A Thing
The Green SheetGreen Sheet

The Green Sheet Online Edition

January 08, 2024 • Issue 24:01:01

Insider's report on payments
Does AI have a bubble problem?

By Patti Murphy
ProScribes Ink

Patty Murphy

I may get blowback on this, but the way I see it we are living through an AI bubble, and like other bubbles before it, this one will burst. Before continuing, let's agree this is not necessarily bad news, and that artificial intelligence is not inherently bad.

Machine learning tools and other forms of AI have been used for generations, with success, for banking and payments applications, including fraud prevention and detection. And there will be numerous opportunities for the technology to continue to make contributions to banking and payments in 2024 and beyond.

But the seeming hysteria around AI: fears about how we're going to lose our jobs to AI tools like chat bots; the money: private equity investments ballooning from $6.67 billion in 2015 to $47.4 billion in 2022, according to Goldman Sachs; plus, the squabbles and jockeying for position among titans of tech that seem to have more to do with ego than serious efforts at building out an AI industry—all these things give me pause.

While they are prone to bursting, bubbles can bring important advances. Some prominent companies were born during the dot-come bubble of the late 1990s and early 2000s. Amazon, eBay and Priceline are among them.

What made the dot-com and other bubbles burst was indiscriminate investment strategies. Investors were seemingly throwing money at anything that leveraged this new sales and delivery channel called the internet. The problem was that many of these companies burned through the cash with outsized salaries and cushy perks, but without real products that brought returns on investment.

When the dot-com bubble burst, like so many before and after it, small fortunes made seemingly overnight evaporated just as quickly. Back in August 2023, Edward Stanley, head of thematic research at Morgan Stanley, suggested a similar outcome could await AI.

He wrote in a research note that investors should be wary of the "bubble-like euphoria" around AI.

As I noted, bubbles aren't always bad. Back in the 1980s and 1990s fiber optic technology formed the basis of a bubble involving telecommunication firms. That bubble burst but left in its wake massive amounts of fiber optic cable still in use today or waiting to be lit up.

And the dot-com bubble was the genesis of ecommerce, and as has been documented, it soared during the pandemic and continues to grow unabated. Late last year, Adobe Analytics predicted online sales would near $22 billion just in the 61 days between Nov. 1 and Dec. 31, 2023.

So, even if we're in the midst of an AI bubble, and even if it bursts like most bubbles do, it's not the death knell for AI technologies, or for payment initiatives that leverage AI.

Promise and problems

The public frenzy over AI began in late 2022 with the launch of ChatGPT, a chatbot heavily financed by Microsoft. It uses generative AI, similar to automated online customer support chats, that can support text, images and other media using training data and models.

The launch sparked an AI "arms race" of sorts, with big tech companies upping their AI investments by billions of dollars and all tech companies angling to deliver new AI applications. OpenAI, the company behind ChatGPT, was created in 2015 by a group of tech moguls, many of whom have since moved on to other AI projects. On day one, the chatbot claimed over a million users.

To what does ChatGPT owe its success? I posed that question to the chatbot itself. The answer was a bit long to recount here, but these are a few highlights:

  • It is "trained on a diverse and extensive dataset collected from a wide range of sources on the internet."
  • It is built on "transformer architecture," which uses neural networks that can put data into context and predict likely sequences of words in response to prompts or queries.
  • OpenAI makes the technology available through APIs, making it easy to integrate with applications and services.

The response included this caveat: "It's important to note that while ChatGPT has achieved remarkable success, it also has limitations and may generate incorrect or biased responses in certain situations."

After all, the internet, which ChatGPT scours for information, is filled with misinformation. Mistakes are a given. For example, there was the widely reported story last year about an attorney who used the ChatGPT to research court decisions in support of a client's position in a lawsuit. Upon reviewing the filing, the judge discovered that many of the cases cited never existed, and the attorney found himself in hot water.

AI in payments

AI has long played a role in payments. One of the first applications I recall from early in my career, was the use of neural networks to read transaction amounts on checks and biller remittance slips for faster processing.

And machine-learning tools continue to play pivotal roles in combating fraud. Visa revealed in a September blog post that it has spent in excess of $3 billion on AI and data infrastructure to support safe money movement and fraud prevention.

AI in fraud prevention will become increasingly important as more businesses and consumers take advantage of real-time payment networks. Speedy and accurate decisions are critical to the success of these networks. Mastercard gave a nod to this in July 2023 when it said it was leveraging AI technologies to help UK banks stop scammers in real time.

In November, Mastercard unveiled Shopping Muse, an advanced generative AI tool that recreates the in-store human experience for consumers shopping a store's digital catalog. "At Mastercard, we're putting technology and machine learning to work to deliver better outcomes for both brand and consumer," said Raj Seshadri, president of data services at Mastercard.

Visa is on a similar path, stating that it is working on several fronts "collaborating to enhance and improve commerce with generative AI across the consumer shopping journey."

Beyond the payment brands and networks, ISOs and agents may want to look to AI to augment credit decisioning. This becomes ever more important in an age of instant customer onboarding, Allen Kopelman of Nationwide Payment Systems noted in "Let's eliminate friction from merchant services," published by The Green Sheet June 12, 2023, issue 23:06:01. end of article

Patti Murphy, self-described payments maven of the fourth estate, is senior editor at the Green Sheet. She also co-hosts the Merchant Sales Podcast, and is president of ProScribes Ink (www.proscribes.net)

The Green Sheet Inc. is now a proud affiliate of Bankcard Life, a premier community that provides industry-leading training and resources for payment professionals. Click here for more information.

Notice to readers: These are archived articles. Contact names or information may be out of date. We regret any inconvenience.

Prev Next
A Thing