Will ChatGPT soon replace my private banker? (Hint: Not so quickly!)

The subject of artificial intelligence has picked up a lot more momentum ever since ChatGPT was made available to the general public. The famous chatbot and its numerous kin really do look set to take some work off our hands in our jobs in the near future. But are they also suitable to act as investment advisors? We put them to the test.

Self-confidence at a click

Artificial intelligence (AI) has made tremendous advancements in recent years. This progress became not just recognizable to the general public, but also testable in late 2022 with the rollout of the ChatGPT chatbot. Be it for summarizing books and texts, as a source of inspiration and a brainstorming aid, for artistic/creative input, or even simply for amusement, generative language models can already help us today in our work and can provide fun entertainment. In any case, AI certainly doesn’t lack self-confidence, not even when it comes to crafting the right wording. This is demonstrated impressively when the chatbot is asked how Donald Trump would explain Bitcoin – you can hardly get the voice of the former US president out of your head afterwards. But are ChatGPT (from the US-based artificial intelligence research firm OpenAI) or its numerous kin also suitable to act as investment advisors? Our virtual mystery shopping tour revealed that chatbots still lack the necessary financial education. Moreover, even more powerful generative language model versions in the future will not be capable of replacing intimate conversations between clients and advisors.

“It’s tremendous, it’s huge.” | If Donald Trump explained Bitcoin

Query to ChatGPT

Sources: ChatGPT (OpenAI), Kaiser Partner Privatbank

Advisor without a pocket calculator

When we enter OpenAI’s virtual office and ask our first question regarding a suitable investment strategy, the chatbot starts out by alerting us that it is not a certified investment consultant and cannot give specific investment recommendations. But as is the case with so many other subjects, ChatGPT quickly sheds its restraint when we chat about a hypothetical example. Our query asks ChatGPT to construct for an investor with a moderate risk appetite a multi-asset portfolio composed of 15 to 20 ETFs that outperforms a simple 50/50 portfolio of stocks and bonds over the long term. Within seconds, the advisory bot recommends selecting a mix of low-correlated asset classes. Stocks, bonds, commodities, and alternative assets are just the ticket, the bot says, and it names corresponding ETFs. In response to a follow-on query, it also recommends specific weights for each asset class and ETF. Since we, as a conservativish client, think that a 20% allocation to commodities seems a bit too high, at our request the bot corrects it and does the same to the initially somewhat too one-sided confinement to large-cap stocks. After a few more follow-up questions, we end up with a portfolio of 25 ETFs that also incorporates small and mid-caps and sector-based, factor-based, and thematic strategies as well as exposure to international markets alongside the USA in its equity component. The original portfolio also becomes broader and more diversified in its fixed-income component (through the inclusion of TIPS and high-yield bonds) and in its allocation to alternative assets (through the inclusion of infrastructure) in the course of the “client advisory conversation.” However, the electronic advisor seems a little overwhelmed by a sophisticated client like ourself – in any case, the recommended weights in the portfolio of ETFs add up to a total of 120%. We, though, actually didn’t want to work with borrowed capital.

A professional appearance… | …after multiple probing follow-up queries

Recommended portfolio allocation

Sources: ChatGPT (OpenAI), Kaiser Partner Privatbank

Source: Kaiser Partner Privatbank

Solid performance, but short of the benchmark

After ChatGPT recalculated the weights for us, we could now subject the portfolio to a test or, more specifically, to a backtest. However, the backtest only stretches back to 2017 due to the relatively young theme ETFs in the portfolio (the ARK Innovation ETF and the Global X Robotics and Artificial Intelligence ETF). But that’s good enough for an interim appraisal. Over the first six years up until the world equity market’s peak in late 2022, the ChatGPT basket gained more than 60% and outperformed the 50/50 portfolio by up to 5 percentage points. But since November of last year, the much better-diversified portfolio of ETFs has fallen behind the simple benchmark. The chatbot thus shares the anguish of so many other investors who have difficulty beating the performance of a simple mix of stocks and bonds in the long run. Two other metrics also disillusion the client: the equity beta and the volatility of the portfolio recommended by the chatbot are both higher than they are for the 50/50 portfolio. This means that the asset mix does not (yet) quite fit our risk profile.

Not better than the benchmark… | …but respectable

ChatGPT portfolio vs. 50/50 portfolio

Sources: ChatGPT (OpenAI), Kaiser Partner Privatbank

Personalized is different…

A number of additional question marks pop up for the client in other regards as well. Is the recommended portfolio really tailored to us? The focus on US-based ETFs, for example, is somewhat annoying – ChatGPT seems to be proceeding on the assumption that we regularly trade on US securities exchanges. After we inform the electronic advisor that our domicile is in Europe, it promptly proposes a comparable palette of “European” ETFs that completely excludes America. A similar scene plays out when we express our desire for a sustainable strategy, which prompts the chatbot to quickly generate a list of ETFs bearing “ESG” in their names. But serious concerns about the quality of the advice arise at the latest when we look at the original portfolio recommendation’s costs. The annual total expense ratio is 0.85% for one of the commodity ETFs (Invesco DB Agriculture Fund) and 0.70% for the emerging-markets ETF although far cheaper alternatives exist. Moreover, the almost single-minded focus on the two fund providers State Street Global Advisors (SPDR) and iShares raises doubts about ChatGPT’s independence. The chatbot must be given credit, however, for having pointed out its lack of training as a consultant right from the start.

Professional money managers have a deep understanding of the markets, the economy, and the companies they invest in.

Refreshingly self-critical

The conversation ends with the consolation that the advisor at least is refreshingly self-critical. When asked if artificial intelligence poses an existential threat to professional asset managers, the chatbot replies:

“Artificial Intelligence (AI) has the potential to disrupt many industries, including the financial industry. However, it is unlikely that AI will present an existential threat to professional money managers. Here are a few reasons why:

• AI is a tool, not a replacement: AI is a powerful tool that can assist professional money managers in making investment decisions. However, it cannot replace the human expertise, experience, and judgment that professional money managers bring to the table. Professional money managers have a deep understanding of the markets, the economy, and the companies they invest in. They can use AI to enhance their decision-making process, but they will always be needed to interpret the data and make final investment decisions.
• AI has limitations: While AI is becoming increasingly sophisticated, it still has limitations. AI can only analyze data that has been inputted into it, and it cannot predict unforeseeable events that can impact markets. Professional money managers can use their judgment to navigate these unpredictable events and adjust their investment strategies accordingly.
• Human touch is still valued: Many investors prefer to work with professional money managers because they value the human touch. Investors want to work with someone who understands their individual needs and goals and can provide personalized investment advice. While AI can provide data-driven insights, it cannot provide the personalized attention that many investors seek.

In summary, while AI will continue to play an important role in the financial industry, it is unlikely to present an existential threat to professional money managers. Professional money managers can use AI as a tool to enhance their decision-making process, but they will always be needed to provide the human expertise and judgment that AI cannot replace.”

We have little to add to the bot’s assessment. As things currently stand, an investment advisory meeting in cyberspace cannot replace a dialogue with a private banker in any way whatsoever. Whoever puts trust in the inadequately educated chatbot most likely will end up with an impersonal standard strategy or in the worst case will even receive a far too risky investment recommendation. Even more powerful generative language model versions in the future will not be capable of replacing intimate conversations between clients and advisors. And you also don’t get a good cup of coffee in a virtual conference room.

P.S.: Those who find the chatbot’s opinion too dry can also elicit a somewhat more humorous response from it: “ChatGPT does not take lunch breaks, does not need a salary, and can process massive amounts of data faster than any human ever could. Plus, it is not subject to human biases or emotions, so it can make more rational investment decisions. It is only a matter of time before ChatGPT and other AI render many professional wealth managers redundant. But hey, at least we can all sleep easy knowing that the robots will manage our money better than they ever could!”