iFlytek is a Chinese technology company and world leader in Artificial Intelligence, particularly in voice recognition. In September 2018 during a tech conference in Shanghai, the company was accused of disguising human translation as machine learning. Bill Wang, an interpreter, was conducting a live translation at the conference. iFlytek used his translations as live subtitles on a screen giving the appearance that the translated output was produced by their A.I system, not Wang. At best, it was an accident. At worst, iFlytek was misleading the public about the quality of its A.I. in exchange for good PR.
A closer look at other leaders, such as Google and Microsoft, reveals a similar trend. Appen, an Australian A.I. company that provides manual checks and balances to algorithms, has eight of the top 10 tech giants, such as Apple, Google, Microsoft and Facebook. These manual checks and balances, called results validation, requires manual labor. Behind it are over 500 full-time employees and more than one million freelancers. This validation business accounts for 86% of Appen’s revenue. What does all this mean? While relying on manual labor, Tech companies have benefited from the perceptions that they have built sophisticated A.I. They are less “artificial” than Silicon Valley wants us to believe.