Can AI Replace a Human Adviser During Market Volatility?

Can AI Replace a Human Adviser?

Artificial intelligence (AI) tools like ChatGPT are increasingly capable of analysing financial data, generating reports, and answering complex questions in seconds. But the real question is: can AI truly replace a human financial adviser?

In this series, ‘Can AI Replace a Human Adviser?’, we examine real-life case studies of clients navigating major life transitions and put AI to the test against a Providend Client Adviser. As these stories unfold, we reveal what AI gets right, what it gets dangerously wrong, and why human advice may still matter more than you think.

Case Study:

Phillip, aged 50, is a senior executive at a multinational corporation, earning a take-home annual income of $360,000 and annual expenses of $200,000. He is married, and his wife is a stay-at-home mum. They have two children aged 16 and 18, and own a private condominium valued at $4 million, with an outstanding mortgage of $1.2 million. Phillip has been investing regularly, with 50% of his portfolio worth $1.5M concentrated in Nvidia stocks, alongside a diversified mix of ETFs, unit trusts, SSB, FDs, and bonds.

Phillip wants to safeguard his family’s financial security while staying on track for retirement in 10 years’ time and funding his children’s tertiary education in Australia. He hopes to navigate market volatility without making hasty decisions, maintain contributions to his investments and retirement accounts, and ensure that short-term market swings do not derail his long-term plans.

Recent market turbulence, especially huge volatility in tech stocks like Nvidia, has caused significant swings in Phillip’s portfolio. While his emergency fund can cover 12 months of expenses, he is uneasy seeing the high concentration in one stock affect his overall portfolio. He is considering whether to hold, sell, or rebalance, and whether his current risk exposure aligns with his retirement and wealth goals.

Emotionally, Phillip feels anxious and stressed about the potential impact on his retirement timeline, his children’s upcoming university expenses, and mortgage commitments. Despite his disciplined approach to saving and investing, the volatility has triggered fears of making emotionally-driven decisions. He recently sold a portion of his Nvidia stocks as it dipped, and entirely missed out on the rebound the next day, making a significant loss.

His question is: “How should I respond to sharp market declines, especially with concentrated tech holdings, to protect my wealth, maintain progress toward my goals, and avoid making rash investment decisions?”

To find out, we posed his exact situation to ChatGPT and asked it to act as their financial adviser. Our Client Adviser, Annette, reviewed the AI-generated plan, analysed its conclusions, and compared them against what a real human adviser would do. Here is what she found.

The Pluses:

ChatGPT listed Phillip’s current position clearly. Income and expenses, assets and liabilities, and emergency funds were listed alongside key goals such as funding his children’s education in Australia over the next two to four years and retirement in 10 years.
ChatGPT pointed out the risk of concentration and greater volatility from holding a single stock.
It acknowledged the anxiety and stress mentioned by Phillip and highlighted that the anxiety felt is a signal that Phillip’s risk tolerance is lower than the actual portfolio risk.
It offered a framework on how to respond to market swings that comprised:
- Setting clear rules before reacting, such as determining what level of single stock or sector concentration is acceptable and aligning it to one’s risk profile and retirement goals, and pausing for 24-48 hours before making any sell decisions.
- Rebalancing gradually by staggering the sale of Nvidia and reallocating the proceeds to a diversified portfolio of ETFs, bonds, or defensive assets, as opposed to panic selling when markets are down.
- Deploying an annual surplus of $160k systematically through regular contributions to retirement and education accounts.
- Hedging for short-term needs by keeping the kids’ education funds in SGD cash funds or short-term bonds, given that the money is required in the next two to four years.
- Reducing stress by refraining from daily checks of the portfolio and focusing on long-term goals rather than short-term noise.
It detailed a goal-specific, number-based action plan with a suggested portfolio allocation that includes a reduction of the Nvidia holdings to 15-20% of the total portfolio.

The Misses:

ChatGPT assumed that the annual surplus of $160k is available to be deployed towards the education and retirement goals. It did not make any assumptions or calculations regarding the income tax payable and current mortgage payments. If we assume that Phillip’s annual gross salary is $400k, his income tax payable, assuming zero tax relief, is approx. $60k per year. As for his annual mortgage liability, if we assume that the mortgage is payable till age 65, and with interest rates ranging between 1.8% to 3%, it works out to be approx. $90k to $100k per year. Based on his reported take-home annual salary of $200k, Phillip’s living expenses would be only $40k for a family of four. This is unrealistically low. If Phillip’s annual expenses are substantially higher, his plan to retire at 60 may not be achievable.
Phillip did not mention the amount of retirement income that he is comfortable with. This is an important fact to establish before a risk profile and asset allocation can be determined. If Phillip is looking to retire with $200k a year from age 60, this may not be achievable, especially if his current annual expenses are closer to $300k, including income tax and mortgage payments.
Another parameter that ChatGPT did not specify is the duration of Phillip’s retirement. Are we planning till age 80, 90 or 100? This has a significant bearing on how much capital is required and the required rate of return for the retirement funds.
While ChatGPT suggested that Phillip reduce his Nvidia holdings to 15% or 20% of his overall portfolio, it did not explain how it arrived at this percentage and why it is optimal.
It further suggested that Phillip sell off Nvidia stock gradually over 6 to 12 months and reallocate the proceeds to buy into a low-risk fund for the children’s education and the rest into medium-term funds for his retirement. In practice, this is not easy to execute as it requires both selling and buying the stock systematically. Phillip could end up selling and not following through with the buying. Managing one’s emotions in the process of selling and buying will also be very challenging, especially when the market is volatile. The recommendation to set pre-set rules for panic moments did not come with further elaboration on what rules are suitable and how Phillip would be able to stay committed to the course of action.
ChatGPT mentioned that Phillip should consider having $250k of medium-term funds but did not explain what these are and how these funds contribute towards his overall retirement portfolio.
With a plethora of funds out there, it is very difficult for an individual investor to decide which diversified funds to select and to have the confidence that the funds will provide the returns required in the future.

What I Would Do Differently

At Providend, we start by having an in-depth conversation with each client. We explore how their values, experiences with money, and approach to investing influence their decisions, while identifying what’s most important and non-negotiable to them. We also take time to listen to their concerns and uncover any obstacles that may be holding them back from living the life they envision. Ultimately, our goal is to align their financial resources in a way that allows them to have peace of mind and live a meaningful life.

1. Retirement

I’d start by asking Phillip if he has discussed retirement plans with his spouse, what he hopes to do in retirement, and how long he would like his retirement income to last. This would help to determine how much retirement income is needed to support his lifestyle.

2. Children’s education

I’m curious about the decision to send his children to Australia for their tertiary education versus studying locally, and whether the kids have decided on a field of study.

3. Home

For planning purposes, it would be useful to know if Phillip would like his mortgage to be paid off upon retirement and if he has any plans to right-size his home at a later stage.

4. Any other goals

I’d like to find out if there are any other short- or long-term goals that Phillip might have that could potentially impact his finances.

5. Insurance Coverage

I’d help Phillip decide on the level of insurance coverage that he needs in the event of death, critical illness, disability, large hospital and surgical bills, and long-term care, based on his care preferences. By comparing them against his existing coverage, we can determine if there are gaps in the coverage and what it would cost to plug the shortfall.

6. Ability, need and willingness to take risk

I’d understand Phillip’s risk appetite and triangulate this against his need for returns and ability to take risk. This will determine the risk level that is appropriate for Phillip’s retirement goal.

7. Clarifying the numbers

The key numbers that I’d like to clarify with Phillip are his income tax payable, mortgage and length of payment, breakdown of household and living expenses, and any potential changes to the expenses in the next few years.

8. Planning for a reliable income stream for life via RetireWell™

In order to generate a reliable income stream throughout one’s retirement, Providend uses a proprietary framework, RetireWell™, that considers both the accumulation phase of the journey as well as the spending phase in retirement. The framework seeks to mitigate the 5 key risks of longevity, inflation, rising healthcare costs, market risk and sequence of return risk, and the risk of over- or underspending.

While ChatGPT can give us a seemingly good plan in seconds, upon closer inspection, it falls short on the details. In a recent The New York Times article, Feeling ‘Amateur’ at Retirement Planning, They Asked AI for Help author Kailyn Rhone wrote that “Despite the perceived benefits, more than half of Americans who acted on financial advice from generative AI told Credit Karma they ultimately made a poor financial decision or a mistake”.

All too often, the guidance offered by AI tends to be overly generic, as illustrated in Phillip’s case. Its wisdom comes from the models that it is trained on and depends on the quality of the prompts or questions that we provide. This can be challenging for those who know less about the topic.

While an AI-generated plan can serve as a useful starting point, a human adviser provides the essential partnership needed to refine and strengthen that plan. They bring not only professional judgment but also the ability to clarify motives, challenge assumptions, uncover blind spots, understand trade-offs, and help foster discipline and accountability. Most importantly, a human adviser can journey with you through life’s ups and downs to ensure that your future remains financially secure.

This is an original article written by Annette Lee, Client Adviser at Providend, the first fee-only wealth advisory firm in Southeast Asia and a leading wealth advisory firm in Asia.

To receive first-hand wealth insights from our team of experts, we invite you to subscribe to our weekly newsletter.

Through deep conversations with our advisers, you will gain clarity on what matters most in life and what needs to be done to live a good life, both financially and non-financially.

We do not charge a fee at the first consultation meeting. If you would like an honest second opinion on your current investment portfolio, financial and/or retirement plan, make an appointment with us today.

GET IN TOUCH