Artificial Intelligence (AI) presents unique opportunities for the payments industry to streamline customer experiences, tailor service offerings and develop new service models;
In May 2019, AusPayNet made a submission to the Data61 consultation, Artificial Intelligence (AI), Australia’s Ethics Framework;
In August 2019, AusPayNet responded to the Standards Australia Discussion paper, Developing standards for Artificial Intelligence: Hearing Australia’s voice.
Point of View
The English Oxford Living Dictionary defines Artificial Intelligence as the “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” Artificial intelligence is a broad field, of which machine learning is a sub-category. Machine learning is a “method of data analysis that automates analytical model building. It is based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.”
Artificial Intelligence (AI) presents unique opportunities for the payments industry to streamline customer experiences, tailor service offerings and develop new service models.
The latest figures released by the International Data Corporation (IDC) indicate that worldwide spending on Artificial Intelligence is expected to reach $35.8 billion in 2019. This will be an increase of 44% over the amount spent in 2018. Organisations are investing more in projects that leverage on AI capabilities, with spending on AI systems expected to more than double to $79.2 billion in 2022. AI spending will see a compound annual growth rate of 38% over the 2018-2022 forecast period.
In Australia, the Federal Government’s 2018-19 budget allocated approximately $29.9 million over four years to support the development of AI. Research analysts House of Brand conducted a study of 200 Australian organisations which included interviews with senior personnel from a variety of companies. The report “Outlook on the Australian AI market landscape in Australia” indicates Australian organisations are forecast to spend significantly more on AI between now and 2022.The number of companies expected to invest more than $1 million in AI will almost double, rising to 13% in 2022, compared to 6% today.
Artificial intelligence is a key pillar of innovation for the payments industry. Payments companies globally continue to investigate the use of AI, looking to leverage on its benefits to improve the consumer experience or improve back-end processing.
Ability to analyse large amounts of data – Artificial intelligence has the capability to analyse large amounts of complex data in real time, beyond factors such as time, velocity and amount. Machine learning systems can learn from each transaction, constantly improving and becoming more effective over time.
Automated decision-making – Artificial intelligence software (i.e. chatbot) uses natural language processing technologies to conduct a conversation with the consumer. This AI application can be used to streamline interactions with the consumer and create a personalised user experience in a highly automated and scalable manner.
A recent report by the World Economic Forum, The New Physics of Financial Services outlines two use cases of AI by the payments industry:
Authentication practices – Facial recognition and voice are examples of AI applications currently used to authenticate payments. These applications provide a seamless consumer experience that automate the purchasing process.
Recommendation and Automated decisions based on payments data – Payments data provides a very clear picture of a consumer’s wants and needs. AI enabled optimisation engines can generate highly specific and accurate recommendations by using machine learning to analyse a consumer’s payment data. By combining payments providers’ unique datasets with machine learning capabilities, merchants are able to gain targeted insights into a consumer’s purchasing behaviour.
Artificial intelligence is also quickly becoming a valuable tool to help reduce fraud and secure e-commerce transactions. In Australia, we are experiencing unprecedented growth in eCommerce transactions. In 2018, device-not-present transactions represented 22% of all card transactions. Online spending also exceeded $25 billion in the 12 months to February 2018. As the eCommerce economy grows, card-not-present fraud is still the most prevalent type of fraud, accounting for 85% of all fraud on Australian cards.
AI can be used alongside an organisation’s existing fraud protection system to detect fraudulent transactions online and prevent fraud from occurring. It has the capability to analyse transaction patterns and detect unexpected commonalities between fraudulent and non-fraudulent transactions. AI is also an effective tool in reducing the number of false positives in fraud protection, therefore increasing conversion rates without adding friction to the payment.
The future of AI-based fraud prevention relies on the combination of supervised and unsupervised machine learning:
Supervised machine learning examines events, factors and trends from the past. Historical data trains supervised machine learning models to find patterns not discernable with rules or predictive analytics.
Unsupervised machine learning finds anomalies, interrelationships and valid links between emerging factors and variables.
As investment in artificial intelligence systems rises, there are challenges that AI presents that need to be addressed:
Decision making and Liability - As AI use increases, it will become more difficult to differentiate responsibility for decisions. Organisations need to determine who should bear the risk when mistakes are made that cause potential losses.
Transparency - When AI systems are used to make important decisions, it may be difficult to unravel the reasons behind a specific course of action. The clear explanation for machine reasoning is necessary to determine accountability.
Bias - AI systems can only process and analyse information they are given. Results produced are dependent on the data the system is fed. The systems can entrench existing bias in decision making systems. Caution must be taken to ensure that AI systems evolve to be non-discriminatory.
Data protection and Privacy - The potential of AI is rooted in its access to large data sets. As AI systems are trained on one specific data set, organisations need to consider what happens when it applies these learnings to a new data set.
For AI to deliver on its promise, there needs to be a multi-layered approach to solving these challenges.
AI Ethics Framework
Our submission to Data61’s discussion paper, Artificial Intelligence, Australia’s Ethics Framework, AusPayNet highlighted the importance of transparency and accountability of any network. Payment networks need to ensure the highest level of security due to the sensitivity of information being transmitted. As such, an AI Ethics Framework should consider how consumer data is used by an AI system and the output generated. It should also address the roles that manufacturers and users of AI will play to achieve a transparent AI system with clear a liability model.
Our submission to Standards Australia’s discussion paper, Developing Standards for Artificial Intelligence: Hearing Australia’s Voice, recognised the importance of developing a national strategy for AI. The implementation of frameworks are critical to manage security and build trust, so Australians can embrace AI innovation and realise the full benefits offered. Our submission further highlighted that security and interoperability should be the key focus areas when setting AI standards. A layered approach that incorporates both standards (national and international level) and governance models is critical to the continued adoption and trust in AI systems.
A co-ordinated approach can help build consumer trust and acceptance, so they are more open to embracing this technology. This will in turn, translate into wider adoption of AI globally.