Inside Morgan Stanley's Big AI Test

“AI will disrupt a lot of industries, but the way we think about it is certainly as a productivity lift,” Sal Cucchiara, Morgan Stanley chief information officer of wealth management and investment management technology, tells ThinkAdvisor in an interview.

The firm, which kicked off a pilot test of OpenAI GPT-4 earlier this year with 1,000 of its financial advisors, will roll out the software to all its financial advisors [16,000 as of 2020] by the end of the third quarter, Cucchiara, a managing director, says in the interview.

Even before the pilot, Morgan Stanley was an early artificial intelligence adopter, introducing it in 2020, for example, in its Next Best Action system for customized messages to clients.

The advisors have been using AI in recent years for product offerings as well, Cucchiara says.

“OpenAI and large language models are just the next generation of AI’s advancement,” he adds, noting that Morgan Stanley is “really aggressive in thinking about how we use AI but taking a really conservative approach to rolling it out.”

Cucchiara is responsible for the firm’s IT strategy and the development and delivery of all investment and wealth management business lines, including E-Trade.

The firm is making a strong effort to safeguard its artificial intelligence from generating inaccurate information by using Morgan Stanley’s internal research and the firm’s own content exclusively.

That is, it is “not commingling the internet or anything external to Morgan Stanley,” stresses Cucchiara, who before joining the firm in 2016, was with Bank of America Merrill Lynch for 20 years, serving as chief information officer and head of the wealth and banking solutions technology organization.

In the Morgan pilot test, advisors essentially have a “virtual assistant” that can access hundreds of thousands of documents, including advice and forecasts published by the firm’s Global Investment Office, which provides portfolio guidance.

A key advantage is that advisors are able to access AI responses’ source material to validate the veracity of every answer the system churns out.

The software also synthesizes accessed information to “get to the heart” of advisors’ questions, eliminating the need to pore over long documents for a specific sought-after answer.

The wirehouse’s foray into the new generation of AI mainly focuses on advisors obtaining basic information for administrative tasks to save them time.

However, in the pilot, advisors have the capability to record phone discussions with clients — with their consent — a transcript of which AI synthesizes and summarizes. That document can then be sent to the client after the advisor approves it.”

For the near future, the firm is looking at offering investing opportunities using AI based on client behaviors, such as their actions when visiting the Morgan Stanley website.

ThinkAdvisor recently interviewed Cucchiara, who was speaking by phone from the firm’s downtown New York City office.

AI is “all about how to take ideas that advisors have and get them to clients in a digital way,” he says.

“Having the advisors at the center and making sure they own the content that’s generated really propels our confidence about the accuracy of responses,” Cucchiara maintains.

Here are highlights of our interview:

THINKADVISOR: Advisors worried when robo-advisors came along: They thought they’d take over their jobs. Are financial advisors feeling the same way now about AI?

SAL CUCCHIARA: I was surprised that advisors don’t see this as a disrupter to them. But they do see the power of how it can enable them. They’re eager to learn how it can certainly empower them.

AI will disrupt a lot of industries, but the way we think about it is certainly as a productivity lift.

What do you see in the future for AI as a tech business?

This is really exciting unchartered territory in technology. There’s going to be lots of advancement in this space. The number of companies that are coming up right now to support AI is significant.

A lot of tech companies have pivoted their business models to be in support of AI and how they can support its growth.

You’ve been running an AI pilot test with your advisors. When do you expect to make this technology available to all your advisors?

Right now, we’re in the middle of piloting. This first stage is in the hands of a thousand advisors.

[But] we’re going to start ramping up. In the second half of this year, AI will be rolled out to all advisors, and that could be as early as the third quarter.

Will you be rolling it out gradually?

We could go market by market, but regardless of how we roll it out, by the end of the third quarter, it should be in the hands of all our advisors [in the U.S.].

What’s the bottom line when it comes to benefits to the advisors? Does it mean that advisors will have more time or give clients better advice?

We think it’s going to really help their productivity to serve their clients. If there are things they’re doing of an administrative nature — like searching for information on how to — this will, hopefully, eliminate a lot of that.

We know that it’s going to continually improve how the advisors service their clients, and it also will improve their productivity to take on new clients. This will only get better.

Please discuss the sources of the information that your AI provides.

We have our own infrastructure with OpenAI. We aren’t training their models on any of our intellectual property or our information. They don’t retain any of that.

That’s how we can get really comfortable with the security aspects of OpenAI.

Financial advisors, for example, are able to search and get access to information published and stored by our Global Investment Office [analysts, strategists, economists giving advice to guide portfolios] around economic, company or industry outlooks.

What specifically can the advisors do having AI other than access info?

We not only allow advisors to get access to information, but we can synthesize it in a way that gets to the heart of the question they’re asking to give them a thesis or bullet points or more detailed answers.

What else do advisors gain?

Advisors have been operating in a structured world. With AI, we can have unstructured data: We can use all the documents — whether procedural or product information or [as noted] information that comes out of our Global Investment Office.

We have hundreds of thousands of these documents. Advisors have to go and surf [the web] to get answers sometimes. Now we’re going to make it much easier to get access and synthesize the information so that they can enhance how they’re servicing the client.

We’re not only giving them the answer but mapping it back to the actual [original] document that was spun out: We’re giving them the answer but also going back to the part of the document from which we derived the answer.

This synthesized [approach] is almost like having a conversation with someone.

Instead of just reading the document, you’re also getting the specific answer [to your question].