.svg)
Since its November launch, ChatGPT has sparked a frenzy of interest with its ability to create human-like speech, while raising questions about data privacy and its implications for the labour market and wider humanity. VIXIO considers what generative AI could mean for payments.
ChatGPT, the AI chatbot from Microsoft-backed OpenAI, has made headlines all around the world since its release in the autumn. After two months, more than 100m people signed up for the service and, as of February, it had registered 1bn visits to its website.
It has been called the fastest-growing consumer internet app ever. According to experts, generative AI like ChatGPT or Google’s Bard has the potential to revolutionise payments.
Criticisms have also emerged. Several companies, including J.P. Morgan and Amazon, have prohibited their employees from using ChatGPT at work over concerns about data privacy, and two weeks ago Italy’s data protection authority banned ChatGPT in the country.
Meanwhile, more than 1,000 AI experts, including Elon Musk, have urged a delay in the training of AI systems more powerful than GPT-4, OpenAI’s latest release, until further research proves “effects will be positive and risks manageable”.
How do payment firms use generative AI?
The use of generative AI has a large potential to streamline work in various industries, making processes faster, cheaper and more efficient.
AI chatbots, such as ChatGPT or Bard, have “immense potential to revolutionise the payments industry”, according to Iu Ayala Portella, CEO of AI consultancy Gradient Insight.
“These chatbots can handle customer queries, offer support, and even complete transactions without the need for human intervention,” Portella said.
Among those embracing the innovative technology are fintech giants Stripe and Klarna, which announced collaborations with OpenAI last month to improve their service offering.
Stripe uses GPT-4 to enhance Stripe Docs to help developers find links to relevant docs and give them specific pieces of information or a summary of the relevant parts of the documentation.
Klarna offers customers highly personalised product recommendations via an integrated plugin for ChatGPT. In practice, this means that users can ask ChatGPT to give them a shopping recommendation, which then provides personalised results on Klarna’s embedded shopping tool.
“AI chatbots can analyse customer behaviour, preferences, and transaction history to offer personalised recommendations and promotions,” Portella noted.
This personalised approach can increase customer loyalty, resulting in higher customer retention rates and ultimately more revenue for businesses, he added.
Although AI has already been used for personalisation, generative AI has the ability to bring that to a new level, Len Covello, chief technology officer at loyalty technology provider Engage People, told VIXIO.
In the case of a loyalty programme, rather than just displaying personalised content on the screen, new AI tools can function like a personal assistant, continuing the engagement, reaching out to the customer and understanding the way in which the customer wants to be communicated with.
It means that if you have 30m customers, “you can effectively have a person for each of those 30m individuals and loyalty interactions would be very specific to the customer, very individualised and very engaging”, Covello explained.
Additionally, an AI chatbot could be integrated into financial institutions' customer service to allow the customer to easily navigate payment issues, check their account balances and receive instant support
Meanwhile, the ability to monitor massive data sets and produce suggestions in real time has the potential to make fraud mitigation and risk assessment scores more efficient, Covello added.
But what are the risks associated with chatbots?
As mentioned before, data privacy and security are among the main concerns related to the use of generative AI.
Italy’s data protection agency highlighted that “there appears to be no legal basis underpinning the massive collection and processing of personal data” that is used to train AI algorithms.
Additionally, ChatGPT suffered an outage on March 20 which exposed payment-related information of 1.2 percent of the ChatGPT Plus subscribers.
Another risk is AI models can inherit biases from the data they are trained on. If the training data is biased, the AI model may make biased decisions, leading to discriminatory outcomes.
Meanwhile, chatbots may be exploited by fraudsters to design more credible texts, enhance their social engineering practices or carry out company or government impersonation scams at scale.
“We have already seen evidence of ChatGPT being used to form all the components needed for a fraud scam,” Doriel Abrahams, head of risk at fraud prevention software, US, at Forter, noted.
“While ChatGPT and other generative AI can sound promising, it’s important to understand that the same promise could be explored by criminal experts for the purpose of online and payment fraud,” Abrahams emphasised.
“Imagine fraud rings working together with generative AI and scenarios like the holiday season’s rise in scams.”
“The reality is that in online fraud, humans are often the weak link. With generative AI like ChatGPT, it’s inevitable that fraudsters will exploit human emotion in various pig butchering scams, romance scams, business email compromise schemes and even deepfake phishing,” Abrahams added.
Is it just hype or a transformative invention?
“The hype is there for a reason,” noted Covello, who said we are only scratching the surface of how this new technology could be used.
In a world where people are moving more and more online and these chatbots can understand queries and provide meaningful responses, it seems inevitable to rely on technology with less human interaction.
According to Portella, the benefits of AI chatbots in the payments industry are numerous and their potential to transform the industry “cannot be ignored”.
The fact that the governments of large economies have already taken note of the new technology and are moving quickly to address concerns may also help responsible generative AI tools flourish.
The EU is considering far-reaching legislation on AI, called the Artificial Intelligence Act, which would classify AI systems by risk and mandate various development and use requirements.
Meanwhile, on Tuesday (April 11), the US Department of Commerce launched a consultation on the risks and benefits of AI and algorithmic systems.
“Responsible AI systems could bring enormous benefits, but only if we address their potential consequences and harms. For these systems to reach their full potential, companies and consumers need to be able to trust them,” Alan Davidson, assistant secretary of commerce for communications and information, commented.
