Title: The Challenges of Using ChatGPT for Stock-Picking: A Cautionary Tale
Chatbots have become increasingly popular in various aspects of our lives, including investing and stock analysis. However, relying solely on these AI-driven tools may not always yield accurate results. In a recent experiment, I discovered the limitations of ChatGPT, particularly when it comes to providing reliable stock data and analysis.
For instance, when I asked both Bing and Google for the compounded annual growth rate of several locally listed stocks, the formula used was correct, but the data output was nonsensical. Surprisingly, despite having access to share price data, both platforms provided random price points that had no resemblance to reality.
Even when I sought information on the balance sheet and debt situation of different locally listed companies, the responses were nothing but gobbledygook. It was disheartening to see that even a simple query like how much debt does British American Tobacco have and what’s the expiry profile proved to be a challenge for these AI models.
While it is true that Bing’s data only extends up until September 2021, the inaccuracies were not limited to that timeframe. Bing had started using its own search data for more up-to-date information; however, this feature has been discontinued at the time of writing. Despite being my preferred model out of the two, even Bing fell short in providing reliable and accurate answers.
Nevertheless, I did find some usefulness in Bing. It served as a valuable tool for generating investment ideas, particularly in the electric vehicle (EV) sector. By seeking a list of exchange-traded funds focused on this theme, I obtained practical insights. Additionally, diving deeper, I explored research reports on EVs, gaining a better understanding of the key players in the industry and the companies fueling its growth. This served as a solid starting point for further research and analysis.
As a result, I continue to use Bing cautiously, treating it more like an advanced search engine rather than a smart AI model. It requires double-checking and scrutiny of all answers provided. Undoubtedly, this process is time-consuming, but failing to do so renders the entire exercise useless.
In conclusion, the use of ChatGPT for stock-picking poses certain limitations and challenges. While helpful in generating investment theses and providing initial insights, caution must be exercised when relying on these AI models for accurate and reliable data. It is imperative to approach their responses as a starting point for further research and analysis, ensuring all information is meticulously verified. In doing so, we can overcome the perils of ChatGPT and make more informed investment decisions.
