Watch the news for an extended period and you’re all but certain to hear Artificial Intelligence (AI) mentioned. Crunchbase reports that $50 billion was poured into AI startups in 2023, and many believe that Artificial Intelligence could eventually prove to be as transformative as the Internet. Long a boogie man in Science Fiction, respected technology leaders have warned that while AI offers tremendous potential, it could also be a threat to humanity. No one knows what the future holds, but AI will have a major impact on society, like it or not.
Among many other things, AI is transforming how merchants and security experts fight fraud, process payments, and manage disputes, among other things. AI-powered chatbots can handle customer support, for example, and responsive customer service typically reduces chargebacks. AI may also make it possible to develop more sophisticated transaction monitoring like fraud filters. Indeed, card networks like Mastercard are currently developing tools for said applications.
Let’s take a closer look at some of the key concepts surrounding AI and how they’re applied to fraud, disputes and chargebacks.
What is Artificial Intelligence?
First, let’s make sure we understand what’s meant by Artificial Intelligence. The technologies currently being used are not “sentient” per se; AI programs can’t think for themselves and aren’t conscious like a living human being. Some researchers are exploring the possibility of artificial consciousness, but even if that’s possible, it’s likely a long way off.
AI however simulate complex human thinking and problem solving in a limited way. In some cases, AI programs can use machine learning concepts to learn and adapt. AI can also use Large Language Models (LLM) and other tools to create content. Algorithms can be written to address various challenges and analyze data. Still, most AI systems in use right now are quite narrow in what they do; they’re not nearly as dynamic as human intelligence.
Researchers are working on Artificial General Intelligence (AGI), which would have far broader abilities and would be able to simulate many cognitive functions in a flexible way that matches or exceeds human abilities. A fully realized AGI may be able to accomplish any mental task that a human could. For now, we’ll have to wait to see what the future holds as far as such advanced AI is concerned. Current AI tools, however, may still prove to be very useful.
Will AI Prove Transformative?
Predicting the potential for new technologies is tricky. When the Internet first emerged, many companies and people rightly recognized that it would transform how we shop, communicate, share knowledge, and more. In 1994, Bill Gates believed that the Internet had limited commercial potential within the next ten years, but by 1995 he was describing it as a tidal wave. More recently, some felt that the blockchain would quickly turn out to be as transformative as the Internet. Yet while blockchain solutions have proven innovative, so far they haven’t lived up to the hype.
Still, early AI tools have proven to be highly effective. AI content generators, like ChatGPT, can write convincing lines of text, and even produce decent fictional stories, poems, and other bits of content. AI is a long way from winning a Pulitzer, but early progress has been promising.
Crucially, companies that fail to embrace AI to fight fraud, boost customer service, and otherwise improve operations, may slowly fall behind their competitors. They could also become favored targets for hackers and other criminals. In the long run, AI may offer competitive advantages to early adopters. Meanwhile, those who fall behind may end up stuck behind.
Using AI-Empowered Customer Service to Reduce Chargebacks
One of the most promising early uses for AI and reducing chargebacks has been the use of chatbots. Chargebacks sometimes stem from miscommunication. If shipping is delayed, a customer might think that their order was never filled or that it got lost in the mail. They might write to your customer service department, but if you don’t respond quickly, they could file a chargeback.
AI chatbots can be very effective for digging up information, like tracking data and “talking off” disputes. The customer may have been able to find that information on their own if they put in the effort, but the fact is, they prefer to have someone (or in this case, something) do it for them. If AI can do that work, it could improve outcomes for everyone. Roughly 40% of merchants are already using chatbots and setting up such solutions is relatively simple. Facebook’s chat program makes it easy for brands to use chatbots to handle customer relations via a merchant’s Facebook page. IBM reports that chatbots can reduce customer service costs by 30% and that chatbots can handle up to 80% of routine tasks and customer inquiries. This is crucial because customers often expect quick responses. Indeed, about a third of customers expect a response to social media inquiries within a half hour or less.
You can also program AI chatbots to escalate challenges and inquiries under certain conditions. For example, if a customer mentions that they are going to file a chargeback, the AI bot might connect them with a human representative right away. The customer rep can then talk off a dispute. This way, the merchant can avoid chargebacks, fees and a rising chargeback ratio. If a chargeback ratio gets too high, merchants get hit with severe penalties and can be dropped by their acquiring bank and banned from card networks.
AI chatbots can help customers with a wide range of tasks, such as order tracking, payment information, questions about prices, and more. By clearing up confusion, it may be possible to prevent chargebacks.
AI-Empowered Crime Fighters
Spotting fraud is difficult even if you know what to look for. Many merchants handle hundreds or thousands of transactions (or more) per day. It’s often relatively easy to set up alert systems to identify suspicious transactions. AI can go beyond simply filtering data for red flags, and instead can dynamically analyze patterns to uncover suspicious activity. AI and automation tools may also be able to communicate with customers automatically to gather more information or confirm their identity.
Card networks like Visa and Mastercard are pushing forward with AI initiatives. Mastercard recently announced that its Decision Intelligence Pro generative AI program can analyze over a trillion data points and resulted in a 20 percent increase in accurately detecting fraud compared to the older Decision Intelligence program, which itself has been highly effective at uncovering fraud.
As AI advances, crime fighting tools will continue to improve. That said, AI could also be used by criminals and fraudsters to supercharge criminal activity.
The Downside of Artificial Intelligence
Clearly, AI could offer many benefits. However, AI could also present serious risks. Cybersecurity experts can use AI to identify and close security gaps, yet at the same time, criminals could use AI to exploit vulnerabilities, such as poorly written code in a software program. AI might make it much easier to identify and exploit “zero-day” exploits, which are vulnerabilities unknown to the original developer/owner.
Many fraud campaigns rely on phishing and similar tactics. Traditionally, phishing has been labor-intensive. This often meant writing long messages, scheduling emails, teaching victims how to send money, and carrying out various other onerous activities. Now, AI can write and send messages, respond to inquiries, offer support for sending payments, and more. One report found a greater than 1,200% increase in phishing emails since ChatGPT rolled out to the public.
If AI boosts productivity for criminals, it could lead to more fraudulent transactions, and thus more chargebacks. Google has warned that zero-day exploits, phishing, and other cybercriminal activities could become more widespread and harder to defend. Google has even warned that AI might actually be able to write better phishing emails than human criminals.
Ultimately, Artificial Intelligence could become an arms race for evil, as criminals use it to perpetrate crime — forcing the hand of merchants and security experts to answer back with AI of their own. Even if good ultimately beats evil, a lot of merchants, cardholders, and other stakeholders could suffer along the way. That said, those parties that try to ignore AI could suffer even more extensive losses by falling ever further behind in the arms race.
Conclusion: AI Could Be Transformative
Early indications suggest that AI will have a profound impact on the economy and human society at large. Statista concludes that the total market size for AI will grow exponentially, from about $242 billion in 2023 to $740 billion in 2030. Many believe that we have only just begun to understand AI’s full potential. Merchants, cybersecurity experts, anti-fraud specialists — indeed all stakeholders — should keep an eye on AI and leverage it to address pressing needs, like reducing payment fraud and chargebacks.