The AI makes it easier to analyze the CEO’s speech patterns and tone. Does he really mean what he is saying?

In 2020 an algorithm that was designed to spot hidden clues in spoken words pinpointed hesitant uncertainty in the words of IT executives discussing supply chain outages and a shortage of semiconductor chips. Though they maintained that all was fine, the algo didn’t think so. (Reuters)

According to language pattern software specialist Evan Schnidman, the tone of the executives didn’t sync with the positive content of their remarks. Sure enough, their words proved hollow when a few months later the auto industry was facing the prospect of cutting production due to a shortage of chips.

NLP a new tool for fund managers

Reuters interviewed 11 fund managers who were using or testing NLP technology to test speeches by CEOs for inconsistencies. Their objective: to overcome the limitations of financial and corporate documents that had already been in the public domain and heavily analyzed, and instead to gain an advantage by using NLP, a new tool.

Computer-driven Man Group (LON: MAN) confirmed to Reuters that the investment management firm was focused on NLP as a major research area.

NLP technology analyzes the audible tones, cadence and emphases of spoken words alongside phraseology. Unstructured data such as calls with analysts, unscripted Q&As, and media interviews is NLP’s favorite hunting ground.

It also can parse the transcripts of speeches and interviews to glean valuable insights.

A model employed by Bank of America (NYSE: BAC) pores through earnings calls for tell-tale phrases such as “cost cutting” and “cash burn” – which can often suggest the potential for future default.

Indeed, back-testing did show a high correlation between the use of such jargon and the probability of default.

However, companies, CEOs and their PR departments are getting smarter too, and increasingly they take recourse to well-scripted media and earnings calls that do not give away much – even to NLPs.

Related Story: Wu Dao 2.0, China’s Fire-Breathing New AI System, Is 10X Bigger Than GPT-3