how to use AI as a virtual financial adviser

From chatbots and virtual assistants to fraud detection^[1] and risk management^[2], artificial intelligence (AI) is now being used in many areas of finance^[3]. But what could an AI system like ChatGPT do for your bank balance?

AI tools might seem overly complex or expensive to non-experts, but advances in natural language processing and machine learning^[4] could turn ChatGPT and similar products into virtual personal finance assistants. This would mean having an expert on hand to help you make sense of the latest financial news and data.

Staying on top of business news and financial market trends is important for making informed investment decisions and gaining an edge in the markets. Companies already use these tools to perform what finance professionals call “sentiment analysis^[5]”.

This involves analysing financial news and statements to generate insights and predictions for investors about shares and other investments. For example, Morgan Stanley’s AI models^[6] analyse a wide range of data – including news articles, social media posts and financial statements – to identify patterns and predict stock prices.

Researchers have started to explore the potential of AI tools like ChatGPT, but given how new this technology is, much of the academic research remains in the early stages. A recent preprint study^[7], the results of which have not been reviewed by other academics, tested ChatGPT’s predictions about stock market performance based on sentiment analysis of news headlines.

ChatGPT determines if a headline is good, bad or irrelevant for a firm’s stock prices and computes a score. This research found a high correlation between ChatGPT’s responses and stock market movements, showing some ability to predict the direction of returns.

AI tools may also be able to help investors decipher monetary policy announcements, providing insights into their potential effects on financial markets. Another recent preprint^[8] evaluated ChatGPT’s ability to understand what announcements from the US central bank, the Federal Reserve, might mean for financial markets.

It compared this to professional investors’ efforts to do the same. The study found that, particularly when ChatGPT models are fine tuned, they are more accurate than other machine learning models used by professionals to analyse and understand “Fedspeak”.

Monetary policy decisions, such as interest rates or asset purchase programmes, can have a big effect on financial markets^[9]. So AI’s ability to assess what central bank announcements on policy changes^[10] will mean for financial markets could provide valuable insights into the effects of these actions. This could help you make more informed investment decisions.

Tailored financial guidance

The ability to identify trends in specific market sectors^[11] could also be helpful for people seeking more tailored financial guidance.

For example, an AI tool could be used to analyse financial data, such as balance sheets and income statements, from technology companies. It could identify patterns that might indicate opportunities or problems. An investor could then adjust their portfolio, potentially increasing returns or even just helping to reduce exposure to certain risks.

In addition to analysing market trends, AI could also be used to build an investment portfolio tailored to an individual’s specific investment goals and risk tolerance^[12]. Using information on your preferences such as your current financial situation and risk attitude, for example, the AI could generate a customised portfolio that accounts for the level of return you’d like to make, but also the kinds of risks you’d like to avoid.

Your assistant, but not your only guide

AI tools show tremendous potential as personal financial assistants, but also present some challenges^[13].

There are several factors that AI tools may not be able to account for, such as unexpected events or changes in market conditions, as well as human behaviour. A tool like ChatGPT cannot fully comprehend the intricacies of human language and conversation, which can lead to responses that lack depth and insight.

There is also a need for greater transparency^[14] about how these tools make decisions. For an investor to leave their portfolio in the hands of one of these “robots”, they would need to be able to understand how, for example, it reaches its conclusions and what data it uses. Some financial planning companies already offer robo-advisors^[15] – services that use algorithms to design individual investment plans – that can also do this, but, of course, you pay a fee to the financial advisers for this.

The potential for bias^[16] in the recommendations of these tools must also be considered. ChatGPT’s training data may have underlying biases that could affect its predictions. The accuracy and reliability^[17] of ChatGPT’s predictions need careful evaluation given recent reports that it has repeated disinformation.

No single model or algorithm can predict financial market movements with complete accuracy. So AI tools like ChatGPT should only be used to supplement your own judgment, not as a replacement^[19].

While AI could be an excellent aid for investing, it is important to do your homework thoroughly about potential investments, understand and accept the right level of risk for you, and diversify your portfolio when deciding where to invest.

References

1. ^{^} fraud detection (www.ft.com)
2. ^{^} risk management (www.elibrary.imf.org)
3. ^{^} areas of finance (www.forbes.com)
4. ^{^} natural language processing and machine learning (oxsci.org)
5. ^{^} sentiment analysis (dl.acm.org)
6. ^{^} Morgan Stanley’s AI models (www.cnbc.com)
7. ^{^} study (papers.ssrn.com)
8. ^{^} Another recent preprint (papers.ssrn.com)
9. ^{^} big effect on financial markets (onlinelibrary.wiley.com)
10. ^{^} central bank announcements on policy changes (www.federalreserve.gov)
11. ^{^} identify trends in specific market sectors (www.forbes.com)
12. ^{^} individual’s specific investment goals and risk tolerance (www.businessinsider.com)
13. ^{^} present some challenges (www.forbes.com)
14. ^{^} need for greater transparency (www.sciencedirect.com)
15. ^{^} robo-advisors (www.investopedia.com)
16. ^{^} potential for bias (hbswk.hbs.edu)
17. ^{^} accuracy and reliability (www.theguardian.com)
18. ^{^} GaudiLab/Shutterstock (www.shutterstock.com)
19. ^{^} supplement your own judgment, not as a replacement (ifamagazine.com)