On Friday, August 21, 2020 stories broke on the Bloomberg & Financial Times websites worth reading. The stories (two in total) are packed with information anyone following the development of consumer products built with natural language processing (NLP) and/or natural language understanding (NLU) tech should carefully consider. When combined, the data points to big obstacles to further improvement in NLP & NLU tech. The difficulty of overcoming them any time soon looks almost insurmountable. The story about Amazon and its decision to “pause hiring [for its Alexa unit] amid pressure to make money” validates this conclusion.
In a story titled “Amazon Alexa Unit Pauses Hiring Amid Pressure to Make Money” Priya Anand reports for Bloomberg:
“Alexa group job postings on Amazon’s website are down 43% since March 16, according to a Bloomberg review, compared with a 37% decrease overall in openings.”
Anand also reports Amazon officially denied the hiring freeze. Given Jeff Bezos’ one pointed focus on developing “right” solutions, regardless of cost, if Anand’s story is correct, the inescapable conclusion has to be management has made a decision there is little further to be gained from hiring more NLP & NLU experts for the Alexa business.
Another story, also published on the same Friday, this time by the Financial Times, adds credibility to this notion. This story is titled “Boom time for the smart speaker“. The writer is Adrian Justins. Students of Dr. Clayton Christensen’s theories of innovation and disruption of markets will be quick to see Harman Kardon’s $2,500 Citation Towers as an upmarket entry. Harman Kardon has not been leading in NLP & NLU technology, but one of its partners, Microsoft, has certainly tried very hard to become a leading player in the market.
As Dr. Christensen makes clear in his book, “The Innovator’s Dilemma“, when manufacturers of disk drives moved upmarket they were abandoning the low end business over to competitors. These manufacturers took their steps up to pricier products when they concluded they couldn’t innovate, further, to defend against competitors with a cheaper solution for resource constrained customers.
If both of these stories are right, then concluding NLP & NLU technology is at a wall makes sense. Don’t be surprised if there are few, if any breakthroughs in this market segment any time soon.
Like this story? Here is a link to another you may find helpful?