AI in 2020: Artificial Intelligence? Or just plain artificial?

While some facets of the sci-fi classic have come to pass, such as voice-command technology, and others are currently unfolding, such as the destructive effects of climate change, the dramatic AI advances that fool Harrison Ford still feel a long way off.

But that hasn’t stopped AI from becoming one of the buzziest terms thrown around by technology and financial services experts. Fintech startups use the word to make products appear cutting-edge and established firms find they have to offer “AI” to remain competitive.

The products they are referring to as AI, though, are often little more than new data models or even just outright vaporware — technology sold on a conceptual level that hasn’t been built yet, according to many technology leaders in the advice industry.

“There are a lot of packages out there that are more ‘artificial’ than ‘intelligence’,” said Raj Madan, BNY Mellon Pershing’s managing director of technology. “It’s not AI, just pure data analytics.”

The challenges facing the advice industry when it comes to implementing AI are less about obtaining AI technology, and more about firms not having the kind of data needed for AI to operate efficiently. That said, AI is expensive to invest in and hard to define.

For example, “object character recognition,” in which a machine can interpret text or images from an image or document, used to be considered cutting-edge AI, but is now commonplace. That doesn’t mean there aren’t promising developments out there.

Unlike other buzzy technologies like blockchain that have few proven cases of actual usage in financial services, there are genuinely practical applications for AI in this industry, said Mark Casady, general partner and founder at fintech venture capital firm Vestigo Ventures.

“AI is very investable right now,” Mr. Casady said. “There are a lot of smart entrepreneurs that know how to use it.” Misconceptions abound, though, about what the technology is capable of, he said.

Some advisers think AI can be an all-knowing tool that can instantly generate new business or cure whatever ails a firm.

Other advisers jump on the idea of AI without understanding the kind of data the technology requires.

“This one-size-fits-all, this magic AI pill you throw and all of a sudden it produces results, it’s just not true,” said Doug Besso, HighTower’s chief technology officer. “You need data, it has to learn from something.”

Because the technology can be complicated, it’s often difficult for advisers to distinguish which technology has promising business applications, and which has only promises.

“There’s a lot of noise in the ecosystem and it’s not easy; I struggle myself.

There’s a lot of opaqueness from a technology standpoint,” said Ashish Braganza, LPL Financial’s executive vice president of market intelligence, data and analytic solutions. “A healthy dose of skepticism is always good.”

Academia distinguishes artificial intelligence by two categories: weak AI and strong AI. Strong AI typically refers to a machine that can apply intelligence to any problem, though some academics reserve the term for a machine that can experience consciousness, if such a thing ever becomes possible.

Weak AI is a machine that implements a limited mind-simulating human cognition by automating tasks and analyzing data. If the tool is focused on solving a single specific problem, it’s often called “narrow AI,” which is how academics typically define most modern applications of AI.

Digital assistants like Siri and Alexa are examples of narrow AI. These don’t rise to the level of weak AI because they operate within limited, predefined rules.

Of course, most advisers aren’t interested in academic debates on the definition of artificial intelligence, Mr. Madan said. They just want technology that will effectively help them manage their businesses and grow revenue.

For many technology decision-makers in the financial advice industry, the most important part of AI is machine learning — when a machine can improve its functionality over time and adapt to changing circumstances without manual human input.

“If you throw a ball at someone’s head, you’re going to hit them in the head if they don’t know to move their head out of the way,” Mr. Madan said. Once it learns to avoid the ball, that’s a form of intelligence, whether it comes from a person, animal or machine, he said.

Pershing has succeeded in using machine learning to improve its ability to detect anomalies, Mr. Madan said. Beyond the most obvious cases of fraud detection and data security, the firm is looking at how AI can help advisers pick up on irregular client behaviors.

For example, if a client is suddenly downloading all of their statements or has stopped logging in to check trades as they normally do, it could be the sign of an attrition event. At the very least, the AI could signal to advisers that it’s time to check in with that client.

Mr. Besso said his firm is experimenting with Salesforce’s Einstein AI platform to see if detecting anomalous behavior can help prevent a client from leaving an adviser. The idea is to use the Salesforce client relationship management system to look for strange behaviors, such as slower response times to meeting requests or not opening advisers’ emails.

Once a client starts withdrawing funds, it’s already too late, Mr. Besso said.
The hope is AI can detect a pattern before that point.
“Getting clients is hard, so saving them [from leaving] would be much better,” he said.

Instead of the nebulous term “AI,” Mr. Braganza said the technology team at LPL prefers “AAAI,” referring to Assisted Augmented and Autonomous Intelligence, to better articulate what the technology is capable of doing. Assistance technologies refer to programs like Alexa and Siri, while autonomous represents various forms of robotic processes that automated traditionally manual workflows.

The augmented category is what Mr. Braganza is most excited about today.
These are prescriptive technologies designed to help advisers by finding and recommending new opportunities, like the Next Best Action engine introduced by Morgan Stanley in 2018.

LPL is currently testing a tool with its internal sales team that can tell representatives who they should talk to, why, and provide them talking points before the meeting, Mr. Braganza said. Late last year, the firm implemented a new algorithm that can recommend products to end clients.