourages our acquiescence in the power of the scientific, economic and

political elites who control computer technology and claim a scientifically
legitimated right to decide its future course of development. The core ideas of AI,
expressed in the simple and immensely powerful metaphors of the human mind
and social organizations as 'information processing systems', are rapidly spread-
ing beyond the immediate context of writing 'intelligent' programs for computers
to serve as the premises for theorizing in academic disciplines from psychology to
molecular biology to management 'science'. Popularization in introductory
textbooks and in the often naive and over-heated coverage of the mass media is
helping to make AI metaphors a part of popular culture.
This essay has argued that AI constitutes an ideological discourse about power
focused on problems of order and control in complex social organizations.
Whether in the automated factory, on the automated battlefield, or in the
automated cabinet, AI represents the apotheosis of the two hundred year effort
to bring the management of the bureaucracies of the nation state and capital
under the infallible and predictable mastery of science (Berman, 1990). The
political agenda contained in AI, all the more seductive because so many AI
specialists seem genuinely unaware of its presence and blind to its implications,
reaches toward the depoliticization of decision-making in organizations in favour
of objective rational calculation. For its enthusiasts, AI will ultimately produce
intelligence machines that eliminate entirely the need for fallible and recalcitrant
human intellect, and "will worry about all the really important problems for us
(for us, not with us)" (Boden, 1985: 103). AI is part of a wider 'ideology of
technical control' embracing such disciplines as "cybernetics, information theory,
game theory, systems analysis, operations research and linear programming"
which extended "mathematical formalization into the realm of social problems
[and] brought with it a sense of new-found power, the hope of technical control of
social processes to equal that achieved in mechanical and electronic systems"
(Edwards, 1988: 252). In democratic societies, however, the artificial intelligent-
sia and their supporters in business and the state must be denied the mystification
of technological discourses and the magical view of scientific 'wizardry' which
protects the authoritarian and anti-human implications of their activities from
critical public scrutiny.
Artificial Intelligence and the Ideology of Capitalist Reconstruction 113

Acknowledgement

A n e a r l i e r v e r s i o n o f t h i s p a p e r w a s p r e s e n t e d at t h e W e s t C o a s t M a r x i s t S c h o l a r s
Conference, University of California, Berkeley, November 1 3 - 1 5 , 1987. M y
thanks to Paul Edwards and Les Levidow for their helpful comments and
suggestions.

Endnotes

1. Babbage's conception of the capitalist task division of labour (often known today as 'the Babbage
principle') was directly connected to his design for computing machinery on the principles of
hierarchy, sequence control and iteration. See Berman, 1989.
2. For studies of the role of the state and capital in the development of hegemony during the first
industrial revolution see Foster, 1974: chs. 6-7: and Corrigan and Sayre, 1985: chs. 6-7.
3. Some AI specialists now complain about the distoring effects of media exaggeration on their
work, but whenever they try to talk about the future effects of AI on society, they fall back into the
same tendency to inflated and self-serving claims that make it clear that they, and not the media, are
the root of the problem. See for example, Schank, 1984: 28, 34, 192-212,223-25.
4. Schank, one of the most theoretically austere of AI researchers, admits in his book that "I am not
unconcerned with business and money. In fact, I am president of a private company in the business of
selling AI programs." (1984: xi).
5. Whether these expert systems are fully functional, i.e., work as well or better than human
experts, and whether such systems can achieve such a level of functionality in the future, remain
matters of dispute. My own view is that they cannot do so, given the degree of tacit, intuitive non-
verbal knowledge and ideosyncratic subjective experience that forms the basis of human judgement in
the practice of professional skills. However, and this is the crucial point, from the perspective of the
wider objectives of capital, they may not have to work as well in the narrowly technical sense in order
to be acceptable replacements for human actors. There have been instances in the past in which
capital has resorted to technology which was actually inferior in operation to human labour in order to
achieve the goal of breaking the resistance and either bringing under control or eliminating particular
groups of skilled manual workers. (Winner, 1980: 124-25) Hubert and Stuart Dreyfus grasp this point
when they note that the real objective may be merely 'competent systems', operating at the level of
the expert novice where a somewhat inferior performance is organizationally acceptable. This
redefines the parameters of human and machine capabilities, limiting the vision of human managers
and professionals and undermining their faith in their own intuition and expertise. (1989: 126, 138).
6, There is an ironic paradox to be noted here. There is no way in which we can ground the argument
for the possibility of AI upon the rational analysis of existing evidence; quite the contrary, in fact,
given the persistent failure to achieve promised goals. Instead, the premises on which AI rest are what
Roszak calls 'master ideas', fundamental assumptions or principles based on no information
whatever. (1986: 91-92). The very idea of AI seems to be a conviction, an act of faith, whose very
existence and the passion with which it is pursued belies the computational, information-processing
model of mind on which it rests.

References

Athanasiou, T. (1985). Artificial Intelligence: Cleverly Disguised Politics. In: T. Solomonides and L.
Levidow (eds.) Compulsive Technology: Computers as Culture, London: Free Association Books.
Athanasious, T. (1989). Artificial Intelligence, Wishful Thinking and War. In: L. Levidow and K.
Robins (eds.) Cyborg Worlds: the Military Information Society, London: Free Association Books.
Baggage, C. (1832). On the Political Economy of Machinery and Manufactures, 3rd ed. enlarged,
London: Charles Knight.
Berman, B. (1990). Perfecting the Machine: Instrumental Rationality and the Bureaucratic
Ideologies of the State, World Futures, 28.
Berman, B. (1990). The Computer Metaphor: Bureaucratizing the Mind, Science as Culture, 7.
Boden, M. (1985). The Social Impact of Thinking Machines. In: T. Forester (ed.) The Information
Technology Revolution, Cambridge: MIT Press.
Braverman, H. (1974). Labor and Monopoly Capital, New York: Monthly Review Press.
114 B.J. Berman

Corrigan, P. and Sayre, D. (1985). The Great Arch: English State Formation as Cultural Revolution,
Oxford: Basil Blackwell.
Dreyfus, H. and S. (1989). Why Computers May Never Think Like People. In T. Forester (ed.)
Computers in the Human Context, Oxford: Basil Blackwell.
Edwards, P. (1988). The Closed World: Systems Discourse, Military Strategy and Post WWII
American Historical Consciousness, AI & Society, 2.
Flamm, K. (1987). Targeting the Computer: Government Support and hzternational Competition,
Washington: Brookings Institution.
Flamm, K. (1988). Creathzg the Corizputer: Government, Industry and High Technology, Washington:
Brookings Institution.
Fleck, J. (1982). Development and Establishment in Artificial Intelligence. In: N. Elias, et al. (eds.)
Scientific Establishments and Hierarchies: Sociology of Science, Vol. VI, Dordrecht: D. Riedel.
Foster, J. (1974). Class Struggle and the Industrial Revolution, London: Methuen.
Freeman, C. (1984). Prometheus Unfound, Futures, October.
Gilder, G. (1990). A Technology of Liberation, in R. Kurzweil. ed., The Age of Intelligent Machines,
Cambridge: MIT Press.
Kaplinsky, R. (1984). Automation: the Technology and the Society, London: Longman.
Knee, C. (1985). The Hidden Curriculum of the Computer, in T. Solomonides and L. Levidow, eds.,
Compulsive Technology: Computers as Culture, London: Free Association Books.
Kurzweil, J. (1985). Artificial Intelligence: an Ideology for the Information Society, Studies in
Communication and Information Technology, Working Paper #1, Kingston: Queen's University.
Kurzweil, R. (1990). The Age of Intelligent Machines, Cambridge: MIT Press.
Littler, C. (1982). The Development of the Labour Process in Capitalist Societies, London:
Heinemann.
Marx, K. (1977). (first ed. 1867), Capital, New York: Vintage Books.
Mosco, V. (1989). The Pay-Per Society: Computers and Communication in the Information Age,
Toronto: Garamond Press.
Pylyshyn, Z. (1984). Computation and Cognition: Toward a Foundation for Cognitive Science,
Cambridge: MIT Press.
Ritchie, D. (1984). The Binary Brain: Artificial Intelligence in the Age of Electronics, Boston: Little
Brown.
Roszak, T. (1986). The Cult of Information, New York: Pantheon.
Schank, R. and Childers P. (1984). The Cognitive Computer: On Language, Learning and Artificial
Intelligence, Reading: Addison-Wesley.
Schiller, H. (1986). Information and the Crisis Economy, New York: Oxford University Press.
Shaiken, H. (1985). Work Transformed: Automation and Labour in the Computer Age, Lexington:
Lexington Books.
Simon, H. (1982) The Sciences of the Artificial, Second Ed., Cambridge: MIT Press.
Turkle, S. (1984). The Second Self" Computers and the Human Spirit, London: Granada.
Weizenbaum, J. (1985). The Myths of Artificial Intelligence. In: T. Forester (ed.) The hlformation
Technology Revolution, Cambridge: MIT Press.
Winner, L. (1980). Do Artifacts Have Politics, Daedalus. 109.
Wood, S. (ed.) (1982). The Degradation of Work, London: Hutchinson.

Correspondence and offprint requests to: Bruce J. Berman, Queen's University, Kingston, Ontario,
K7L 3NG Canada. Fax: (613) 545 6848