Failure to act causes cyber-scientific management that optimizes every second of every day for maximum profit

Relating outcomes to competences, which are then optimised through machine learning, and the rise of ‘use sovereignty’ are ways in which learning is displaced from the individual worker and increasingly from work-based learning in terms of collectives of workers. Paradoxically, learning seems to be occurring everywhere in the contemporary factory but it is not happening at a human level as epistemologically CBET and machine learning are distinct from what we would understand by learning. This displacement of learning might occur not in terms of a single computational or organisational ‘mind’ co-ordinating activities but rather as a system of AIs and machine intelligences that collectively seek to optimise profit and cost functions determined by the employer who ‘wireheads’ these into the organisation. In terms of human outcomes, AIs work to build up a ‘grammar’ of outcomes, behaviours and combinations. This is the new ‘scientific management’ where humans are ‘chopped up’ into outcome entities, either individually or collectively, foreshadowed by Marx in Grundrisse: The worker's activity, reduced to a mere abstraction of activity, is determined and regulated on all sides by the movement of the machinery, and not the opposite. The science which compels the inanimate limbs of the machinery, by their construction, to act purposefully, as an automaton, does not exist in the worker's consciousness, but rather acts upon him through the machine as an alien power, as the power of the machine itself. The appropriation of living labour by objectified labour – of the power or activity which creates value by value existing for-itself – which lies in the concept of capital, is posited, in production resting on machinery, as the character of the production process itself, including its material elements and its material motion. The production process has ceased to be a labour process in the sense of a process dominated by labour as its governing unity. Labour appears, rather, merely as a conscious organ, scattered among the individual living workers at numerous points of the mechanical system; subsumed under the total process of the machinery itself, as itself only a link of the system, whose unity exists not in the living workers, but rather in the living (active) machinery, which confronts his individual, insignificant doings as a mighty organism. In machinery, objectified labour confronts living labour within the labour process itself as the power which rules it; a power which, as the appropriation of living labour, is the form of capital. (Marx, 1993, p. 693, my italics) What is relevant in the above quote is not just how machinery colonises labour (the ‘acting upon’ of machinery) but also how labour unity both between and within the individual labourer is ‘scattered’ among the individual workers. As the British philosopher Nick Land describes the process of AI takeover in industry: [Page 647]Modern production seems like a dream of cyborg colonization work, a dream that makes the nightmare of Taylorism seem idyllic. Industrial machines dismantle the actuality of the proletariat, displacing it in the direction of cyborg hybridization, and re the plasalizingticity of labour power. The corresponding extraction of tradable value from the body sophisticates at the interface, dissociating exertion into increasingly intricate functional sequences, from pedals, levers and vocal commands through the synchronization of production-line tasks and time-motion programs, to sensory-motor transduction within increasingly complex and self-micromanaged artificial environments, capturing minutely adaptive behavior for capital. Autocybernating market control guides the labour-process into immersion. (Land, 2018, p. 435, my italics) Here, Land takes further Marx's 19th-century considerations of what is happening to labour in the process of industrial production. Labour in all its forms (mental and physical) becomes ‘dismantled’ into discrete components through AI realising the myriad ways in which it can be combined (its plasticity) where human behaviour is captured. In each of these quotes, the idea is that labour is consistently pushed to the limits of human embodiment as it becomes captured by machinery into a ‘grammar’ of human labour far surpassing the ideas of scientific management. In these scenarios learning at work does not exist within human entities but in ‘chopped up’ human assemblages that would be integrated with machines and data in many different forms. The learning takes places within various AIs in terms of a ‘skill bank’ whereby various entities would be charged with ‘producing’ types of behaviour. As stated above, this would be an impoverished conception of human learning where learning would become reconceptualised as pure outcome. In the field of consumption, AI is already building up data on our behavioural patterns through platform capitalist tools such as Facebook, Twitter and Google. Reactions and behaviours are quantified and used to build up a profile not only of our individual behaviour but of a model of human behaviour. Existent disciplines such as Psychology and Sociology in universities appear to be rather insignificant compared to how accurately social media giants can predict (and modify) attitudes and behaviours based on the predictions of ‘Big Data'. In the workplace, quantification also gathers and uses data on how workers behave and respond. As considered above, in many ways this is an extension of scientific management whereby the movements of workers were laboriously logged and tested in a series of scientific experiments. Photography was used to map and chart each human movement as part of a production process so that each factory worker was subject to optimisation and eventually computer technology developed more complex techniques of process and production management which were used to record and simulate worker behaviour. As technology improves, modes of quantification have become increasingly extensive (Moore et al., 2017). High performance digital cameras and sensors are used to record worker movements at sub-perceptual levels. Smartwatches can record worker biometrics and allow instant feedback on worker attitudes. Worker interactions and vocalisations can be measured by using Alexa-type devices in the workplace. Employee sentiment can be measured by conducting text analysis of their social media posts. Such devices can also be used to measure worker resistance (including union organisation) and the worker's capacity for disruptive activity. These are all forms of worker surveillance. Through these means, labour power production is becoming even more extensive to reflect not only a wider variety of physical and mental attitudes and behaviours, but also worker attitudes and emotions. Compliance and satisfaction can be measured not only through a periodic worker satisfaction survey but constantly and [Page 648]consistently during every moment of an employee's day. This is not just a process of quantifying each individual worker, it also involves building up a model of human behaviour in general. Zuboff (2019) argues that ‘Surveillance Capitalism’ is not just a method of increasing productivity in certain circumstances but also a method for producing a residue, or surplus, from these surveillance activities in which workers are dispossessed of skills. Zuboff refers to this activity as ‘behavioural surplus': Surveillance capitalism begins with the discovery of behavioral surplus. More behavioral data are rendered than required for service improvements. This surplus feeds machine intelligence – the new means of production – that fabricates predictions of user behavior. These products are sold to business customers in new behavioral futures markets. The Behavioral Value Reinvestment Cycle is subordinated to this new logic. (Zuboff, 2019, p. 97) Zuboff primarily considers the realm of consumption in her analysis of Surveillance Capitalism, but it is also possible to apply this work to the behavioural surplus acquired in the workplace. Surveillance produces streams of data which can be combined with other data pools and resources to produce (through machine learning) insights into aspects of human labour that could not be gained by a single observation alone. Owners of this data (who might not be the employer who uses the technology) can create new ways of understanding the mechanisms of worker productivity and the ‘nudges’ by which the labour power of workers can be enhanced. These can be combined with AI knowledge of other entities in the production process to specify minute interactions between those entities to optimise production. This is what Zuboff (2019, p. 97) refers to as the ‘Behavioural Reinvestment Cycle', whereby behaviour is used to build platforms that are reintegrated into the production process to further optimise worker behaviour. One existing example of this is the use of so-called ‘fatigue functions’ to optimise human performance in production. A ‘fatigue function’ is a mathematical equation derived from empirical observation which shows when a human worker is reaching the point of ‘failure’ in terms of being unable to carry on with their task successfully. At present fatigue functions are usually determined by using existing data on worker fatigue so that a production process can be optimised. Studies which have tested worker fatigue in real manufacturing environments under experimentally controlled conditions can be used to specify how workers should behave in reality. With the advent of consistently gathered biometric and behavioural data on workers in real time, fatigue can be measured consistently, and the worker is consistently experimented on by AI that can ‘nudge’ workers towards better performance. The accumulation of intelligence through surveillance on each individual worker is reinvested constantly into a superior system. It must be noted that the idea of an intelligence that can catalogue and build a grammar of individual skills and capacities to be used in the workplace is not a new one. The early years of the Soviet Union produced a number of experiments in creating a catalogue of human skills and abilities, with their origins in the work of Alexi Gastev. Gastev sought to record and catalogue all manifestations of physical labour (Ings, 2016, p. 79) which could be combined in a ‘performance'. Through this manual cataloguing performances could be choreographed, and workers would learn through rehearsal to prepare for factory work. In the 1980s there were similar attempts in terms of CBET to create a universal database of competences that could be used for any occupation. Plans were made for a vast lexicon of computerised records which could be used to select exactly the right set of competences for a particular work task (Preston, 2017, p. 34). These early experiments foreshadow the contemporary situation where individual behaviours are under surveillance and disaggregated, but the essential difference is that learning is transferred from the worker to the AI, and that the real learning that takes place in the workplace [Page 649]is occurring at the level of machines rather than at the level of humans. This evolution and confluence of workplace learning with AI and machine learning is increasingly making the prospect of ‘cloud-based’ services for skills more likely. Rather than individuals possessing knowledge of discrete skills, the organisation uses ‘Behavioural Value Reinvestment’ to create a consistently evolving learning platform regarding worker behaviour. However, there is no reason why this evolving knowledge of skills and abilities should remain at the level of the individual organisation. As is the case with other services which can be accessed by a digital organisation, there are plans for ‘skills clouds’ whereby employers of various types could ‘draw down’ the particular skills components that they require in their workplace. For example, Workday, a NASDAQ-listed company with a market capitalisation of $40 billion, has produced a ‘skills cloud’ system that uses machine learning to categorise skills between organisations in order to form a common ‘skills language': Cristina Goldt, vice president, HCM products, Workday commented: ‘Nearly every organization wants to develop and reskill their workforce to grow their people and their business, but they lack a fundamental understanding of the skills they have and the skills they need. Our skills cloud tackles this issue head on with machine learning algorithms that bring calm to the chaotic language of skills. This will ultimately help customers connect skills to people in a more meaningful way to improve how they get work done, develop existing workers, and allocate talent to better meet evolving business needs. (Source: www.enterprisetimes.co.uk/2018/10/03/workday-unveil-skills-cloud/) The capacity for ‘skills clouds’ to grow is not only premised on technology but also on the growth of new forms of precarious employment. Recruitment agencies, who often act to ‘sift’ employees to large employers make use of skills tests, using AI and machine learning, to filter workers into job positions and in doing so gain wider behavioural knowledge of worker skills and abilities. Similarly, the growth of ‘platform capitalism’ in terms of employers who use mobile applications to track and reward workers (such as Uber or Deliveroo) gain constant data on employee behaviour and skills. This leads to a number of separate ‘skills clouds', with the possibility of amalgamation at a higher level. The implication of this for learning at work is clear in that learning is disappearing not just from workers, but potentially also from employers as it is integrated into the ‘skills cloud’ by multinational companies.