Would you trust a piece of programming, an algorithm, with your hard-earned savings to make them grow? The AI Powered Equity Exchange-Traded Fund that just launched in the US will let EquBot LLC, an artificial intelligence (AI) entity, pick stocks. EquBot has used IBM’s Watson AI technology to create a set of rules to analyse upto 10 years of data on thousands of stocks. The breadth of data-sources it pours over should amaze you—market sentiments, regulatory filings, news articles and even social media posts.
It then ranks the stocks on the basis of a probability forecast of its profitability in the current conditions. It banks on information, of which there has been an explosion in the last couple of decades, thanks to the internet. You would agree that a machine would far comprehensively, efficiently and objectively process that information than any investment advisor ever could. Hence, index-tracking ETFs have become the flavour that forward-looking fund managers are seeking. Reuters reports, starting 2019, the Chartered Financial Analyst Exam will add questions on AI, automated investment services and data-mining from uncoventional sources.
All this sounds terribly exciting. EquBot, if it is self-teaching, will become progressively sharper—and there would be many who are already imagining “fatter than ever before” returns. All that is good, but a word of caution. Whatever inputs AI and self-teaching systems use, at least as of now, are human generated. It could be garbage in, garbage out—unless EquBot can parse information for genuineness. For instance, trawling the depths of social media, EquBot is likely to net what is now infamous as fake news.
With some concept of criticality in the picture, AI may factor in to its calculations information that is fake but has spread wildly—sometimes even mainstream media picks such misinformation up. While many are enthusiastic, there are others who would rather make haste slowly. While IDFC Mutual Fund is using robot guidance for managing investors’ money, it has a fund manager who can override the AI-derived signals and take the final call. Man and machine, both may go wrong—the latter because of what the former feeds it. The choice of who to go with, at present, is a difficult one to make.