Which industry sectors might be affected?
The potential of AI is driving rapid take-up of artificial intelligence across a range of sectors. What are some of the likely use cases across sectors? Here we outline some sector-specific AI use cases that are already deployed, and some which may be adopted in the future. (Not all the currently operating applications we describe are entirely autonomous or involve machine learning yet, but their use reflects a trend for future adoption of autonomous AI within the relevant sector.)
Energy
AI will impact upon the Energy sector in a number of different ways. Most importantly, it will be able to facilitate ‘optimisation’, leading to enhanced efficiencies, cost reductions, and the ability of energy infrastructure to predict with greater accuracy changes in supply and demand. AI can also help the industry address safety issues – there will be no need to send out maintenance teams in the middle of winter to repair offshore wind farms if the work can be done by AI.
The combination of AI and electric vehicles is likely to result in disruption at a scale we have not yet seen. We are now looking at so-called ‘brontobyte’ data volumes as the possibility that every vehicle creates large quantities of data becomes a reality. If the Transport sector is electrified this would have a profound impact upon the oil and gas sector (and stranded assets would also become a reality).
Anne Lapierre, Global Co-Head of Energy
Use cases already deployed in the Energy sector, or which are likely to be seen in the near future.
These include:
- energy trading: AI will optimise energy trading. Pilot energy trading platforms using distributed ledger technologies are already emerging (for example, in Brooklyn, USA, LO3 Energy has teamed up with Siemens to create a pilot microgrid; and financial institutions are piloting commodity trading platforms). The next logical step will be to use machine learning to predict trends and autonomously execute transactions;
- oil and gas: advanced data analytics and machine learning will help optimise exploration and production. AI is also expected to facilitate problem identification and solutions in real-time, scanning databases and implementing solutions. Unmanned so-called “Floating LNG” is already a under development, with concepts already published;
- facility operation and maintenance: self-learning weather forecasting models (such as SolarPulse 3.0) are already being used to predict output from solar and wind power plants, but in future smart sensors in energy infrastructure (such as transmission lines or gas turbines) will be used to enable better predictive maintenance; machine vision will be used to facilitate onsite maintenance checks in conjunction with drone technology to access remote areas (such as gas pipelines or offshore wind farms); and AI may make wholly unmanned power stations possible;
- distributed energy resources: machine learning will assist with the integration of variable generation, and could optimise their participation as demand-side response. For example, in the UK, Herriot Watt University and Upside Energy have recently been awarded grant funding to develop algorithms for grid prediction and demand response portfolio management;
- utilities: AI will facilitate automated trading and consumer tariff switching, making virtual power plants possible; AI will also be able to optimise smart appliances (and other demand-side response), energy storage, and distributed energy sources;
- system operation and network management: intelligent networks or “smart grids” will contribute to energy security by optimising the usage of electricity generation, storage, and demand-side response, improving forecasting, minimising the risk of outages, and enhancing system security.
Infrastructure, mining and commodities
AI-enabled robotics will be widely deployed in the Infrastructure, Mining and Commodities sector. We expect to see significant advances in safety, in the quality of prospecting decision-making, and in sustainability in some areas.
Nick Merritt, Global Head of Infrastructure, Mining and Commodities
Use cases in the Infrastructure, Mining and Commodities sector include those relating to mining, waste, and precision agriculture.
These include:
- mining: many mining businesses have invested in AI technology in recent years to increase productivity and bring huge cost savings to the industry. Uses of AI in this sub-sector include: (1) autonomous drills and haulage (trucks and trains), often managed by operators many miles away in central control centres; (2) robotics and sensor-based ore sorting, including automated real-time mixing of aggregates from multiple mines; (3) drones for exploration, surveys, and mine operations; and (3) big data analysis for more accurate prediction of mine locations. AI-enabled robotics are increasing safety and reducing costs, leading to the possibility of mining previously untenable areas (such as deep underground and deep sea beds);
- waste: processes using AI are emerging that can be used to enhance sustainability in the waste sub-sector (including recycling), particularly in relation to “smart cities”. AI may be used in waste collection, including driverless rubbish collection vehicles;
- precision agriculture: AI is being used to target fertilizers to specific fields and crops, using drone technology.
Financial institutions
Banks and financial institutions possess and generate vast amounts of data. Their internal (human) processes often lend themselves to automation. Banks depend on legacy systems which can impede their ability to extract full value from data (such as customer insights). They are also subject to a great deal of regulatory oversight to protect both the interests of customers and to guard against systemic financial risk. Their regulatory environment is characterised by requirements for enhanced risk management and resulting increased costs.
Increasingly banks and financial institutions see AI as a way of addressing many of these burdens as well as providing the foundation for new product development and augmenting highly contested market share.
James Bateson, Global Head of Financial Institutions
AI has considerable scope for application in relation to the Financial Institutions sector.
Use cases include:
- stress testing: in 2016 a leading bank used an AI system to help it pass the U.S. Federal Reserve’s stress test (it had failed the previous year). Using AI-driven software, the bank adopted a new approach, helping it to identify, validate and select variables and models which ensured that: (1) business logic was built into the process; and (2) the bank had accurate, defensible revenue forecast models which stood up to the Federal Reserve’s scrutiny;
- robo-advisers: these are online wealth management services that provide automated, algorithmic portfolio management or investment recommendations. They are becoming a particular focus for regulators. In the UK, for example, the Financial Conduct Authority has set high standards for robo-advisers (which have to meet the same standards as are applicable in relation to face-to-face advice). German-based Deutsche Bank is reported as “having launched AnglageFinder, a robo-advisory service that asks customers about their investment objectives, terms and risk appetite and puts together customised asset allocations based on their responses.” Robo-adviser start-ups are launching in many jurisdictions – for example, in Canada, at least eleven robo-advisory start-ups were launched in 2015 – 2016;
- robot data entry: although not confined to the Financial Institutions sector, data entry is a significant cost for banks and financial institutions, and accordingly they are now beginning to use software "robots" to replicate the actions of humans in relation to entering data. Once trained, the system can automatically manipulate data, interface with other systems, and process transactions. Such systems typically include digital image recognition tools for capturing data;
- consumer finance: Australia’s ANZ Group is reported as having used IBM's Watson AI computer “as a tool for financial advisors to use as they liaise with customers…. ANZ trained Watson by loading it with information including product disclosure statements, market data, financial statements and terms and conditions of its wealth products. It narrowed down to a thousands-long list of questions what customers were likely to ask their advisors and created a thesaurus to help Watson understand non-standard terminology that might be used in a client query”;
- monitoring: there is a growing array of AI systems with the ability to learn while monitoring markets for unusual behaviour. For traders, those patterns can signal emerging volatility. For regulators, AI may help detect those acting on inside information or engaging in market manipulation;
- regulatory impetus: new regulation will increasingly be the impetus for AI deployments (in part, because of the data generated). For example, in Europe MiFID II / MiFIR (legislation which is applicable in EU Member States from 3 January 2018) impose measures in relation to market participant activities, and demand vast amounts of data. MiFID II introduces closer regulation and monitoring of algorithmic trading;
- payments industry: AI may be deployed in sanctions screening, anti-money laundering, fraud prevention, customer retention, message repair and exception handling, payment reconciliation, and payment validation and authorisation;
- capital markets: AI “RegTech” tools may be used to monitor market abuse and other compliance issues. AI can mimic human emotional responses (in the context of investment decisions). A logical analysis of market sentiment is already available.AI could potentially learn from data generated from such sources to anticipate human investment behaviour. For example, Manulife Financial is reported as having collaborated “with Nervana Systems, a U.S. company, to develop an AI-based application to help asset portfolio managers analyse high-volume online information, financial news, emails and other types of social data, and to make decisions based on the above data much more quickly than what is possible on their own”;
- insurance: insurers are investing in ways to consider how AI can be applied to both the underwriting and claims stages of the product journey. In particular “chat bots” linked to complex algorithms are expected to improve the customer experience in the underwriting phase, and both speed up processing and reduce fraud at the claims stage.
Life sciences and healthcare
By analysing huge quantities of clinical trial data and medical records, AI has the potential to transform the development of drugs and the provision of healthcare in the Life Sciences and Healthcare sector. Stacey Martinez, Global Co-Head of Life Sciences
Use cases for AI in the Life Sciences and Healthcare sector.
These include:
- big data analytics: Google’s DeepMind Health Project and Zebra Medical Vision for Oncology are leveraging Big Data analytics to detect and diagnose disease and provide personalised treatment plans. Such so-called “precision medicine” increases the efficacy of treatments; Microsoft is reported as having “launched Project Hanover with the Knight Cancer Institute to develop AI to personalise drug prescriptions for patients, using machine learning to understand how a tumour is reacting to treatments and is working with researchers in Redmond, Washington to use AI to program cells for fighting cancer and other disease”;
- predicted outcomes: in the U.S., Veterans Affairs is reported as having used “AI to predict medical complications and treat combat wounds, leading to better healing and lower medical costs”;
- virtual doctors and nurses: technology is being developed to help those with chronic conditions to monitor their condition continually. In the UK, the University of Essex and the UK’s National Health Service are reported as “developing automated online doctors to treat patients which would save millions of pounds per year in health care costs. The automated doctors will provide medical advice to patients who seek medical information online regarding self-treatable conditions such as colds or the flu, which consume £2 billion annually of the national health budget.”8 MD, a Norwegian company, has developed a “personal health assistant” which offers users personalised advice about their medical complaints, taking into account the user’s symptoms, and then matching these against a “map” of clinical data about illnesses compiled from public sources and contributing doctors;
- robotic-assisted surgery: Google is developing with Ethicon (a division of Johnson & Johnson) a robotics-assisted surgical platform to develop improved healthcare delivery in operating theatres. These “surgical robots” are intended to give operators greater control and accuracy than is possible by hand, thereby minimising trauma and scarring, and enabling quicker post-surgical healing.
Technology and Innovation
The technology and Innovation sector businesses can use AI to reconstitute their own business models - by, for example, automating customer support or gaining customer insight. There are also huge opportunities for them to develop and sell AI solutions to other businesses. Nick Abrahams, Global Head of Technology and Innovation
Particular use cases for the Technology and Innovation sector.
These include:
- intelligent automation: AI in conjunction with robotics can undertake complex or time-consuming physical tasks formerly undertaken by humans. Such functionality is already being deployed as so-called “robotic process automation” by outsourcing vendors in order to achieve labour arbitrage (reductions in headcount);
- mobile assistants: AI can transform smartphones into mobile assistants. Tech giants are currently competing to develop the dominant mobile assistant for consumers;
- diagnostics systems: AI systems with natural language processing ability and recourse to user manuals can: (1) assist engineers in diagnosing and suggesting solutions to difficult technical issues; and (2) learn from the process over time in order to achieve better outcomes;
- cyber security: AI is being used to predict which files are malware and to look for patterns in how data in the cloud is accessed, and report anomalies that could predict security breaches;
- intelligent sales: a well-known silicon chip manufacturer uses AI to segment customers into groups with similar needs and buying patterns. It uses such information to prioritise its sales efforts and to tailor promotions;
- augmented governance: the board of one well-known global tech giant is understood to have used AI as a “co-pilot” to improve board decision-making in relation to safeguarding the interests of stakeholders and shareholders.
Transport
AI is already widely deployed in a wide range of transport-related applications. The regulatory implications of this are only now beginning to be explored by national governments. In the absence of specific legislation, complex liability issues could arise. Consumer confidence will be a critical factor in uptake in relation to passenger transport and autonomous vehicles. Harry Theochari, Global Head of Transport
In the Transport sector, AI has potential for deployment in use cases across a large range of sub-sectors.
These include:
- shipping: AI may enable unmanned ships to operate (albeit with remote human monitoring). Shipping businesses are considering using AI to analyse market conditions to forecast more accurately (and price for) market movements, find out more;
- commercial aviation: although auto-pilot systems have existed for many years, new types of highly sophisticated AI-enabled auto-pilot systems (using ANNs) are being developed to enable “intelligent flight”. Such systems will be capable of learning and responding to unexpected or unusual flying conditions entirely autonomously;
- unmanned aerial vehicles (UAVs): manufacturers of UAVs for the military are considering ways to use AI-enabled UAVs to respond actively and intelligently to battlefield and environmental scenarios. Major logistics businesses and retailers also wish to deploy fleets of UAVs for autonomous deliveries;
- rail: driverless trains have been deployed on light rail systems for a long time. The rail industry is now employing AI in other areas. For example, the nightly engineering work on Hong Kong’s subway system is scheduled and managed by an AI program designed to use human knowledge taken from a variety of experts and to comply with all local regulations;
- road: autonomous (driverless) vehicles on roads are currently being trialled in many countries. The autonomous vehicle industry is projected to be “worth US$2.6 trillion a year within 15 years, due to advancements in AI and disruptive business models.” See Norton Rose Fulbright, Autonomous Vehicles for more information.