With constant content generation comes the necessity for automated analysis systems. This is especially true in the high frequency trading (HFT) of stock. In this week’s VentureBeat, Lexalytics CEO Jeff Catlin explores the potential pitfalls of utilizing AI for speed instead of accuracy in stock market transactions. What happens when this AI is presented with a gag headline, like Tesla’s April Fools announcement about their brand new “Tesla Watch”? “[Unstructured data analysis] is still a relatively unsophisticated field that can deliver unsophisticated results,” says Catlin in response to the market confusion caused by the Tesla announcement.
Catlin is wary of an interesting potential pathological case like a feedback loop forming between automatically generated unstructured data (like press releases) and the AI that analyses it. “It’s probably already happened at some level…” Continues Catlin, “Automated text generation racing to keep up with rapidly fluctuating conditions caused by automated trading based on text mining of automatically generated text.” But, pathological feedback loops aside, these robots will improve in the foreseeable future. “We are likely 5 to 10 years from something resembling a complete contextual understanding,” he says. “One that has the niceties in it to understand humor and sarcasm, and that trees grow up (not down).”
If you’re curious to learn more and would like some tricks on how to protect your company from erroneous and potentially harmful text mining then head over to Venture beat and check out Jeff’s article here.