Last reviewed 17 April 2019
Unnerved by HAL in Stanley Kubrick’s 2001: A Space Odyssey? Charmed by R2-D2 and C-3PO in Star Wars? Like the films, the role of Artificial Intelligence (AI) in modern life brings both opportunity and cause for concern. Laura King considers the role of AI today and tomorrow, as well as some of its inherent risks.
AI is a type of computer science that allows machines to sense their environment and then think, learn and act like humans. As a cornerstone of the so-called fourth industrial revolution, AI is a rapidly growing sector and a recent analysis by PricewaterhouseCoopers (PwC) suggested it has the potential to contribute a staggering $15 trillion to global gross domestic product by 2030. Improved automation, the provision of better information to the workforce, and increased demand from customers due to the creation of personalised goods are just some of the benefits expected. However, such a sea-change rarely happens without impacting on something or someone, and in this case, the same report also estimates that by the mid-2030s, 30% of jobs are at risk of automation, with poorly educated workers the most likely to be affected.
With such growth and impact, it’s no wonder that businesses are asking what opportunities AI might bring, as well as whether disruption brought about by AI might threaten how they currently do business.
Environmental sustainability and AI of today…
Whether we like it or not, AI already surrounds us. From movie recommendations, to spam filters, chatboxes and internet searches, we are all already interacting with AI. In these examples, most of the AI is based on “deep learning” that uses pattern recognition to make predictions and recommendations. And this is just the tip of the technological iceberg. Within the realm of environmental sustainability, AI also promises big things.
For example, in October 2018, United Utilities started using AI across its entire water network in the North West of England. The AI platform analyses data such as weather and demand for water to decide on the best ways to run pumps, detect burst pipes, and minimise the risk of discoloured water. The roll out followed a three-month pilot which demonstrated energy saving costs of 22%.
Other companies have used AI to optimise demand side response contracts and build intelligent networks that connect energy producers, consumers and storage. In the field of renewables, AI is allowing companies to make better predictions of renewable energy generation from weather forecasts.
Transport too, is expecting big changes with the promise of automated vehicles and technologies that can learn and respond to driver behaviour to improve road safety. Although fully automated vehicles might still be a way off, aviation and ferry companies are already using the technology to identify fuel-efficient routes for air and sea travel.
And it’s not just business who are benefiting. AI is also contributing to global efforts to improve the environment and conditions for society. For example, Global Fishing Watch helps provide much-needed transparency on the locations of fishing vessels to help sustainably manage fishing and marine environments, and researchers at Stanford University have found a way to process high resolution satellite images to locate communities living in poverty in five African countries, helping to improve how relief is delivered.
AI of tomorrow …
Although AI shows considerable promise, it is certainly not an easy win, requiring expertise, good clean data and significant processing capacity. Early adopters include tech companies, energy-intensive industries and manufacturing, but the benefits of AI do not need to stop there. The Ella MacArthur Foundation, for example, has identified AI as a potential game-changer for the circular economy, and analysis of big data has the potential to result in developments such as real-time sustainability reporting.
Can I use more AI in my business?
When deciding if AI can be developed as part of the business or organisation, the following questions need to be considered.
AI as a strategy — Is AI a good fit for the business, and where does it fit into the business model? Is there support (both financial and strategic) from top management?
AI governance — Does the business have a sound governance structure for AI? For example, how does the organisation monitor and set standards for the new technology?
Data for AI — What data does the business have access to — is it of high enough quality and quantity for AI to be able to analyse, learn and act on?
Technology — Does the organisation have the technology capabilities to process the data?
Resources — What resource is there for AI within the organisation? Considerable expertise is needed that might initially be outsourced, but to be successful, staff need to be involved too. Going forward, it is important to make sure the company has the specialist expertise needed, but also that it also has the resources to up-skill and train its workforce.
Proceed with caution
Despite its obvious benefits, AI is not without its ethical challenges.
One of the key concerns around AI is how to make it accountable for its decision-making. At the moment, many applications are described as “black boxes” with no explanation for the reasoning or rationale behind decisions. This might be acceptable if the recommendation is something relatively minor, such as which film to watch next, but if the decision-making has wider, or ethical, implications then this situation is clearly untenable.
Developments to move towards what is known as “Explainable AI” are already underway, driven by this very real need for transparency and trust: it’s clear that AI cannot move forward or be useful if its output does not inspire the confidence of customers or stakeholders.
Second, more work needs to be done to develop digital ethics, whereby there are checks and balances in place to make sure that decision-making complies with moral codes and standards. This is especially the case with risks around automated bias, for example, if machines identify and act on patterns in data that are based on unfair bias, such as those related to gender or race.
Data protection and security is also increasingly important. This is not only crucial from the point of view of security breaches. Individuals also need to have adequate protection from companies and organisations looking to collect ever-more data for training and improving AI machines and algorithms.
Finally, organisations need to be alert to the consequences of AI. The impact on jobs is one such example, but unintended consequences are another cause for concern. For example, automated cars may improve fuel efficiency and therefore be better for the environment, but will they also increase the number of journeys made, negating any benefits?
To help protect individuals, AI is regulated for under the General Data Protection Regulation, and the Government has also developed its own Digital Ethics Framework for use in the public sector. It has also set up the Centre for Data Ethics and Innovation which will look at how data is used to help maximise the benefits for society.
AI is an emerging trend that will change the course of business. When considering whether AI has potential in an organisation, start by considering whether the right governance is in place, as well as what AI can do for the business. Consideration will then need to be given to the data quantity and quality available, processing capacity and resourcing of the project.
AI has the potential for great things, but there are also very real concerns, and it is for governments, businesses and developers to create systems and frameworks that are mindful of the potential pitfalls such as unintended consequences, trust and accountability, unintended bias and data security.