Alphabet Google: AI Wins at Go; Loses in Other Contests

May 25, 2017

I learned that Google has “bet the farm” on smart software. When I read this statement, I wondered if ads were now going to be 100 percent artificially intelligence-ized. Humans would not be needed for conference presentations about organic search is wonderful but Adwords make everything better. Humans would not be needed to visit ad broker and ad agency people to explain that Google Adwords work much better than Facebook’s offering. Humans would not be needed to take or dodge voice calls from advertisers whose online accounts seemed to be shrinking at the same time leads from those ads were drying up. Humans. Unnecessary but for the task of figuring out what to do with fake news and other assorted online depravities.

I also learned that Google’s smart software defeated a mere human in Go. One of the photos I saw of the alleged numero uno in the life and death world of old fashioned board games showed a fairly unhappy human. “Google’s A.I. Program Rattles Chinese Go Master as It Wins Match” explains that I am using the wrong word. The now disgraced Ke Jie was rattled. Nah, I will stick with disgraced.

But the really big news about Google’s smart software appeared in “Why AI Gets the Language of Games but Sucks at Translating Languages.” The main idea is that Google’s acquisition DeepMind has algorithms which can win at Go. The rest of Alphabet Google cannot do very good translations.

The criticism of Alphabet Google’s translation system seems harsh. The write up asserts:

Neural machine translation (NMT) is Google’s response to the quest for more accurate translations. NMT technologies focus on the whole sentence instead of its components (word, phrases) in isolation by combining those components in the most naturally used manner. When AI technologies are applied to this process, NMT is also able to learn from other completed translations by analyzing their structure and how they change over time to pick up on subtleties and nuances.

I like the NMT acronym. But the write up explains that the reality of Google’s system is a bit less slick. Here’s a passage I highlighted in True Blue (a color reserved for the most informed technical statements backed by diligent research and verified data). The “diligent” part was a contest between Google’s smart software and some human translators. The “verified” part is that humans decided who translated  text better. The result? Here you go:

The reviewers stated that about 90 percent of the NMT-translated text was “grammatically awkward,” or perhaps not obviously wrong but definitely never the kind of translation produced by any educated native speaker. Many linguists and translators will be relieved by the resounding success of the humans in this latest battle against the machines. It’s inevitable that, as NMT develops further, technical content — which follows strict content guidelines and terminology — may soon be near perfectly translated without requiring much human post-editing, if any.

For now, Alphabet Google is in the game. Alphabet Google has not won the game. Just like IBM Watson, winning a “game” is different from doing real things in the real world for real people. Footsteps, but the human prey has not be killed… yet.

Stephen E Arnold, May 25, 2017

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta