Iโve recently come across a lot of criticism about using AI in financial-economic analyses. Iโd like to gather pros and cons in this post after sharing my personal and professional perspective.
๐๐ฎ๐ญ๐๐๐ญ๐๐ ๐๐๐ฆ๐จ๐ง๐ข๐ณ๐๐ญ๐ข๐จ๐ง โ๏ธ
To start, I find it objectively ridiculous and outdated to demonize the use of AI in analyses for no reason.
Whether it’s isolating the key points of earnings releases, researching current or historical data on the internet, interpreting a chart, or systemizing certain variables, AI is now an established reality. Denying it would be like rejecting Word’s spell check, Excel formulas, or search engines.
Besides, AI is the driving force behind a stock market that would otherwise still be stuck in 2021โand even its detractors are silently speculating on it.
๐๐ก๐ ๐๐ข๐ฌ๐ค๐ฌ โ ๏ธ
Like all technology, AI also carries risks, specifically when it comes to economic-financial analysis, impacting both fund managers and investors:
A. Risk of Not Absorbing Knowledge from the Analysis Results: Itโs the danger of leaving behind the essence and blindly accepting the final summary provided without truly understanding it.
B. Failure to Verify Data Underlying the Summary: Occasionally, AI makes mistakes. Therefore, whatever data it references, a double-check is recommended.
C. Providing Unclear Instructions and/or Adding Biases in the Instructions: AI is very literal. If the user isnโt clear or simply shifts focus to the desired outcome, the final analysis is bound to be unreliable.
๐๐๐ฌ๐ฉ๐จ๐ง๐ฌ๐ข๐๐ข๐ฅ๐ข๐ญ๐ฒ
I believe anyone using AI for analysis and public sharing must, at the very least, fulfill the three points mentioned above to avoid improper use, which could lead to poorly informed investment decisions.
๐๐จ๐ฐ ๐ญ๐จ ๐๐จ๐ง๐๐ฎ๐๐ญ ๐ญ๐ก๐ ๐๐ง๐๐ฅ๐ฒ๐ฌ๐ข๐ฌ?๐งฉ
Hereโs the framework I propose:
1. Identify the Problem
2. Research the root causes through personal (professional or experiential) knowledge
3. Draft an analysis plan
4. Create a personal expectation regarding the completeness/logic of the draft plan (the core of the reasoning) and its output
5. Enter the problem identified in point 1 into AI, reviewing the risks listed above (A, B, C)
6. Compare the personal expectation from step 4 with the AI output in step 5
7. Identify potential discrepancies or gaps and delve deeper (moment of professional growth)
8. Finalize the analysis plan
9. Implement the final analysis plan
10. Collect feedback and, if needed, reinsert it as a new point 1, initiating a positive recursive cycle of value and knowledge generation
By the way, I donโt think Iโve discovered anything exceptional here; itโs just the classic thesis-antithesis-synthesis process applied to AI.
๐๐จ๐ง๐๐ฅ๐ฎ๐ฌ๐ข๐จ๐ง
Surprise, surprise! Like any tool, AI’s reliability depends on how itโs used. And, how itโs used or criticized depends almost entirely on the userโs intelligence.
I vote for the oldest concept in the world: let performance do the talking, as thatโs not only honest but also the main thing investors look for.
If anyone wants to add something in the comments, Iโm eager to open a constructive discussion.
๐ ๐๐ ๐
Do I incorporate AI into my trading? Yes, Iโm a proponent of algorithmic trading and obsessed with perfecting it.
Do I use AI in my analyses? Yes, I use AI as a thinking partner (challenging) and learning source (mentoring).
Do I use AI to write content faster and more accurately? Yes, my native language is Italian, and I use AI for English
Do I use AI to gain popularity on eToro? Yes, because it lets me be faster and more accurate, a win-win for me and my followers.
Am I sure Iโve always used AI correctly? No, because Iโm human, but I strive to improve
