Part 1 of a 3-part Series on AI in Asia
AI Arrives in Asia: Let the Games Begin!
“Asia’s business landscape is poised not only to benefit greatly from AI’s rise, but to also define it,” reports MIT’s Technology Review
CES 2020: Chattering robots, gadgets and widescreens
Black Friday is this week, which means that CES 2017 is just around the corner (January 5-8). If you go to CES 2017, remember to take lots of selfies together with all that gear surrounding you, because it’s all about to change very quickly…forever and for the better.
Much of that change will arrive from Asia, although Asia is just waking up to find it penned in, not penciled, on its to-do list.
By the time CES 2020 rolls around, which will be quicker than you think, a remarkable change will have taken place with each and every one of those oh-so-cool consumer gadgets covering the monster 2.5 million square feet of the CES show space: Every piece of electronics will be carrying on conversations with CES visitors and there won’t be a remote in the house.
Imagine millions of square feet of exhibits without a single remote anywhere in sight. It boggles.
When that LG 86-inch HD widescreen speaks to you in your own native tongue, and addresses you by name, because it already has accessed your CES registration form and then matched that info to your LinkedIn profile picture, you’ll instantly get the drift about what has transpired during the brief intermission between CES 2017 and CES 2020.
AI (artificial intelligence) will have taken over everything. To a lesser or greater degree a chunk of AI will have been infused into every electronic device in the joint.
In fact, if you had not been so cheap and paid the fee, the 32-inch Roku in your hotel room would have done much the same. CES 2020 would then have been much less of a surprise, except for the fact that being enveloped by 2.5 million square feet of AI is an experience well worth the price of entry to CES.
For many, CES 2020 will be more like Disney World: A rumpus room of wall-to-wall AI attractions so fun and so worthy of the annual jaunt to Vegas.
Chances are that you will pay the fee when you return to your hotel room that evening, and allow the TV to connect up all your devices, even to the point of replaying that morning’s Skype call with your client in Barcelona.
Halfway through the Skype replay, you’ll realize how dated your selfies from CES 2017 have become.
The question will also begin to dawn on you about how all those consumer electronics manufacturers in Korea, Japan, Taiwan and China were able to get all that AI into their electronics so quickly; and you’d be surprised to know that the factories where all those electronics were made were AI-driven smart factories.
Here in 2016, the week of Black Friday, if you had read, Asia’s AI agenda: How Asia is speeding up global artificial intelligence adoption, you’d be understandably puzzled about how Asia had come so far so fast.
That report from MIT’s Technology Review surveyed over 60 Asia-based senior executives to “gather perspective on the impact of AI and robotics on Asia’s business landscape.”
“Overwhelmingly, respondents feel technological advancements in AI and robotics will have very positive effects on most industrial sectors in Asia.” Yet fewer than 25 percent had made investment in AI until recently. Now manufacturers in Japan and Korea are “looking to infuse all manner of consumer electronics with intelligence.”
Since no manufacturer wants to be held hostage to buying AI for their electronics from 3rd parties—which is exactly what will happen if they don’t act fast—they are all acting fast. “Asia’s quest for technology innovation and sustained economic growth drives tremendous investment in AI.”
Don’t snooze on AI
AI sneaked up on most of Asia’s electronics manufacturers because they had gone to sleep on it back in the late 1980s to early 1990s, during AI’s long winter of discontent.
They slumbered still as the new millennium rolled in. By then it was mega-size supercomputers and central processing units (CPUs) for massive computations on the size of the universe or for genetics or for theoretical physics or material science; but for the lowly air conditioner, stove or refrigerator, not so much.
Quietly in the late 2000s, GPUs, graphics processing units—powerful “gamer” chips—showed themselves to be “20 to 50 times more efficient” than traditional CPUs. They quickly became AI’s chip of choice.
Artificial intelligence (AI) and its subset of machine learning and its subcategory of deep learning began to enable computers to perform tasks through of all things, learning. Neural nets, big data, analytics. How brain-like.
But isn’t that the stuff of connected cars, autonomous driving, Siri, Cortana, and the impossible task of reading my handwriting?
But what about the basics? Practical stuff? Show me something or I’m giving up on you.
Chess, Jeopardy...and then Go
Who knew and who cared “way back” in 2014 about Demis Hassabis and his teeny UK startup, DeepMind Technologies, and what possible advantage could DeepMind’s neural network offer a widescreen TV? Google knew, and Google acted; Google bought the AI tyro for $500 million.
Two years later, AI everywhere benefited when DeepMind’s AlphaGo beat up on grandmaster Lee Sedol 4–1 in the ancient, human-centered board game of Go.
The damned thing learned how to play Go better than a human!
With chess and Jeopardy already vanquished and publicly skewered, the West-developed AlphaGo and its beat down of the East’s most venerated and venerable brain game was beyond dramatic.
It hurt. It was a deep blow that somehow wounded our sense of the human brain’s unique singularity. Maybe akin to exhuming a speck of Confucius and resurrecting his memory so that it can be ransacked for more analects.
South Korean President Park Geun-hye said of Sedol’s defeat: “Thanks to the ‘AlphaGo shock’, we have learned the importance of AI before it is too late.”
Indeed.
Suddenly Asian manufacturing was awake; deep learning wasn’t just the province of the usual, high-tech suspects: Google, Baidu, Microsoft, and Facebook. It was now headed everywhere and into everything. Refrigerators, stoves, microwaves; and yes, home robots.
For example, we all know and accept the fact of life that data centers are heavy power users. When DeepMind increased the energy efficiency of Google’s data centers by 15 percent, eyes began to open very wide. By extension, could DeepMind do the same for a household’s annual power efficiency or a nation’s power grid? The answer is an obvious yes.
What then could these new deep-learning efficiencies offer up if trained in the direction of conserving water in Southeast Asia, or maximizing arable land in China or the health and welfare of aging populations in Japan and Korea?
The applications myriad. The effects immense
President Park could well have amended her “before it is too late” admission to reflect the true status of AI in Asia that the timing of its arrival couldn’t have been better. Just as Asia is ramping up the hardware side of its assault on the Fourth Industrial Revolution, along comes a game of Go to show the way forward for software.
Fortuitous for Asia, technology has at last yoked the massive computational power of Moore’s Law to the exabyte’s of data—images, video, audio, and text files everywhere on the Internet. These vast billions and billions of gigabytes, which are so indispensable for neural nets to learn from and for humans to benefit from that learning, are now ready for the first leg of a fantastic journey.
All of which prompts Technology Review’s Asia’s AI agenda to conclude: “Asia’s business landscape is poised not only to benefit greatly from AI’s rise, but to also define it.”
Heady stuff, a compliment such as “define it”, but that’s just the limitless opportunity looking back at Asia.
In short, the age of computers teaching themselves has arrived. As Jen-Hsun Huang, Nvidia’s CEO succinctly puts it: “You essentially have software writing software.”
Frank Chen, a partner at the Andreessen Horowitz venture capital firm, cuts to the business chase inherent in all of this machine knowing and doing, and software writing software: “We’re now living in an age,” Chen observes, “where it’s going to be mandatory for people building sophisticated software applications,” to hear the following: ‘Where’s your natural-language processing version?’ ‘How do I talk to your app? Because I don’t want to have to click through menus.’ ”
Most prominent in those near-term conversations will be those carried on by humans with consumer electronics: those appliances, TVs, gizmos and gadgets crammed into every corner of CES 2020.
In fact, CES 2020 may well not allow any digital appliance, TV, gizmo or gadget entry to the expo floor if it cannot intelligently converse with a human.
As for the ubiquitous remote, that fifty-year-old maze of crazy buttons and confusion, it will finally and thankfully be history. You won’t find a single one at CES 2020. They won’t be allowed in! Humanity may even refuse to make any more of them. Hooray for us!
And if chatting with humans is an expectation placed upon digital appliances, what then will be the expectations placed upon a personal or home robot?
Certainly, Pepper rolling up to you at CES 2020 with a meek, “Konnichiwa. Saikin dō?” (“Hello. What’s up?”), won’t cut it by a long shot.
It could well turn out that personal robots might have to pass an interview or take an entrance exam to be admitted to CES 2020.