Читать книгу Making Sense of AI - Anthony Elliott - Страница 14

1 Sceptics show some recognition that AI is sweeping through industries, enterprises and public life, but AI is not viewed as revolutionary. On the contrary, ‘no significant change’ is the motto.

2 For many authors of a sceptical persuasion, AI as a transformative power is recast as little more than marketing hype or a myth.

3 Rather than a transformed world economy powered by AI, sceptics advance a business-as-usual model comprising technological advances on the one hand, and adaptation by the labour force on the other hand.

4 There is an emphasis upon workplace change as involving the twin forces of people and machines, employees and technology.

5 It is implicitly acknowledged that AI poses a risk to some jobs (mostly routine, unskilled work, according to sceptics), but in general the position advanced is that AI will create more jobs than it destroys.

6 There may be some spillover from AI breakthroughs that impact society, culture and everyday life – especially the globalizing forces of communication. Nonetheless, AI is primarily a technological process which principally impacts the economy in limited and partial ways.

7 For many of a sceptical persuasion, traditional economic power is paramount and the actions of national societies are important too. Accordingly, the globalizing dimensions of AI are treated as contingent on these economic and national state factors.

Central to this transformationalist perspective is an emphasis on the social relations impacted by AI. That is to say, the technologies associated with AI are understood to reshape not only institutions and organizations but also identities and intimacies. Another way of making this point is to say that the AI revolution is as much about entertainment as it is about the economy, as much about meaning and morality as it is about money and manufacturing. For lifestyle change is likely to be of key importance in the spheres of both professional and personal life when assessing the impacts of AI, or so argue transformationalists. As Erik Brynjolfsson and Andrew McAfee write of these massive changes in The Second Machine Age: ‘Computers started diagnosing diseases, listening and speaking to us, and writing high-quality prose, while robots started scurrying around warehouses and driving cars with minimal or no guidance.’⁷ Brynjolfsson and McAfee capture well the idea that digital transformation is not only about the economy, industry and corporate life, but crucially also about sociality, everyday life and power. The advance of AI is, in a word, generative. The digital revolution creates different kinds of work and different sorts of skills and gives rise to different ways of living from those of even the very recent past.

Transformationalists question the idea that economy and society can be adequately grasped from the business-as-usual perspective advanced by sceptics. For the extensive penetration of the global economy by digital forces has fundamentally altered its operations and dynamics. Transformationalists generally underscore the essential significance of the digital revolution, an historic moment in the worldwide transformation of manufacture and services. This involves locating contemporary patterns of globalization within the new technological revolution, and a dazzling variety of terms has been coined to capture these momentous shifts – including ‘Industry 4.0’, ‘digital capitalism’, ‘algorithmic governmentality’, ‘bot economy’ and ‘automated society’. Three aspects of change tend to be emphasized in the transformationalist literature: the radical transformation of manufacture and services, of consumption and citizenship, and of public policy. In transforming both the conditions and consequences of economy and society, according to this argument, AI, robotics and other forms of automation have revolutionized corporate life and businesses across the world. Advances in machine learning algorithms and big data in particular have underpinned extraordinary innovations in the manufacturing of goods and services as well as the emergence of new industries, and consequently jobs and employment have come under assault as never before. The impact of smart software, and of social media more generally, has significantly transformed the consumer economy itself. At the same time, these unparalleled technological innovations directly impact upon issues of ethics and governance. Recognizing how closely the impact of AI on jobs and public policy are intertwined, governments worldwide have sought to introduce a raft of measures geared towards enabling robust engagement with the digital revolution.

If you accept the argument that AI involves the transformation of manufacturing and services between and across the world’s advanced economies and societies, then it follows logically that there will also be a wholesale shift at the base of the job skills pyramid, with very broad employment implications as well as the prospects of massive unemployment. As AI reorganizes the global economy, so transformationalists argue, blue- and white-collar jobs alike increasingly evaporate. The transformationalist story of what this will do to jobs and the future of employment is, however, multilayered and complex. For some transformationalists, the economic consequences of increasing automation are clear: the proportion of the labour force in manufacturing will decline sharply in all the industrialized countries. Martin Ford, in The Rise of the Robots, equates AI and automated technology directly with the threat of jobless futures. From telepresence robots to the digital offshoring of high-skill jobs, Ford sees a relentless AI-driven technology trend towards rising unemployment and greater inequality.⁸ Seeking to shift the debate beyond the conventional solution that increased education and training will facilitate better adaptation by workers into new, higher-skill roles, Ford argues the case for a new economic paradigm, one based on a guaranteed income or living wage that incentivizes risk-taking and entrepreneurship. Similarly, Richard Baldwin’s The Globotics Upheaval views the disruptive impacts of digital technology as wall-to-wall, resulting in an unparalleled displacement of jobs worldwide. In Baldwin’s telling of the transformationalist narrative, however, these negative impacts will be mostly short-lived, opening the way for a more optimistic prognosis of automated technology in the long run. As Baldwin comments:

I view AI . . . as a good thing once we can get through the transition. People’s jobs will be more interesting because all the robotic repetitive stuff will be done by machines. Things that can be done remotely will be done remotely and that will allow us to do things where we actually have to be together. So, ultimately, I think it will be a very, very good thing.⁹

An excessive zeal also applies to other aspects of the transformationalist position, especially as regards the creation of new jobs. Transformationalists contend that automated production destroys jobs within industrial manufacturing. But, within this literature, there is also the argument that AI is creating new jobs elsewhere in the economy. An extraordinary range of knock-on services and jobs, especially roles performed as ‘digital work’, has been unleashed by the rise of AI – which, in turn, has led to the emergence of new industries, businesses and even occupations. As Paul R. Daugherty and H. James Wilson argue in Human + Machine: Reimagining Work in the Age of AI: ‘In the current era of business process improvement AI systems are not replacing us; they are amplifying our skills and collaborating with us to achieve performance gains that have previously not been possible.’¹⁰ In today’s circumstances, argue some transformationalists, the future of jobs increasingly depends on AI–human collaboration. In this view, the deployment of human–machine hybrid teams dramatically improves productivity and thereby increases prosperity. Other transformationalists highlight that AI and machine learning algorithms (based on big data) underpin the scaling up of many companies across multiple industries today. Such developments drive new customer acquisition, underpin employee retention rates and help create new job opportunities.

Let us turn now to contrast two transformationalist interventions which centre upon the problem of work and employment. The first is Klaus Schwab’s The Fourth Industrial Revolution, issued by the World Economic Forum (of which Schwab is executive chairman). The second is Bernard Stiegler’s Automatic Society, volume 1 of which is subtitled The Future of Work. There is a telling feature about the writing of Klaus Schwab that several critics have noted, and which pertains to the underlying ardour of his transformationalist stance. Schwab makes it abundantly clear that the AI transformation in manufacturing and services is already well under way. The digital revolution, he contends, is producing ‘exponential disruptive change’, and this can be discerned in the prevalence of advanced robotics, machine learning, big data and supercomputers in business and organizational life today. The scope and scale of the digital revolution for Schwab – what he terms the ‘fourth industrial revolution’ – are ‘unlike anything humankind has experienced before’.¹¹ Yet if Schwab’s transformationalism is clearly evident in this diagnosis of our times, his critique of the consequences of AI appears (at least on an initial reading) as scrupulously non-judgemental. Employment is a signal example. Schwab contends that AI ushers in massive efficiency gains and cost reductions for businesses and industry, but also highlights the massive automation of jobs stemming from these very developments. On the one hand, he emphasizes that technological innovation today destroys jobs as never before, whilst on the other hand he underscores that AI unleashes a new era of prosperity through the creation of novel employment opportunities and future industries. He argues that AI disrupts labour markets and workplaces around the world, and yet emphasizes the ability of workers in the new economy to adapt continuously and fashion new skills through lifelong learning.

In other words, Schwab’s approach seeks to capture both the stunning opportunities and threatening risks stemming from AI. Pressed to an extreme, however, his analytic approach is never free from a certain degree of ambivalence, as every social change associated with the digital revolution appears mediated through this both/and logic. This might be said to be the conceptual equivalent of wanting to have your cake and eat it too. Towards the latter sections of The Fourth Industrial Revolution, Schwab’s analytic reserve – where his lack of a conclusion on the consequences of AI becomes a conclusion all of its own – gives way to a more robust transformationalist sensibility. As he concludes:

The digital mindset, capable of institutionalizing cross-functional collaboration, flattening hierarchies, and building environments that encourage a generation of new ideas, is profoundly dependent on emotional intelligence . . . The world is fast changing, hyper-connected, even more complex and becoming more fragmented but we can still shape our future in a way that benefits all. The window of opportunity for doing so is now.¹²

In the end, AI for Schwab is an exhilaratingly progressive affair. He argues that AI has the potential to be institutionalized as a global, cosmopolitan form of life, one to be celebrated rather than castigated.

In contrast to this business-school approach to understanding AI, radical French theory informs Bernard Stiegler’s Automatic Society. Like Schwab, Stiegler holds that the AI revolution is already upon us. AI for Stiegler inaugurates a new social order of ‘total autonomization’, in which production and manufacturing are controlled by software and big data. But unlike Schwab with his stab at analytic even-handedness, Stiegler is out to develop a more full-blooded critique of the destructive aspects of AI for economy and society. He writes, for example, of today’s ‘immense transformation’ whereby ‘capitalism becomes purely computational’, of ‘generalized autonomization and autonomisms’, and of ‘algorithmic governmentality’. Taking his cue from the post-structuralist analysis of ‘control societies’ developed by Gilles Deleuze, Stiegler seeks to lay bare the short-circuiting of minds and spirits – the ‘shock and stupefaction’ inflicted on contemporary women and men – arising from full automatization. Drawing upon quantum physics, Stiegler argues that automatized societies are increasingly locked in a contradictory relationship between entropy (where life-energy dissipates) and negative entropy (the reversal, or undoing, of such decomposition). ‘Automation’, writes Stiegler, ‘has given rise to an immense amount of entropy, on such a scale that today, throughout the entire world, humanity fundamentally doubts its future – and in young people especially so.’¹³ Google Translate, as Stiegler remarks, is a good example of the immense linguistic entropy occurring throughout the world today, as split-second machine translation of the world’s diverse languages into English results in a radical impoverishment of vocabulary. Google’s algorithms simply flatten both the individual and collective use of language. What is at stake, as Stiegler shrewdly points out, is human knowledge in the broadest sense; knowing how to think, reflect, talk, communicate and act in the world.

If for Stiegler Google Translate represents destructive linguistic entropy, the algorithmic automation of society signals massive economic entropy. AI makes it possible not just to economize upon labour, but to fully automate tasks and thus render employees redundant. This is a redundancy of the worker’s expertise, as advanced automation for Stiegler produces a generalized (economic as well as environmental) ‘disorder of hyper-standardization’ – where work, and the value of employees, are determined by calculating probabilities based upon averages. Today’s industrial capitalism, writes Stiegler, is ‘an era in which calculation prevails over every other criteria of decision-making, and where algorithmic and mechanical becoming is concretized and materialized as logical automation and automatism . . . as computational society becomes a society that is automated and remotely controlled’.¹⁴ We are at the beginning of a process of technological transformation that will have a massive impact upon the nature of work, expertise and knowledge – the algorithmic governmentality of 24/7 capitalism, according to Stiegler, will precipitate ‘entropic catastrophe’.

The new technological landscape, however, results not only in doom and gloom. Stiegler also seeks to discern a hidden trend in economic entropy for reversing the devastating impacts of algorithmic capitalism. Emancipation for Stiegler is linked to the primacy of meaningful work, which he sharply differentiates from employment. In this perspective, work is fundamentally meaningful and creative, whereas the bureaucratized terrain of employment is increasingly automated and dependent upon computational software. His argument, broadly speaking, is that the production and transformation of automation prepare the way, paradoxically, for the ‘dis-automatization of society’. In a striking irony, the kind of employment which is bound up with automated entropy also consists in de-automating routines, which can liberate most of the population from exploitative domination. If employment is increasingly the terrain of advanced automation, complex algorithms and computational software on the one hand, work produces value and the creation of something new to society on the other hand. From this angle, Stiegler emphasizes that work consists of practical know-how (savoir-faire), formal knowledge (savoirs formels) and life skills (savoir-vivre). The ‘data economy’ is therefore not the inevitable destiny of automated society; a range of other possible systems can be envisaged. This scenario, Stiegler proposes, is already practicable. We have reached a stage, in algorithmic capitalism, in which the automated forces of production are overdeveloped and new economic models based on the social economy and cultural solidarity – especially through associations, cooperatives and public services, as well as new industries – will create novel, intermittent forms of work and new professions. A non-repressive automated society, Stiegler argues, would become an ‘economy of contribution’.

How is it possible that there should be such significant differences in assessment between two authors associated with the transformationalist position? To begin with, Stiegler’s writings serve as an apt counterbalance to Schwab’s emphases, particularly the former’s penetrating analysis of the very large decline in jobs worldwide resulting from advanced automation. Schwab’s work is explicitly concentrated on how organizations create value, and he repeatedly emphasizes that technological transformation today creates new opportunities and dilemmas – the results of which can lead to positive, shared benefits for all of society. Stiegler on the other hand clearly does intend his analysis to have a very broad application: not just economics and the market, but society and the politics of life itself. Whilst some have dismissed Stiegler’s work as excessively influenced by the jargon of radical French theory, his critique of the production of automation in contemporary social life, as AI displaces labour, remains highly significant. In demonstrating that advanced automation produces an entropic violence of hyper-standardization, Stiegler’s critique arguably confronts head-on the most painfully destructive and debilitating aspects of algorithmic capitalism. We can also see that fundamental lines of difference are to be found among voices advocating the transformationalist position. This is an important point. Contrasting the contributions of Schwab and Stiegler highlights that the transformationalist position is not cut of one cloth.