The author argues that efficiencies of generative AI will produce demand for growth of human-led advice, just as changes wrought by so-called "robo-advisors" did a decade ago.

The following article addresses a question that must be front of mind for many advisors: will AI take my job or be my indispensable assistant? How that question is answered will play a large role in shaping how the wealth sector looks in the next decade and more. 

To get a handle on how AI plays into the role of the advisor, we carry this guest article from Andrew J Evans, chief executive and founder of Rossby Financial, a US RIA. The editors are pleased to share these views; the usual editorial disclaimers apply. To jump into the conversation, email tom.burroughes@wealthbriefing.com

If you’re a financial advisor, generative artificial intelligence won’t put you out of work. But a human financial advisor using AI tools intelligently may very well out-compete you.

I would argue that AI is the first major technological innovation since robo-advisors that seems poised to disrupt every part of our industry. It’s easy to open up ChatGPT, prompt it to generate sophisticated-looking answers, and then think, “Wait, is my job in danger?”

When I say AI, I mean generative AI built on large language model (LLM) technology, the kind that underpins ChatGPT, Claude, Gemini, and other platforms. I’m not talking about AI in the sense of more classic, rules-based algorithms that have actually been a part of our work and lives for years now – that is nothing new and has been integrated into our advisory practices since our industry, and the business world at large, started sharing data over the internet nearly 30 years ago. But when it comes to generative AI, advisors have a hard time accepting it because just about anything can be (and has been) marketed as “AI-powered” these days. 

The widespread fear in our industry is that generative AI may soon do all of our work, from start to finish. A recent research paper from MIT Sloan studied what a fully-automated AI advisor would have to do in order to serve clients as well as, if not better than, human professionals. The biggest hurdle generative AI has to clear is that LLMs cannot understand entities like people. At a high level, these models look at very large data sets to build associations and develop rules that determine what the most likely answer might be.

It’s true that AI often produces excellent results. LLMs are quite good at summarizing text. They are less good at math, or at representing text in wholly AI-generated images. But in general, this technology looks at what users tell it and then delivers what the LLM calculates to be a likely answer. But there is no human-level reasoning or empathy under the hood. LLMs inherit the biases and flaws of the data they ingest without really understanding it.

The MIT research suggests solutions: a large cache of (likely proprietary) financial data, paired with a sort of rapidly-computed evolutionary simulation that would allow generative AI to “teach” itself concepts of empathy and ethics from first principles. But here we run into the other limitation of generative AI as a total replacement for human work: it is costly to develop and use, both in terms of money and energy. Ami Badani, CMO of chip design firm Arm Holdings, warned that generative AI’s power demands could account for 25 per cent of all US electricity by 2030. When OpenAI CEO Sam Altman says future development would need fusion power to sustain itself, you can guess the costs are steep. Even if truly empathetic, accurate, unbiased, self-aware AI were to be created with this technology, advisory firms would have to weigh their productivity against the cost of keeping them plugged in.

For these reasons, I strongly doubt generative AI will replace human advisors. The AI hype cycle puts a lot of stock in what the technology might one day be able to do. The good news, for us, is that the technology as it exists today can already add plenty of value to an advisor’s work. As an RIA, our firm uses Boosted.ai and Alphathena to augment our investment management capabilities. Generative AI speeds up our operations with rapid first drafts of key documentation. These tools work because they understand that generative AI is a powerful support tool, and not a replacement for every aspect of our work. Autopilot did not eliminate the need for pilots – someone still has to land the plane.

Instead of replacing us, I find it more likely that these business-supporting AI tools will become additive for advisors aiming to grow their businesses and improve their clients’ experience. Firms will need to hire or outsource people who know how to maximize the generative AI components of their tech stacks. I think the comparison to robo-advisors is apt. Did robos replace us? No. Many of them have been commoditized as business segmentation tools, or quietly shuttered. But you can’t deny the impact they had on our profession.

Robos arguably kicked off a trend of fee compression that forced financial advisors to get serious about financial planning, behavioral finance, more sophisticated investment solutions, and a growing lineup of ancillary services that add value to the advisor-client relationship. I believe the efficiencies of generative AI will trigger a similar wave of growth for human-led advice, not its extinction.