The advent of ChatGPT has sparked a massive debate about artificial intelligence (AI). Can we regulate it? What does it mean for the future — will we all lose our jobs? How can AI be used in a way that helps humans, not destroy them?
And yet, this debate is not new: we have been facing these questions in financial markets for years. In the previous century neural networking threatened to replace traders and investment bankers.
The issue with AI is that it is nebulous: AI is not one thing, but many. It’s a bit like cloud computing: using AI is not a generic, it’s a specific. Adding to this conundrum is that financial markets and technologists are great at coming up with three-letter acronyms (or TLAs for short), that confuse things even more.
TLAs, like all jargon, are a great way to complicate and baffle those who aren’t are deep in the weeds and working in this area full-time. CDOs, ETFs, FTP, HTML and even AI can confuse.
The issue with AI is that it is nebulous: AI is not one thing, but many
So, let’s add another one: LLM, which stands for large language model. LLMs, according to US marketing and sales firm TechTarget, are a type of AI algorithm that uses deep-learning techniques and massive data sets to understand, summarise, generate and predict new content. It is the basis of things like ChatGPT, Bard and many others.
So, how are financial firms using LLMs?
Among many examples, I’m going to pick out just two: American Express (AmEx) and JPMorgan.
According to Laura Grant, vice-president of product development for emerging platforms and AI at AmEx Digital Labs, the company is rolling out pilot projects with LLM-based AI to understand how it can help with its ‘3 Ps’: making a product more personalised, proactive and predictive.
JPMorgan is using LLMs in many different areas. In April, chairman and CEO Jamie Dimon revealed that the bank has more than 300 AI use cases in production for risk, prospecting, marketing, customer experience and fraud prevention. “AI and the raw material that feeds it, data, will be critical to our company’s future success,” he said in an April letter to shareholders. “The importance of implementing new technologies simply cannot be overstated.”
What gets me here is that, in both cases, banks and financial services companies are using AI to enhance the customer experience, augment the human and automate the mundane. That makes absolute sense, but how far can we push this?
Consider flash trading; we can push this to an extreme where machine competes against machine in the time it takes to trade. If you want the best deal in financial services, you have to beat the other to the deal. In dealing and business, this becomes even more critical. If you get the right price at the right time, you’re the winner.
But, with generative AI already being so pervasive, what happens if generative AI gets you the deal or best price before others? Take the consumer example of automated bidding services on eBay, the auction site, which will guarantee your bid within a second of the item closing, beating anyone who hopes to bid manually.
Expand this example to trading and you will get markets that are no longer trading fairly. They are connected by bots that bid and manage on our behalf. They are automated and structured in a way where those who don’t have the technology simply can’t compete.
If you don’t know the acronyms, the technologies and the systems being used, then you shouldn’t be in these markets because you can’t compete.
On the one hand, AI will benefit customer services, as evidenced by AmEx’s experiment. On the other, it is a competitive weapon that may create markets that are biased towards insiders.
All in all, I always return to a common theme around AI: it should augment the human and automate the mundane.
