BLOG Amazon Go: From walking cookies to reflective human beings

This student paper, in the form of a blog, by Elmira van den Broek is a great example of what our students can deliver for a course on technology & innovation. Elmira is a 1st year student in the 2-year research master ‘business in society’, and who selected the specialization course offered by Marleen Huysman. The research master can be considered as the first part of a PhD project within one of the subfields in business.

 

UNDERSTANDING AMAZON GO AS AN OUTCOME OF ON-GOING INTERACTION BETWEEN INDIVIDUALS, TECHNOLOGY, AND INSTITUTIONS

Amazon has just come up with a solution to the worst aspect of the grocery store experience: waiting in line to buy your items. Even in the exceptional cases of no lines, the checkout process is still time-consuming: you have to unpack your items, wait for all the items to get scanned, have to pay, sign a receipt, and bag up your items again. This takes a lot of time, a lot of waiting, and is mostly boring. Why not eliminate this whole scenario and replace this by a streamlined, easy and speedy setup?

This is what the “Just Walk Out” technology of Amazon offers us. The new concept convenience store – Amazon Go – promises no shopping lines, no check-out, and ultimately, no employee intervention at all. Customers will automatically register with their smartphones as they enter the store, can take items off the shelf, and leave the store when finished, in which their Amazon account is automatically charged. This extremely clever concept will eliminate the usual bottleneck of cashiers and registers that typically stand between shoppers and the store exit. The idea of self-checkout has been around for years, but the technology that powers Amazon Go differs in its reliance on similar technology used in self-driving cars: computer vision, sensor fusion, and deep learning (www.amazon.com, 2016). All possible surveillance methods will be used to profile customers’ needs in the most efficient way, including face recognition. By being able to combine business efficiency (reducing labor costs) with enormous advantages for consumers (time), Amazon has created an inevitable strong business model, that is expected to be highly successful.

The world’s leading e-commerce company has already introduced many things that the world now takes for granted, ranging from online buying books, to the emergence of customer reviews, and cloud computing. With its rapid innovations and aggressive competitive strategies, Amazon has expanded its reach into many aspects of everyday life and gained a substantial amount of market power. ‘How far can Amazon go?’, asked The Economist back in 2014 (www.economist.com, 2014). While this question was triggered by Amazon’s introduction of its first smartphone in 2014, it is still applicable, and perhaps even in a greater extent. As argued by the author of “Rise of the Robots: Technology and the Threat of a Jobless Future”, Martin Ford, Amazon Go will have disruptive consequences for the entire retail industry (CNBC Power Lunch, 2016). By providing an easier way to shop, Amazon Go does not only have implications for the future of convenience stores – Amazon has crushed local bookshops before –, but also for the future of work, what Amazon knows about us, and societal issues such as privacy and mass consumption.

 

Figure 1. Amazon Go technology. Source: http://www.theverge.com/.

This gives rise to important questions: How will Amazon Go shape the way people work and organize? And what are the implications of this new concept for customers? I will shed light on these issues by diverging from the view on technology as solely an objective, external force – which risks reducing humans too easily to cultural and social dupes – or as an outcome of human action – which risks assigning technology too little a role in making history – by drawing on Orlikowski’s Structurational Model of Technology (1992). This perspective will help to present a more complete image of the implications of Amazon Go, by viewing it as an outcome of on-going interaction between individuals, technology, and institutions.

THE STRUCTURATIONAL MODEL OF TECHNOLOGY

People can differ in their views on the impact of a new technology or innovative concept on organizations. The most dominant view considers technology as an objective, external force that determines organizational properties. Organizations cannot escape its influence and will be affected by technology in the same way. A second view emphasizes the human action aspect of technology. Technology is here seen as the outcome of dynamic, social interaction, in which similar technologies can result in a variety of outcomes (Leonardi & Barley, 2010). However, these views offer a rather incomplete image of how technology influences the structure of a workplace. For example, identical technologies can trigger similar dynamics and yet lead to divergent forms of organization (Barley, 1986). Orlikowski (1992) attempted to solve this issue by bringing both views together in her Structurational Model of Technology. She emphasized the duality of technology – technology is created and changed by human action, yet it is also used by humans to achieve action – and the interpretive flexibility – there is a flexibility in how people design, interpret, and use technology, but this depends on the different actors and socio-historical contexts (Orlikowski, 1992: 405). Thus, technology does influence organizational structure, but this influence depends on the specific institutions in which it is embedded. This third, and more recent, view on technology is presented in Figure 2.

 

Figure 2. The Structurational Model of Technology applied to Amazon Go, based on Orlikowski (1992).

As Figure 2 shows, human agents, technology, and institutional properties all have a share in the impact of a new technology. Individuals are linked to technology, due to their influence in designing, developing, appropriating, and modifying a certain technology (Arrow I), but also because technology facilitates and constraints their actions (Arrow II). Technology, in turn, influences the institutional properties of an organization, by for example changing organizational structure (Arrow III). These institutional properties also affect how people in turn design or use a technology (Arrow IV). In the case of Amazon Go, we can expect that the impact of the Just Walk Out Shopping technology is influenced by how people act on it: will customers value and use cashierless supermarkets? Will employees and management be able to sustain and coordinate these stores? And what is the impact of Amazon Go on the everyday lives of individuals and their institutions: how will it influence the norms, values and rules of society? To reflect on these various implications that arise due to interaction between technology and social action, I will draw on this Structurational Model of Technology of Orlikowski (1992). In specific, I will pay attention to the rise of Amazon Go as an outcome of human action and institutional properties (Arrow I and Arrow IV), Amazon Go as a facilitator and constrainer of individuals (Arrow II), and the impact of Amazon Go on institutions (Arrow III).

AMAZON GO AS AN OUTCOME OF HUMAN ACTIONS AND INSTITUTIONS

Individuals interact with technology in two ways: they are involved in the design of a technology, by building into technology certain interpretive schemes, facilities, and norms. They are also involved in the use of a technology, by assigning shared meanings to it (Orlikowski, 1992). How individuals design and use a technology is influenced by the institutional properties, such as shared values and norms. In the case of Amazon Go, the new concept convenience store was motivated by:

‘Four years ago we asked ourselves: what if we could create a shopping experience with no lines and no checkout? Could we push the boundaries of computer vision and machine learning to create a store where customers could simply take what they want and go? Our answer to those questions is Amazon Go and Just Walk Out Shopping.’ (www.amazon.com, 2016)

However, this answer remains rather unsatisfactory, since Amazon already has a shopping experience with no lines and no checkout, which is used by millions of people every day: Amazon.com. The developers of the concept of Amazon Go seem to be motivated by a type of customers that do not have the patience for one-click online shopping and waiting for delivery: customers that desire an online store that is offline. This taps into norms, values, and preferences on an institutional level: seeing, touching, and smelling products before buying them, combined with a high-level of service. Furthermore, it can be expected that developers of Amazon Go have been influenced by Amazon’s business strategy, which has authorized and legitimized the development of innovative, data-driven products, which span a web of technology around consumers. Located in an institutional climate of exponential, digital, and combinational technological progress (also referred to as “the Second Machine Age” by Brynjolfsson & McAfee, 2014), Amazon’s interpretive schemes and norms will be aligned with innovative competitors. Amazon Go is thus argued to be the product of human action of developers and customers, which has, in turn, been influenced by institutional properties.

AMAZON GO AS FACILITATOR AND CONSTRAINER OF HUMAN ACTION

Amazon Go facilitates and constrains individuals in various ways. In particular, the technology will have important implications for employees and customers.

Once developed and deployed, the Just Walk Out Shopping technology will become part of the everyday working routines of employees. Actually, not for a lot of employees. Amazon Go will make it possible to eliminate much of the staff needed to operate a store. This is expected to pose a serious threat to some of the 3.4 million Americans who work as cashiers (Bureau of Labor Statistics, 2016). Although precise details on the actual employee functions of Amazon Go stores are still unknown, it can be expected that some roles are still demanded: think of managers, technicians, but perhaps also concierges that can offer product information or a little human interaction to customers. These roles are expected to be facilitated and constrained in various ways. First, since Amazon Go is designed to map more-or-less rational, predictive shopping behavior of customers, such design decisions will discipline employees’ execution that work with these systems. As shown by Pine and Mazmanian (2014), tight coupling between formal representation systems and everyday work practices can be problematic for organizational effectiveness, by forcing employees to create a “perfect” but inaccurate account of work. For example, when prompt and quick action is required, employees might desire to technically deviate from the protocol, and rely on their own expertise and knowledge. However, this can result in time, burden, and coordination hurdles as this is not a “page-able” situation, and formal representation systems will conflict in case of deviations (Pine & Mazmanian, 2014). This can constrain flexibility of employees in work practices.

Second, since low-skilled functions (e.g., cashiers and shelf fillers) are expected to be automatized or replaced by virtual work, separation between people and objects will occur (Bailey et al., 2012). This will have various implications for work structure and the division of labour. For example, Barret et al. (2012) showed that a new technology can reconfigure boundary relations among occupational groups, with important consequences for skills, jurisdictions, status, and visibility. In the case of Amazon Go, a shift from low-skilled, monotonous tasks to knowledge management or algorithmic decision-making can be expected, related to an increase in status of the latter group. The separation between people and objects can have several unintended consequences. First, by making co-location and direct access to physical products unnecessary due to advanced virtual representations of customers and products, coordination problems among employees can arise (Sergeeva et al., 2015; Bailey et al., 2012). Employees may find it more difficult to adjust to other’s actions, or to interpret “errors” in virtual representation systems, such as shoplifting. Also, the introduction of sophisticated digital systems can result in the deskilling of professionals and loss of expertise and tacit knowledge (Sergeeva et al., 2015; Newell & Marabelli, 2015). By limiting learning through practice, for example stocking shelves or interacting with customers in a store, interdependencies between technology and human action are created. This can modify individual’s ability to learn new tasks, and to adapt to the work place or even society (Dall’Alba and Sandberg, 2010; Nicolini et al., 2003; in Newell & Marabelli, 2015). Thus, Amazon Go is expected to facilitate and constrain employees in various ways, ranging from eliminating functions (e.g., cashiers), to affecting work practices, coordination and deskilling.

Amazon’s expanding reach of products also facilitates and constraints the actions and behavior of individual consumers. So far, Amazon’s insatiable appetite has benefited the lives of many customers, and the introduction of Amazon Go is expected to tap into one of their major irritations: long waiting-lines. By bringing together shopping habits online as well as offline, Amazon is able to facilitate customers’ needs in a better and more efficient way. However, concerns can be drawn as Amazon grows in size and power, in which there is a slightly creepy feeling that Amazon knows too much about its users already (www.economist.com, 2014). By using the technology of Amazon Go, customers become ‘walking data generators’ (McAfee & Brynjolfsson, 2012), in which big data is generated during their everyday shopping routines. As argued by Marc Hijink in NRC, customers become transformed to “walking cookies” in an innovative logistic machine when touching a product (Hijink, 2016). By making it possible to track products that are grabbed in an impulse and returned to the shelves, even customers’ secret desires become exposed. This makes all anonymity of shopping becoming obsolete.

Amazon Go will transform you from a customer to a “walking cookie” in an innovative logistic machine (Hijink, 2016)

Algorithms play a key role in generating this customer data. They predict what a person will do, think and like based on the current or past behaviors of individuals (Newell & Marabelli, 2015). Algorithms offer a lot of strategic value for businesses and are used in various ways, such as the number of “friends” on Facebook being used to predict a person’s credit risk (www.google.com/patents/US8560436). However, the value for individuals and society is less clear. Although consumers can benefit from targeted information and personalized products and services, the predictions of these algorithms often represent a “black-box”. Users are not involved in deciding what to measure and produce, and even employees and management of corporations often have limited understanding on what is included in the algorithm and why (Newell & Marabelli, 2015). Customers should be aware of the trade-offs they make between exploring new opportunities of digitalization of everyday life, and the costs of giving up privacy, freedom, and independence (Newell & Marabelli, 2015). One the one hand, the Just Walk Out Shopping technology allows customers to shop more quickly than before, relieving them from the frustrating experience of waiting lines. On the other hand, the technology constraints the customers in that there is limited to no employee interaction possible, and that customers represent “walking data generators” that have to be aligned with “perfect” formal representations. This constraints the emotional, or human aspect of daily shopping routines.

THE IMPACT OF AMAZON GO ON WHAT WE TAKE-FOR-GRANTED

Once the Just Walk Out Shopping technology is taken-for-granted, it has become institutionalized, and its use by employees and customers will influence institutional structure. This will have implications for convenience stores and society in general.

Due to its dominant position as the world’s leading e-commerce company, Amazon is expected not only to set new organizational practices and structures for its own stores, but also to transform these among other organizations. This will change the structure of signification, because the knowledge and assumptions embedded in the technology deployed directs the manner in which problems are interpreted and work is conducted, structure of domination, by distributing more resources to functions which involve technical and knowledge management tasks, and structure of legitimization, by sanctioning manual, time-consuming systems and propagating algorithmic, virtual representations (Orlikowski, 1992). This will set the trend of making the visible world increasingly invisible, by relying on data-driven, algorithmic predictions (Orlikowski and Scott, 2014), and eliminating direct employee intervention.

There is nothing innocent about making the visible invisible (Orlikowski & Scott, 2014)

Moreover, the institutionalization of Amazon Go has the potential to change practices and routines of how customers prefer to shop. By providing a low-barrier, easy way to do groceries, this can alter norms and values on consumption. For example, by lowering barriers of shopping and distancing individuals from manual actions such as checking out at the counter, this facilitates easier and perhaps even more consumption. Increased consumption has negative environmental impact – by increasing pressure on natural resources (Wiedmann et al., 2015) – and increases risks of people spending beyond their means (Evans & Schmalensee, 2005). Moreover, institutionalization of data-driven, online-offline combined algorithmic shopping experiences can contribute to normalization of using sensitive customer data. However, as argued by Orlikowski (1992), individuals do not always use technology as it was intended. For example, customers chose to adopt text messaging in the United Kingdom, while this was developed as an engineer-communication tool (Ansari & Phillips, 2011). In the case of Amazon Go, customers can undermine or transform the embedded rules and resources, by for example enforcing rules regarding data-protection or prizing employee interaction in stores. The possible tensions and instabilities that arise at an institutional level have the potential to change and transform in return the actions of individuals, such as modifying the Just Walk Out Shopping technology, or deciding where to consume.

CONCLUSION

The impact of Amazon Go is the outcome of the technology it deploys, as well as the individuals that design and use the concept, and the institutional environment. This makes it possible to see that technology is not solely an objective reality: it is a product of its time and organizational context, that reflects the interaction with various individuals such as consumers and employees, and the knowledge, interests, and conditions of society. By discussing the emerge of the concept of Amazon Go from this perspective, we are able to look at the duality and flexibility of technology. It helps to see Amazon Go as more than an online experience surrounded by brick walls: it is an on-going design that is physically and socially constructed by individuals, and that in turn influences human action and institutional properties. This allows us to escape the deterministic view on Amazon as a powerful data-giant that spans an ever-growing web of technology around our lives, but also makes us aware of the limits and opportunities of human action and the trade-offs that we make as society. By creating awareness on this duality between technology and social action, we are perhaps able to transform ourselves back from “walking cookies” to “reflective human beings”.

 

REFERENCES 

Amazon. (n.d.). Frequently Asked Questions. Retrieved from https://www.amazon.com/b?ie=UTF8&node=16008589011.

Ansari, Shahzad, and Nelson Phillips. “Text me! New consumer practices and change in organizational fields.” Organization Science 22, no. 6 (2011): 1579-1599.

Bailey, Diane E., Paul M. Leonardi, and Stephen R. Barley. “The lure of the virtual.” Organization Science 23, no. 5 (2012): 1485-1504.

Barrett, Michael, Eivor Oborn, Wanda J. Orlikowski, and JoAnne Yates. “Reconfiguring boundary relations: Robotic innovations in pharmacy work.” Organization Science 23, no. 5 (2012): 1448-1466.

Barley, Stephen R. “Technology as an occasion for structuring: Evidence from observations of CT scanners and the social order of radiology departments.” Administrative science quarterly (1986): 78-108.

Brown, John Seely, and Paul Duguid. “Knowledge and organization: A social-practice perspective.” Organization science 12, no. 2 (2001): 198-213.

Brynjolfsson, E., & McAfee, A. (2014). The second machine age: Work, progress, and prosperity in a time of brilliant technologies. WW Norton & Company (chapter 2: “The skills of the new machines: technology races ahead”)

Bureau Of Labor Statistics (2016, March 30). Occuppational Employment and Wages – May 2015. Retrieved from https://www.bls.gov/news.release/pdf/ocwage.pdf.

CNBC Power Lunch. (2016, December 6). Technology creates and destroys jobs: Martin Ford [Video file]. Retrieved from http://video.cnbc.com/gallery/?video=3000573952.

Evans, D. S., & Schmalensee, R. (2005). Paying with plastic: the digital revolution in buying and borrowing. Mit Press.

Garud, Raghu, Sanjay Jain, and Philipp Tuertscher. “Incomplete by design and designing for incompleteness.” Organization studies 29, no. 3 (2008): 351-371.

Hijink, M. (2016, December 12). De supermarkt zonder kassa. Retrieved from https://www.nrc.nl/nieuws/2016/12/12/de-supermarkt-zonder-kassa-5750238-a1536248.

Leonardi, Paul M., and Stephen R. Barley. “What’s under construction here? Social action, materiality, and power in constructivist studies of technology and organizing.” The Academy of Management Annals 4, no. 1 (2010): 1-51.

Newell, S., & Marabelli, M. (2015). Strategic opportunities (and challenges) of algorithmic decision-making: A call for action on the long-term societal effects of ‘datification’. The Journal of Strategic Information Systems, 24(1), 3-14

Orlikowski, W. J., & Scott, S. V. (2013). What happens when evaluation goes online? Exploring apparatuses of valuation in the travel sector. Organization Science, 25(3), 868-891

Orlikowski, Wanda J. “The duality of technology: Rethinking the concept of technology in organizations.” Organization science 3, no. 3 (1992): 398-427.

Pine, K. H., & Mazmanian, M. (2014, February). Institutional logics of the EMR and the problem of perfect but inaccurate accounts. In Proceedings of the 17th ACM conference on Computer supported cooperative work & social computing (pp. 283-294). ACM.

Sergeeva A., Huysman M., Faraj S. (2015) Transforming work practices of operating room teams: the case of the Da Vinci robot. Accepted for publication in ICIS 2015 Proceedings.

The Economist. (2014, June 21). How far can Amazon go? Retrieved from http://www.economist.com/news/leaders/21604550-it-has-upended-industries-and-changed-way-world-shops-it-should-beware-abusing.

Yoo, Y., Boland Jr, R. J., Lyytinen, K., & Majchrzak, A. (2012). Organizing for innovation in the digitized world. Organization Science, 23(5), 1398-1408.