Aspire Thought Leadership! Wondering about impact of AI in business?. Find out more on how fortune 500 companies use this approach. Come right in for
The literature that studies impact of AI in business and economic outcomes more generally starts by arguing that AI (namely machine learning, robotics and associated technologies) is an example of a General Purpose Technology (or GPT). The concept goes back to GPT as a pervasive technology, attracting complementary innovations and susceptible to further improvements.
They provided many examples of GPTs that have significantly impacted economies such as the steam engine, electricity and computers. GPTs tend to increase the productivity of businesses across most industries and spur several complementary innovations that enhance the positive effect of the GPTs on productivity by creating new supply chains or allowing businesses to exploit economies of scale.
 |
| Impact of AI in Business |
AI (and in particular machine learning as the underlying methodology) appears to fit the description as pointed out that machine learning, in general, is particularly suited for automating tasks where prediction matters and they can be applied to most tasks currently performed by humans. Besides, AI systems embedded with machine learning can learn and improve over time. [Machine Learning introduction]
Interestingly, the improvement process can be led by an algorithm itself rather than by a software engineer. For instance, machine learning algorithms can identify the best functions linking inputs and outputs and can do so even without supervision. Also, machines can share knowledge and learn from each other: once a machine acquires skill in one location, it can be replicated across digital networks thanks to cloud computing availability. Finally, machine learning systems can spur a variety of complementary innovations. Indeed, unsupervised machine learning can help engineers develop a broader set of additional applications that can enhance existing supervised machine learning algorithms’ capabilities.
The Impact of AI in Business
Most of the economic analysis on Artificial Intelligence has focused on the impact that robotics (or automation in general) may have on existing jobs and through this route on economic growth. The impact of automation on tasks, productivity and work and focusing on economic growth and automation. Automation is assumed to be exogenous, and the incentives for introducing AI are related to cost reduction.
Automation and Labour Market
Automation will displace workers in tasks. Interestingly, it is not only routine tasks that are replaced in this model but also tasks that require high skills. Factories manufacturing microchips and circuit boards are using AI-equipped with high-resolution cameras that outperform the human eye. As a result, labour demand will decrease (despite the increasing productivity of labour), and employment in equilibrium will be much lower, everything else being equal. The increase in productivity of the existing workforce will expand the economy, which will increase the demand for jobs whose tasks cannot be automated. Additionally, increasing automation triggers an increase in capital investment which will generate labour demand in industries like robotics and engineering. Finally, automation will generate new jobs which will support robots and their maintenance (for instance). Importantly, even if there are factors that may slow down the displacement effect of AI on labour, it is essential to bear in mind that the adjustment will be costly as workers will be searching for new jobs and will need to retrain. However, the speed of adjustment is endogenous as it will be determined by firm-level decisions about the adoption of ArI and by workers decisions about education and training.
What is the Impact of Automation on Economic Growth?
the positive impact of automation on growth may be constrained by the fact that sectors with relatively slow productivity growth may experience increases in their size. This phenomenon has been compared to Baumol’s “cost disease” which refers to the fact that sectors with slow productivity growth may increase in size even if they do not grow faster than the other industries that experience fast productivity growth. When Baumol’s observation is applied to models where fast productivity growth is triggered by automation, sectors with slow productivity growth slow down economic growth. As a result, the labour share remains substantial even if the extent of automation is pervasive.
There are no models that focus on other types of AI (machine learning is the key example here) and their impact on the activities of the businesses. In reality, most of the understanding of the impact of AI on economic outcomes is driven by the analysis of automation and its impact on productivity growth and through that route on economic growth. While these results are perfectly reasonable in the context of what we expect the impact of automation on economic outcomes to be, they do not account for the fact that artificial intelligence is not only automation and that businesses use AI in ways that are well beyond the cost-saving paradigm that underpins most of the research on the economic impact of AI. Indeed, businesses may use AI to improve consumer experience or deliver services that are closer to what the consumers want. The impact of these alternative uses of AI on productivity and ultimately economic growth is not well understood; however, to be able to do so, it is crucial to understand how AI shapes business models and how incumbents continue to create value with AI. In other words, the focus on the analysis has to move to business models even if the concept itself does not have any grounding in economic theory (that mostly focuses on pricing and value capture). Still, studying business models makes sense in economic models where innovation can be disruptive and therefore, incumbents will have to identify new mechanisms to survive by developing new value propositions.
- Value Migration and Business Models
As suggested previously, a full assessment of AI’s impact on economic outcomes requires an understanding of how organisations embed AI into their business model.
New entrants adopt radically different business models that rely on innovative uses of AI, and therefore the mechanisms used by incumbents to create value may no longer be fit for purpose. There are many examples of this: Uber and Airbnb are well-known entrants that use a different business model than incumbents thanks to AI and eventually, end up dominating their respective industry segments. How can incumbents react? Several authors have pointed out that business model innovation is the primary mechanism that incumbents can use to offset value migration’s negative implications. Our main claim here in the context of AI-driven value migration is that business model innovation is not limited to introducing new products or new processes. Therefore, the purpose of this subsection is to identify different patterns of business model innovation which enable businesses (and organisations in general) to rebuild their business logic. We will first define the concept of the business model and its components; we will then introduce the concept of industry-level value migration. Finally, we will provide a framework to analyse business model innovation and taxonomy that summarises the main types of business model innovation we can observe in real life.
Impact of AI in Business: Defining Business Models
Every organisation has a business model that describes how value is created, delivered and finally captured by the organisation. A business model addresses the critical question: “how does one build a competitive advantage and generate a profit?” Business models can be considered a concise representation of an organisation’s internal thinking of creating value for its stakeholders. Business models allow firms to identify the connections between choices and performance and ensure that internal decisions are consistent among each other. These can be related to the value chain structure or the value proposition offered to customers.
Driving factors behind the interest in business models include the growth of the Internet (as an alternative distribution channel) e-commerce, and more generally the fact that value capture mechanisms are different from those used in manufacturing industries. Still, there is no consensus on what a business model is although both practitioners and academics agree that a good business model has to generate value while enabling value capture.
A business model represents how an organisation creates, delivers and captures value in conjunction with other partners. In their view, without a well-developed business model, businesses will fail to capture value from their activities. Indeed, developing a successful business model is insufficient to assure competitive advantage as an imitation of business models is of ten straightforward. A business model can create a competitive advantage if it is hard to replicate.
Most papers on business models emphasise that business models are made of different components that may explain how businesses do business. Again, there are several descriptions of these components.
- A customer value proposition,
- A profit formula,
- Key resources and
- Key processes that allow consistently delivering the value proposition.
The literature suggests that the same business can use multiple business models, and they differ along with different distribution costs, arrangements to source inputs and satisfying customer needs.
Value Migration and Business Model Innovation
Initial work on business models had assumed a static relationship among the many elements that made up a business model and did not consider the impact of the environment on such relationships. At the same time, organisational level decision making is acknowledged to be influenced by the business environment and its changes. Therefore business models cannot be immune from this process as in real life, business models change continuously because of technological innovation, changing regulation, changing industrial structure and so on. As pointed out by several authors, to understand how changes in the environment alter the relationship among the different business model elements, it is essential to adopt a different perspective that focuses on the drivers of business model innovation.
Research on business model innovation has tried to articulate the relationship between industry-wide phenomena (like the emergence of new technology) and business models and how the resulting strategic choices available to firms unfold. Therefore, literature has developed the concept of value migration that is the change of the sources that create profits. From the con- sumers’ standpoint, the value can migrate from established incumbents to new entrants with better products. While some firms absorb value from other firms due to changes in their business models, others will lose value to other firms because of business models that have become less competitive or outdated. The mobile phone industry summarises very well these different phenomena. Nokia has been a long- standing incumbent in the mobile phone industry and did not realise that new entrants such as Apple and Samsung would become central players. With time, value creation moved away from the hardware, and as a result, Nokia’s business model would not be able to create value as its vital mechanism for value creation was negatively affected by the new entrants.
Conceptually, business model innovation is the construct that has been developed to explore the relationship between value migration and business model. Business models tend to change over time for many reasons, and therefore business model innovation is defined as the change in operations and value creation that leads to an improvement in business performance. Theoretically, there is a lack of clarity about what business model innovation is. Indeed, some authors suggest that business model innovation is creating a new business model while other authors point out that small changes in the business model qualify as business model innovation. Business model innovation refers both to new business models developed by new companies and to changes in the existing business models. Another element of ambiguity in the literature refers to the scope of business model innovation, although there is an agreement that at least one dimension has to change radically before the change can be qualified as innovation.
Business model innovation follows value migration: changes in business models might be needed when there are structural changes in industries; in these cases, firms need to think how they create and capture value by identifying the best business model that allows keeping creating value when this is migrating between firms. A company that has successfully changed the business model following technology disruption is Netflix. The company started as a rental company for DVDs which became quickly successful in its line of business. However, as streaming services started to emerge, the company successfully jumped into this segment of the video industry. Two critical elements of the business model had to change: the products offered by the company and the pricing structure.
Business Model Innovation and AI
The diffusion of AI in an industry tends to trigger value migration across different segments of the industry: it does so by facilitating the emergence of new businesses that use AI not only to reduce the costs associated to production (through automation, for instance) but to change the mechanisms through which value is created and delivered. Still, the role of AI in driving business model innovation has been explored only recently. The focus is mostly on new entrants (like Uber and Airbnb) and their business model.
However, the effect of AI on business model innovation goes far beyond that. Besides enabling business model innovation by facilitating the entry of new competitors, AI technologies can also change the relationship between the different components of incumbents’ business models. We argue that to understand business model innovation following the adoption of AI, we need to explore the mechanisms that connect different components of the business model and how these change due to the deployment of AI. Three main dimensions to business models have been identified: value creation, value delivery and value capture.
Value creation and delivery focus on customers and how they are engaged while value capture refers to how value is monetised. Therefore, to discuss business models, we need to focus on these dimensions, and so we explore what it means for business model innovation and its relationship to technical innovation.
Business Models Innovation and AI: A Taxonomy
The current literature on business model innovation focuses on external antecedents linked to changes in the business environment. So far, very few papers have explored the role of technology, and it is only recently that the introduction of new technologies has been studied as a factor driving business model innovation. Work on classifying business model innovation and technology has proceeded along two lines. On the one hand, some researchers suggest that technology can be incorporated into a business and have a positive impact on performance. On the other hand, some other authors see the concept of a business model as separable from technology. However, there is no such empirical work in the case of business model innovation driven by AI. Therefore, they will provide a taxonomy of the new business models that have emerged because of AI-based on the qualitative literature on AI and business model innovation.
There are many more examples of businesses that use AI to reduce costs or boost their revenues. Can we identify some patterns around the different types of business model innovations that employ AI? To study the impact of AI on business model innovation, we use the framework: they describe for each type of impact the effects of AI on value creation, value delivery and value capture. This way, they identify four ways AI can change the elements of the business models. They start from an incremental change going to a radical transformation of the business model, which implies a change of all the business model elements. Importantly these four models are not mutually exclusive, and businesses can employ them simultaneously. Besides, business model innovation will be studied in the context of incumbent firms that prefer to reconfigure the elements that constitute their business model because of AI’s introduction. In general, they also want to distinguish incremental business model innovation (where businesses tend to keep the same customer base while expanding their activities in such a way that new customers are attracted) from radical business model innovation (where businesses leave behind the core of their activities and move to new markets).
 |
| How does AI affect business |
Processes and Automation
In this case, AI is mostly used by businesses to improve internal processes without changing the whole business architecture. The main goal of the business is to increase efficiency, and it does so by automating processes. Relationships with suppliers, for instance, can be streamlined through the use of bots. Overall, the emphasis of this type of business model innovation is on automating existing processes to capitalise on the company’s existing knowledge and resources. The automotive industry is the best example: robots have been used for a while in the assembly lines of companies such as Jaguar-Land Rover and BMW. In this case, value creation is generated by better connections among machines, increasing efficiency of the workforce and transparent management.
Efficient use of internal resources is the crucial mechanism for both value capture and value delivery to customers. Finally, value capture is driven by more efficient processes and use of resources.
Improving Customer Interface
AI is deployed to understand customers and their needs. Virtual reality and bots are the key technologies here, and they follow the initial investment in automation. Value delivery is facilitated by segmentation (based on data science workflow analysis) and the resulting development of long-term relationships with customers. Besides, digital distribution channels can improve consumer sales. At the same time, new services can be created as consumers needs can be easily identified, thanks to the extensive use of data collected through bots. Finally, new services (like dynamic pricing or pay-per-use) generate new revenue streams that allow businesses to capture value. Examples include supermarkets (Tesco, Sainsbury’s) and large retailers (Boots) which use data collected through loyalty cards to segment customers and personalise offers based on their characteristics and preferences. Other examples include Spotify and KFC, which use AI to improve their relationships with customers.
Joining Ecosystems
AI facilitates the creation of virtual marketplaces that allow businesses to create new value networks. In these examples, AI allows integrating knowledge and resources drawn from many organisations and businesses into networks that allow delivering new services or new products to the consumers or other businesses. In this business model, value creation is generated by using real-time information about production, sales and availability of new services. Value delivery is guaranteed to all the businesses that belong to the network by delivering the new services that are intrinsically linked to the platform. Finally, value capture is guaranteed by the revenue streams generated by the new services. The virtual marketplace, which Amazon enables, is the prime example of such an ecosystem; however, other examples include the Appstore or Google Playstore.
Developing Smart Products
AI allows them to develop and commercialise different goods and services, allowing firms to diversify or expand their markets. The emphasis is on the development of AI-powered products which are the critical mechanism for value creation. Importantly, customers are part of the value creation process, and there is a direct relationship between the business and the customers, thanks to AI. Value delivery is generated by the smart products and the innovation in the associated services while value capture is generated by the new revenue streams associated with the new products. All products for smart homes or cars use AI and tend to interact with customers to deliver services tailored to their needs.
General Considerations
There are some expected benefits associated with most AI technolo- gies, namely real-time capability, interoperability and the potential of integrating production systems:
- First, they tend to lower the production costs, which may result in low prices that may be eventually passed on to the consumers. This is an inherent feature of AI that generally tends to replace labour and, therefore, reduce a large share of businesses’ variable costs.
- Second, AI allows businesses to generate new products and services; while it is true that some of these can be “intelligent” variations of existing products, importantly some of these products are innovative in the sense that they did not exist before the advent of AI.
- Third, AI can change the nature of the interaction between con- sumers and businesses. In the past, consumers’ role was that of passive purchasers of goods that could express their preferences only by walking away from specific products.
As business models change with the advent of AI, consumers tend to drive the production of services, and in some case, they are co-producers of services and products. Their preferences get known to businesses at an early stage and businesses can get a detailed picture of consumers’ preferences; this can lead to the development or creation of new products and services that better address the consumers’ needs. In this respect, embedding AI in a business model provides opportunities to create new value delivery mechanisms that better address customer needs. AI allows creating delivery systems that distinguish between two types of customers: those that provide data that allows personalising the services that are eventually offered to the second group. In other words, the notion of customers as being just at the end of the value chain and willing to pay for the products has changed.
Simultaneously, while personalisation of services and products is highlighted as one of the strengths of the AI-powered business models, in real life, we can observe businesses that offer “one-size-fits-all” value delivery proposals along with personalised offers. In these cases, the AI business model has to deliver value to the former group so that the latter can have better services. Implicitly this requires the development of two parallel mechanisms for value delivery. In practice, the two mechanisms can support each other and connections between the two mechanisms matter as they eventually define the mechanisms through which value is captured. Typically, value capture is rationalised in terms of pricing and price discrimination. However, the critical feature of business models that embed AI is that pricing is not the only mechanism to capture value, but it is one of the many that businesses can use. For instance, the use of consumers’ data to personalise offers can be alternative mechanism companies can use to capture value.
AI-driven business models tend to be responsive to user-driven design and, as a result, business models tend to be better aligned to customer value creation. However, businesses need to develop the technical capability to acquire data about their clients and start to think in terms of ecosystems rather than value chains [what is big data?]. Additionally, AI is pushing companies to a change from product to service mindset. This outcome is not new in itself. Products are delivered as a service as the digital part of a hybrid solution. The result is the so-called product- service system concept that shows how development and offering of specific product-service bundles are sold to customers as a solution rather than a product. This practise is widespread in the automotive industry where smart products are sold together with smart services, and as a result, suppliers, customers, and other partners become part of a networked ecosystem.
AI allows the horizontal and vertical integration of the value chain, allowing expanding the firms’ boundaries. In this context, new actors arise. Openness can create communities with similar interests and therefore monetise innovation through other routes. Openness can let users indicate what they want and allow them to be engaged with the business, leading to value creation for both sides. For instance, firms contributing to online communities share their adaptations with vendors, which will improve the next release.
One key question is whether business models that embed AI at their core can be easily imitated. It is an important question because if they are not easily replicated, they can generate a higher return to AI investment until they are. In reality, this is not always the case, and businesses have to find mechanisms that allow them to capture value even when imitators enter the industry.
Choosing a New Business Model
The value migration framework only suggests that firms must pursue business model innovation to remain competitive. Of course, it is still unclear under what conditions business model innovation takes place. Business model change is an experimental search process where elements of the existing business model change in a progressive way. From the perspective of established businesses, changing an existing business model can be problematic, and the existing characteristics of the business model tend to limit the choices available to firms. As a result, the mechanisms for value creation, delivery and capture may change as businesses have norms, behaviours and organisational structures they must modify as they adopt new business models.
Changes in the existing business model require exploration and experimentation. Business model innovation is essentially a reinforcement learning process which has to consider how the environment has changed. While experimenting with new business models is a routine process, it is particularly vital in AI as this is an emerging technology that is poorly understood. Indeed, experimentation allows businesses to acquire knowledge about how the new technology works and allows to manage risks by helping managers to select the best business model given the capabilities of the technology and the constraints posed by the environment. For instance, a company may want to use AI to offer customers a set of personalised services based on subscription. In this case, the company must understand whether the AI tool employed for personalisation is “fit for purpose” and whether the customers are open to such an innovation. To do so, the company may need to experiment with the AI tool and test how it is useful in creating and capturing value (i.e., whether it is accurate in identifying what customers may want and whether customers are happy with such innovation). Importantly, this type of experimentation may allow businesses to know about the technology’s technical characteristics and how the workforce, consumers, and suppliers will accept a particular business model where AI plays a key role.
Experimentation should have a long term view and consider the fact that several initiatives need to be tested and combined so that they can jointly produce value. For example, to allow AI tools to segment customers efficiently, a company might need to set up several sales and marketing initiatives and check which ones can deliver value fast.
Some researchers have advocated running multiple business models when pursuing new opportunities.5 For instance, Universities may have moved into the online delivery model (secondary business model) simultaneously as the traditional face-to-face delivery model (primary business model). Some others have pointed out that by running paral- lel business models, the firm may not see complementarities between them and is often the leading cause for failure. Alternatively, businesses can change the primary business model in line with the second business model. In this case, the firm can transform the primary business model elements in line with changes in the external environment.
Changing a business model requires businesses to acquire knowledge about new technologies and changes in the environment. A lot is known about how businesses go about acquiring new knowledge. Researchers have pointed out that knowledge can be sourced externally, and that competitive advantage can be created by combining external knowledge with internal knowledge. There are two types of knowledge acquisition activities: exploration and exploitation. Exploratory knowledge acquisition aims at developing new competencies, while exploitative knowledge acquisition wants to expand existing knowledge.8 The two activities can complement each other, and indeed, ambidexterity has been pro- posed to combine the two strategies for knowledge acquisition. External and internal knowledge can be combined and can be acquired through alliances or collaboration with other organisations. Collaboration can be a source of knowledge which is quite different from the business’ current knowledge base. Once embedded into the firm’s knowledge base, it can eventually lead to a change in the business routines and the business model. In practice, we do not have results on how businesses acquire AI-related knowledge although we can assume that an ambidextrous strategy can help firms capture external knowledge that allows deploying AI efficiently.
Reim et al. (2020) have highlighted the changes that need to occur when trying to innovate their business model. The authors use the concept of digital transformation to clarify that these changes are not simple IT initiatives but rather a re-design of the business activities to align AI strategies to the overall business strategy. In their view, this type of changes can lead to new business scope and customers. They also highlight the importance of organisational aspects when changing a business model. Pilot projects can help identify organisational bottlenecks and the additional actions needed to ensure employees support and accept the AI tools. The importance of intermediaries between data scientists and business managers has been highlighted by Reim et al. (2020): both suggest that technologists are needed to demystify what AI does and give managers assurance they are still in control. In some sense, the emphasis on intermediaries’ role in this field highlights the fact that the current provision of skills among general management cannot support business model innovations driven by technology. There is an extensive debate on skill shortages in components of data science, but there is not much discussion on the managerial skills shortages in this area [What is data science]. This topic can be a primary area to study in the future as AI becomes more ubiquitous than it is now.
COMMENTS