Ask an AI platform like ChatGPT whether a stock is a good investment, what is the best ETF or how to reduce risk in your portfolio, and an answer arrives almost instantly. It is neatly structured, appears reasoned, and is accompanied by a range of supposedly reliable sources. And you can get it for free.

AI is rapidly becoming an important tool in the investor’s kit. The likes of ChatGPT, Claude or Grok (AI platforms known as Large Language Models or LLMs) can rapidly digest financial reports, explain investment concepts, compare funds, examine economic scenarios and process quantities of information that are beyond the capability of any human.

Used well, it could make investors better informed and unlock opportunities. Used poorly, it could make them confidently wrong.
 

Every signal. All at once.

The incredible efficiency AI can provide is a clear benefit to investors in terms of time, effort, and cost savings. The sheer volume of information and data that can now be analysed, distilled and turned into comprehensible, actionable responses gives the everyday investor access to a level of expertise and processing power that was previously unimaginable.

The possibilities are endless. For example, investors can:

• Ask AI to research an investment before putting money into it.
• Compare investment strategies according to your specific objectives.
• Identify how different ETFs overlap to uncover stock concentrations.
• Reverse-engineer what a share price implies about future earnings growth.
• Build an explicit sell discipline so that emotion is removed from your decisions.

Many investors are already embedding AI into their investment process, using it as a co-pilot, and an always-on adviser to inform and guide portfolio management.  

Is this the end for the financial adviser? Probably not.

When a group of academics from MIT Sloan conducted research into the financial advice LLMs were providing, they were surprised. The financial guidance produced by LLMs was better than they had expected, regardless of whether the prompts came from ordinary users or academics.

The models tended to encourage higher savings, greater participation in the sharemarket, better portfolio diversification, and levels of risk that were more appropriate to an investor's age.

But the LLMs struggled with nuance. The quality of the advice depended heavily on the quality of the prompt. This presents a quandary. To access quality financial outputs from AI, the user must already have sufficient financial knowledge to not only know what’s important and how to ask it but also know enough to interrogate the answers and check their veracity.
 

Risks

There is no absolute way to invest, and no single investment strategy or piece of advice will suit every investor.

Two intelligent investors can examine the same company, study the same financial statements and reach opposite conclusions. One may see an undervalued business. Another may see a business whose best years are behind it.

However, regardless of the complexity of the question, LLMs deliver clear, confident answers. They present information as correct, even when it is not.

It’s called a “hallucination” when an AI model generates information that appears plausible but is inaccurate or invented. But with investing, there is an added complication. An answer can contain completely accurate information and still lead to a poor investment decision.

Across the LLM landscape, the most cited source is Reddit - a website built on user-generated content, opinion and community voting. According to Contently’s analysis, LinkedIn, YouTube, Wikipedia, and “Forbes and editorial publications” round out the top five most cited domains. Most of these sites would not normally make the list for providing accurate, objective investment information.

The bigger risk, however, is not that AI gets it wrong, but that we become less inclined to question it.

Wharton academics recently released a paper that brings Nobel laureate Daniel Kahneman’s work in Thinking, Fast and Slow into the AI era. The study introduced the concept of ‘cognitive surrender’, which is the tendency of humans to accept AI-generated answers with minimal critical oversight.

The researchers found that even when the AI was giving wrong answers roughly half the time, participants’ confidence in their responses went up.

In markets, where certainty is already scarce, that confidence can be a particular risk.
 

Where machines may have the edge

Humans, of course, also have shortcomings when it comes to investing.

Our behavioural biases are well documented. We have tendencies to put more value on recent performance (recency bias), seek information that confirms what we already believe (confirmation bias), and can fall in love with a strategy or stock, resulting in not selling out despite it being in our best interests.  

An algorithm has no sentimental attachment to the shares it bought last year. It does not panic because the market has fallen 5% in a day.

This is one of the reasons ETFs have become so popular. Being rules-based, ETFs provide access to a range of investment strategies – many previously only accessible by institutional investors – while removing emotion from the management of the fund. AI extends that capability.
 

Conclusion

The next evolution of investing is harnessing this computing power, data analysis and machine learning to build dynamic portfolios. This is not about asking a standard LLM which stocks to buy. It is about institutional-grade, purpose-built AI models designed to identify stocks according to defined investment learning, parameters, and adapting as they process new information.

And this technology is already on ASX, with the first AI-powered ETF launched this year. Like every other sector, AI will continue to transform the way we invest.

But AI alone will not be a differentiator. Just as with LLMs, the quality of the output is only as good as the inputs, frameworks and expertise behind it. The investment edge will therefore belong not simply to those with access to the most powerful models, but to those who combine them with the deepest investment expertise. 
 

From ASX

The ASX Investment Products monthly report provides information on the latest ETFs and other listed funds.