Editing IPCC:AR6/WGIII/Chapter-5 (section)

==== 5.3.4.1 Digitalisation ====

<div id="h3-7-siblings" class="h3-siblings"></div>

In the context of service provision, there are numerous opportunities for consumers to buy, subscribe to, adopt, access, install or use digital goods and services ( [[#Wilson--2020b|Wilson et al. 2020b]] ). Digitalisation has opened up new possibilities across all domains of consumer activity, from travel and retail to domestic living and energy use. Digital platforms allow surplus resources to be identified, offered, shared, transacted and exchanged ( [[#Frenken--2017|Frenken 2017]] ). Real-time information flows on consumers’ preferences and needs mean service provision can be personalised, differentiated, automated, and optimised ( [[#TWI2050--2019|TWI2050 2019]] ). Rapid innovation cycles and software upgrades drive continual improvements in performance and responsiveness to consumer behaviour. These characteristics of digitalisation enable new business models and services that affect both service demand, from shared ride-hailing ( [[#ITF--2017a|ITF 2017a]] ) to smart heating ( [[#IEA--2017a|IEA 2017a]] ), and how services are provisioned, from online farmers’ markets ( [[#Richards--2018|Richards and Hamilton 2018]] ) to peer-to-peer electricity trading to enable distributed power systems ( [[#Morstyn--2018|Morstyn et al. 2018]] ).

In many cases, digitalisation provides a ‘radical functionality’ that enables users to do or accomplish something that they could not do before ( [[#Nagy--2016|Nagy et al. 2016]] ). Indeed the consumer appeal of digital innovations varies widely, from choice, convenience, flexibility and control to relational and social benefits ( [[#Pettifor--2020|Pettifor and Wilson 2020]] ). Reviewing over 30 digital goods and services for mobility, food buying and domestic living, [[#Wilson--2020b|Wilson et al. (2020b)]] also found shared elements of appeal across multiple innovations including (i) making use of surplus, (ii) using not owning, (iii) being part of wider networks, and (iv) exerting greater control over service provisioning systems. Digitalisation thus creates a strong value proposition for certain consumer niches. Concurrent diffusion of many digital innovations amplifies their disruptive potential ( [[#Schuelke-Leech--2018|Schuelke-Leech 2018]] ; [[#Wilson--2019b|Wilson et al. 2019b]] ). Besides basic mobile telephone service for communication, digital innovations have been primarily geared to population groups with high purchasing power, and too little to the needs of poor and vulnerable people.

The long-term sustainability implications of digitalised services hinge on four factors: (i) the direct energy demands of connected devices and the digital infrastructures (i.e., data centres and communication networks) that provide necessary computing, storage, and communication services ( [[IPCC:Wg3:Chapter:Chapter-9#9.4|Section 9.4]] .6); (ii) the systems-level energy and resource efficiencies that may be gained through the provision of digital services ( [[#Wilson--2020b|Wilson et al. 2020b]] ); (iii) the resource, material, and waste management requirements of the billions of ICT devices that comprise the world’s digital systems ( [[#Belkhir--2018|Belkhir and Elmeligi 2018]] ; [[#Malmodin--2018|Malmodin and Lundén 2018]] ) and (iv) the magnitude of potential rebound effects or induced energy demands that might unleash unintended and unsustainable demand growth, such as autonomous vehicles inducing more frequent and longer journeys due to reduced travel costs ( [[#Wadud--2016|Wadud et al. 2016]] ). Estimating digitalisation’s direct energy demand has historically been hampered by lack of consistent global data on IT device stocks, their power consumption characteristics, and usage patterns, for both consumer devices and the data centres and communication networks behind them. As a result, quantitative estimates vary widely, with literature values suggesting that consumer devices, data centres, and data networks account for anywhere from 6% to 12% of global electricity use ( [[#Gelenbe--2015|Gelenbe and Caseau 2015]] ; [[#Cook--2017|Cook et al. 2017]] ; [[#Malmodin--2018|Malmodin and Lundén 2018]] ). For example, within the literature on data centres, top-down models that project energy use on the basis of increasing demand for internet services tend to predict rapid global energy use growth, ( [[#Andrae--2015|Andrae and Edler 2015]] ; [[#Belkhir--2018|Belkhir and Elmeligi 2018]] ; [[#Liu--2020a|Liu et al. 2020a]] ), whereas bottom-up models that consider data centre technology stocks and their energy efficiency trends tend to predict slower but still positive growth ( [[#Shehabi--2018|Shehabi et al. 2018]] ; [[#Hintemann--2019|Hintemann and Hinterholzer 2019]] ; [[#Malmodin--2020|Malmodin 2020]] ; [[#Masanet--2020|Masanet et al. 2020]] ). Yet there is growing concern that remaining energy efficiency improvements might be outpaced by rising demand for digital services, particularly as data-intensive technologies such as artificial intelligence, smart and connected energy systems, distributed manufacturing systems, and autonomous vehicles promise to increase demand for data services even further in the future ( [[#TWI2050--2019|TWI2050 2019]] ; [[#Masanet--2020|Masanet et al. 2020]] ; [[#Strubell--2020|Strubell et al. 2020]] ). Rapid digitalisation is also contributing to an expanding e-waste problem, estimated to be the fastest growing domestic waste stream globally (Forti et al. 2020).

As digitalisation proliferates, an important policy objective is therefore to invest in data collection and monitoring systems and energy demand models of digitalised systems to guide technology and policy investment decisions for addressing potential direct energy demand growth ( [[#IEA--2017a|IEA 2017a]] ) and potentially concomitant growth in e-waste.

However, the net systems-level energy and resource efficiencies gained through the provision of digital services could play an important role in dealing with climate change and other environmental challenges ( [[#Masanet--2010|Masanet and Matthews 2010]] ; [[#Melville--2010|Melville 2010]] ; [[#Elliot--2011|Elliot 2011]] ; [[#Watson--2012|Watson et al. 2012]] ; [[#Gholami--2013|Gholami et al. 2013]] ; [[#Añón%20Higón--2017|Añón Higón et al. 2017]] ). As shown in Figure 5.12, assessments of numerous digital service opportunities for mobility, nutrition, shelter, and education and entertainment suggest that net emissions benefits can be delivered at the systems level, although these effects are highly context dependent. Importantly, evidence of potential negative outcomes due to rebound effects, induced demand, or life-cycle trade-offs can also be observed. For example, telework has been shown to reduce emissions where long and/or energy-intensive commutes are avoided, but can lead to net emissions increases in cases where greater non-work vehicle use occurs or only short, low-emissions commutes (e.g., via public transit) are avoided ( [[#Hook--2020|Hook et al. 2020]] ; [[#IEA--2020a|IEA 2020a]] ; [[#Viana%20Cerqueira--2020|Viana Cerqueira et al. 2020]] ). Similarly, substitution of physical media by digital alternatives may lead to emissions increases where greater consumption is fuelled, whereas a shift to 3D printed structures may require more emissions-intensive concrete formulations or result in reduced thermal energy efficiency, leading to life-cycle emissions increases ( [[#Mahadevan--2020|Mahadevan et al. 2020]] ; [[#Yao--2020|Yao et al. 2020]] ).

Furthermore, digitalisation, automation and artificial intelligence, as general-purpose technologies, may lead to a plethora of new products and applications that are likely to be efficient on their own but that may also lead to undesirable changes or absolute increases in demand for products (Figure 5.12). For example, last-mile delivery in logistics is both expensive and cumbersome. Battery-powered drones enable a delivery of goods at similar lifecycle emissions to delivery vans ( [[#Stolaroff--2018|Stolaroff et al. 2018]] ). At the same time, drone delivery is cheaper in terms of time (immediate delivery) and monetary costs (automation saves the highest-cost component: personnel) ( [[#Sudbury--2016|Sudbury and Hutchinson 2016]] ). As a result, demand for package delivery may increase rapidly. Similarly, automated vehicles reduce the costs of time, parking, and personnel, and therefore may dramatically increase vehicle mileage ( [[#Wadud--2016|Wadud et al. 2016]] ; [[#Cohen--2019|Cohen and Cavoli 2019]] ). On-demand electric scooters offer mobility access preferable to passenger cars, but can replace trips otherwise taken on public transit ( [[#de%20Bortoli--2020|de Bortoli and Christoforou 2020]] ) and can come with significant additional energy requirements for night-time system rebalancing ( [[#Hollingsworth--2019|Hollingsworth et al. 2019]] ; ITF 2020). The energy requirements of cryptocurrencies is also a growing concern, although considerable uncertainty exists surrounding the energy use of their underlying blockchain infrastructure ( [[#Vranken--2017|Vranken 2017]] ; [[#de%20Vries--2018|de Vries 2018]] ; [[#Stoll--2019|Stoll et al. 2019]] ). For example, while it is clear that the energy requirements of global Bitcoin mining have grown significantly since 2017, recent literature indicates a wide range of estimates for 2020 (47 TWh to 125 TWh) due to data gaps and differences in modelling approaches ( [[#Lei--2021|Lei et al. 2021]] ). Initial estimates of the computational intensity of artificial intelligence algorithms suggest that energy requirements may be enormous without concerted effort to improve efficiencies, especially on the computational side ( [[#Strubell--2020|Strubell et al. 2020]] ). Efficiency gains enabled by digitalisation, in terms of reduced GHG emissions or energy use per service unit, may be overcompensated by activity/scale effects.

[[File:f91c2c3949403e1b0e7bebcaae4116d6 IPCC_AR6_WGIII_Figure_5_12.png]]

'''Figure 5.12''' | '''Studies assessing net changes in CO''' 2 '''emissions, energy use, and activity levels indicate mitigation potentials for numerous end-user-oriented digitalisation solutions, but also risk of increased emissions due to inefficient substitutions, induced demand, and rebound effects.''' 90 studies were assessed with 207 observations (indicated by vertical bars) including those based on empirical research, attributional and consequential lifecycle assessments, and techno-economic analyses and scenarios at different scales, which are not directly comparable but are useful for indicating the directionality and determinants of net emissions, energy, and activity effects. Sources: [[#Erdmann--2010|Erdmann and Hilty (2010)]] ; [[#Gebler--2014|Gebler et al. (2014)]] ; [[#Huang--2016|Huang et al. (2016)]] ; [[#Verhoef--2018|Verhoef et al. (2018)]] ; [[#Alhumayani--2020|Alhumayani et al. (2020)]] ; [[#Court--2020|Court and Sorrell (2020)]] ; [[#Hook--2020|Hook et al. (2020)]] ; [[#IEA--2020a|IEA (2020a)]] ; [[#Saade--2020|Saade et al. (2020)]] ; [[#Torres-Carrillo--2020|Torres-Carrillo et al. (2020)]] ; [[#Wilson--2020c|Wilson et al. (2020c)]] ; [[#Yao--2020|Yao et al. (2020)]] ; [[#Muñoz--2021|Muñoz et al. (2021)]] .

Maximising the mitigation potential of digitalisation trends involves diligent monitoring and proactive management of both direct and indirect demand effects, to ensure that a proper balance is maintained. Direct energy demand can be managed through continued investments in, and incentives for, energy-efficient data centres, networks, and end-use devices ( [[#Masanet--2011|Masanet et al. 2011]] ; [[#Avgerinou--2017|Avgerinou et al. 2017]] ; [[#IEA--2017a|IEA 2017a]] ; [[#Koronen--2020|Koronen et al. 2020]] ). Shifts to low-carbon power are a particularly important strategy being undertaken by data centre and network operators ( [[#Cook--2014|Cook et al. 2014]] ; [[#Huang--2020|Huang et al. 2020]] ), which might be adopted across the digital device spectrum as a proactive mitigation strategy where data demands outpace hardware efficiency gains, which may be approaching limits in the near future ( [[#Koomey--2011|Koomey et al. 2011]] ). Most recently, data centres are being investigated as a potential resource for demand response and load balancing in renewable power grids ( [[#Koronen--2020|Koronen et al. 2020]] ; [[#Zheng--2020|Zheng et al. 2020]] ), while a large bandwidth for improving software efficiency has been suggested for overcoming slowing hardware efficiency gains ( [[#Leiserson--2020|Leiserson et al. 2020]] ). Ensuring efficiency benefits of digital services while avoiding potential rebound effects and demand surges will require early and proactive public policies to avoid excess energy use ( [[#TWI2050--2019|TWI2050 2019]] ; [[#WBGU--2019|WBGU 2019]] ), which will also necessitate investments in data collection and monitoring systems to ensure that net mitigation benefits are realised and that unintended consequences can be identified early and properly managed ( [[#IEA--2017a|IEA 2017a]] ).

Within a small but growing body of literature on the net effects of digitalisation, there is ''medium evidence'' that digitalised consumer services can reduce overall emissions, energy use, and activity levels, with ''medium agreement'' on the scale of potential savings, with the important caveat that induced demand and rebound effects must be managed carefully to avoid negative outcomes.

<div id="5.3.4.2" class="h3-container"></div>

<span id="the-sharing-economy"></span>