AI, Aiyaiiii: The Howling Gets Louder
September 26, 2014
I found another artificial intelligence write up in my in box this morning. (Believe me, it was more useful than the whining from Lucid Works’ PR firm Launchsquad).
The AI article is “Inside Baidu’s Plan to Beat Google by Taking Search Out of the Text Era.” I love it when folks younger than I are ever so excited about the “end of text.”
The write up focuses on a former Xoogler, Andrew Ng. Obviously like the departures of Babak Parviz (aka Amirparviz, Parvis, and Amir Parviz) and the wonderfully named Sebastian Thrun (no akas in my files, alas), Xoogler Ng is moving on to greener pastures. Well, at least less interpersonally conflicted acreage I assume. No suicide attempts and six month stays in the Stanford Hospital, please.)
The write up focuses on smart software. Now in case you have been living in a country without Internet access, “AI” means artificial intelligence. Smart software used to be the optimal way to eliminate some of the drudgery associated with many tasks. For better or worse, we are back in an era when software is wrapping itself in “artificial intelligence.” When I was in college, I identified several individuals with artificial intelligence. In our bubblicious era, AI means software that thinks, understands, decides, and—wait for it—can figure out what pictures mean. Well, sort of.
Xoogler Ng has joined Baidu, one of Google’s competitors. I assume this does not thrill Messrs. Brin and Page, who want to think about Loon Balloons and Amazon getting into the Parviz-Amir-Parvis business.
Xoogler Ng was working on the Google Brain. Think smart software to generate ad revenue. Xoogler Ng says:
“With the Google Brain project we made the decision to build deep learning processes on top of Google’s existing infrastructure,” he says. “What we’re doing at Baidu is seizing the opportunity to build the next generation of deep learning infrastructure. This time we’re building everything from the ground up using a 2014-base GPU infrastructure.”
Yep, this is a nice way to explain that Google’s ageing architecture is not likely to change soon. This is a tough factoid to swallow because Google is just so darned wonderful. Sort of.
Xoogler Ng also says what few self appointed experts wish to state; to wit:
“Deep learning is a very capital-intensive area, and it’s rare to find a company with both the necessary resources and a company structure where things can get done without having to pass through too many channels and committee meetings. That’s essential for an immature technology like this.”
Yep, immature. It took printing 300 years to shake off its baby crib behaviors. I am not sure if smart software will move too much more rapidly.
Xoogler Ng observes:
Most of these users [folks who rely on mobile devices] haven’t organically learned how to use text-based search as it’s evolved from Ask Jeeves to DuckDuckGo over the past several years. That presents an opportunity to re-think basic assumptions about search, and it extends beyond developing markets. “Text input is certainly useful, but images and speech are a much more natural way for humans to express their queries,” Ng says. “Infants learn to see and speak well before they learn to type. The same is true of human evolution–we’ve had spoken language for a long time, compared to written language, which is a relatively recent development.”
Oh, oh. Another bad news factoid for Google. Text-based ads generate more than 90 percent of the firm’s revenues.
Xoogler Ng adds:
In many cases, text-based search is not ideal for finding information. For instance, if you’re out shopping and spot a handbag you might like, it is far better to take a picture than to try and describe it in words. The same is often true if you see a flower or animal species that you would like to identify.
Given the limitations of von Neumann computing, what can Baidu do? Well, it appears that monitoring of user behavior and image recognition will be one of the first smart things Baidu will explore. After images, then it will be speech. My thought is that speech recognition may be more doable than image recognition, but I am dumb wetware equipped, not smart software equipped.
Xoogler Ng seems to be reasonably confident in deep representation learning. One hopes that his bosses at Baidu are patient.
Stephen E Arnold, September 26, 2014
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
