Just to keep you up to speed, everything is on schedule or ahead of schedule.
In the race toward a superintelligence or ubiquitous AI. If you read this blog or you are paying attention at any level, then you know the fundamentals of AI. But for those of you who don’t here are the basics. Artificial Intelligence comes from processing and analyzing data. Big data. Then programmers feed a gazillion linked-up computers (CPUs) with algorithms that can sort this data and make predictions. This process is what is at work when the Google search engine makes suggestions concerning what you are about to key into the search field. These are called predictive algorithms. If you want to look at pictures of cats, then someone has to task the CPUs with learning what a cat looks like as opposed to a hamster, then scour the Internet for pictures of cats and deliver them to your search. The process of teaching the machine what a cat looks like is called machine learning. There is also an algorithm that watches your online behavior. That’s why, after checking out sunglasses online, you start to see a plethora of ads for sunglasses on just about every page you visit. Similar algorithms can predict where you will drive to today, and when you are likely to return home. There is AI that knows your exercise habits and a ton of other physiological data about you, especially when you’re sharing your Fitbit or other wearable data with the Cloud. Insurance companies extremely interested in this data, so that it can give discounts to “healthy” people and penalize the not so healthy. Someday they might also monitor other “behaviors” that they deem to be not in your best interests (or theirs). Someday, especially if we have a “single-payer” health care system (aka government healthcare), this data may be required before you are insured. Before we go too far into the dark side (which is vast and deep), AI can also search all the cells in your body and identify which ones are dangerous, and target them for elimination. AI can analyze a whole host of things that humans could overlook. It can put together predictions that could save your life.
Now, with all that AI background behind us, this past week something called Google I/O went down. WIRED calls it Google’s annual State-of-the-Union address. There, Sundar Pichai unveiled something called TPU 2.0 or Cloud TPU. This is something of a breakthrough, because, in the past, the AI process that I just described, even though lighting fast and almost transparent, required all those CPUs, a ton of space (server farms), and gobs of electricity. Now, Google (and others) are packing this processing into chips. These are proprietary to Google. According to WIRED,
“This new processor is a unique creation designed to both train and execute deep neural networks—machine learning systems behind the rapid evolution of everything from image and speech recognition to automated translation to robotics…
…says Chris Nicholson, the CEO, and founder of a deep learning startup called Skymind. “Google is trying to do something better than Amazon—and I hope it really is better. That will mean the whole market will start moving faster.”
Funny, I was just thinking that the market is not moving fast enough. I can hardly wait until we have a Skymind.
“Along those lines, Google has already said that it will offer free access to researchers willing to share their research with the world at large. That’s good for the world’s AI researchers. And it’s good for Google.”
Is it good for us?
This sets up another discussion (in 3 weeks) about a rather absurd opinion piece in WIRED about why we should have an AI as President. These things start out as absurd, but sometimes don’t stay that way.