Archive for August, 2015

Artificial Intelligence Post Number 14

August 30, 2015

Here is a book trailer for my new book:

The book is also available on Amazon (at least the Kindle version – the hard copy should be there in a few days).


Artificial Intelligence Post Number 13

August 28, 2015

I have mentioned that semi-automatic or automatic driving cars will be one of the first applications of low-level artificial intelligence. The interest in this area just increased with recent highway death and injury data. Per the National Safety Council, the costs of deaths, injuries, and property damage are up 24% versus a year ago, to $152 billion for the first 6 months. To put this in perspective, the projected 2015 US deficit is $486 billion. If the year continues like the first six months, automobile accidents will cost the equivalent of 2/3 of our annual US deficit. This trend in increased highway deaths is unlikely to reverse given that “experts” believe that the causes are cheaper gas (which encourages people to drive more) along with people using phones and texting while driving, which we do not seem able (or willing) to stop. Note that this upward trend is happening while cars are getting measurably safer! The main problem is the driver, not the car!

Most car companies are developing various levels of computer and sensor driver aids, and even non-automotive companies like Google and Apple are investing heavily. But the company that is perhaps leading in this area is the Israel company Mobileye, which expects to have completely automated car technology within three years. Per their website, their EyeQ chip and algorithms are already in approximately 5 million vehicles. Although this EyeQ chip is not as unique as the IBM TrueNorth chip, it does enable what Mobileye calls parallel routing, up to four paths, between 4 masters and 4 slave ports. Autonomous driving is planned for launch in 2016. This will enable hands-free capable driving at highway speeds and in congested traffic. By 2018 they plan to add country roads and city traffic capability. They say this addition will be enabled by major algorithm changes they are already working on. Note that they believe that the hardware, with relatively minor and identified changes, is already capable.

The reason we are interested in this for AI is that that to truly have an automated car, the level of decision making by whatever computer is involved is extensive. Maybe not truly “thinking,” but it will appear that way if the car can really handle all the various scenarios it will have to deal with. To make the computer decision making as quick and energy efficient as possible, certainly novel computers like those built with the IBM TrueNorth chip will be considered. And the algorithms developed for automatic cars will certainly speed up applications in other areas that don’t have the potential $300 billion annual benefit of self-driving cars.

This work will not go unnoticed by those working towards truly thinking AI, because much of the improvements in computer architecture and advanced algorithms will be able to be applied in other areas, including projects whose goals are to emulate the way the human brain thinks, or its non-biological equivalence.

I have published a novel that puts a possible face on the AI issue. The book is fiction because I don’t have a chrystal ball to predict the future. But it does present a story around a possible scenario. The name of the book is Artificial Intelligence Newborn – It is 2025, and I am Here! It is published by and can be purchased from Kellan Publishing:–Its-2025-and-I-am-Here_p_55.html. For those that prefer reading on a Kindle, go to Amazon: Amazon will have the hard copy available within a few days.

Artificial Intelligence Post Number 12

August 21, 2015

Per Cade Metz of Wired, 8/17/2015, IBM for the first time is sharing their TrueNorth computers with the outside world. They are running a three-week class for academics and government researchers at an IBM R&D lab. The students are exploring the chip’s architecture and beginning to build software. At the end of the training session, the students, which represent 30 institutions on five continents, will each take their own TrueNorth computer back to their labs. Let the games begin!

In the meantime, President Obama has signed the National Strategic Computer Initiative, an executive order that sets the goal of producing an American supercomputer with an exaflop of processing power. This computer will be 20 times faster than the current fastest supercomputer, which is owned by China. This new computer will use parallel processing, but not be as unique as the IBM TrueNorth chip design and will not require new computer code. Nvidia is creating the supercomputer system which is initially targeted for the medical field.

As we are seeing, the efforts on supercomputing and on learning to use chips capable for AI are accelerating.

Artificial Intelligence Post Number 11

August 14, 2015

In my recent posts I have emphasized the changes most likely to happen with the simplest form of AI. There seems little doubt these will happen because we can already see signs of them in the marketplace in advanced robotic manufacturing and the progress towards self-driving cars.

But what about computers that truly think? This is the level of AI (AGI) that is frightening many of our brightest people. Is this just a “sky is falling” thing? After all, you can go back thirty years and find articles, books, and even movies that predicted that by 2001 thinking computers would already be here and causing havoc! Is anything that much different now? Sure, computers are bigger and faster. But all the computers in use still use the von Neumann linear processing architecture developed in 1945. They are all just really fast calculators!

But there IS now a real difference! I have already mentioned IBM’s TrueNorth chip that does parallel processing and attempts to emulate what the human brain does. And a company called HRL is developing a chip that comes even closer to emulating the brain in that its internal connections adjust to new data – it learns from experience, much like a child! But none of these approaches are truly IDENTICAL to the biological brain. Are they close enough in design and application to actually become thinking entities? I don’t think (pun intended) that anyone truly knows, because we don’t really know how to even define “thinking.” If these chips are loaded up with data and given goals, will they independently find unique paths to reach those goals? Of course, if they do, that also could be a problem. If you ask a thinking AGI computer how to solve global warming issues identified as being caused by humans, their advice to kill all of mankind may be valid but not welcome! Or even if their advice is less terrifying, if they suggest shutting down all coal-firing power plants for example, is this viable politically even if it may be theoretically possible? And to eliminate this kind of impractical advice, do AGI computers require some morality judgements based on human values? We cannot even agree on what those are within the human race. We quickly get into religious and philosophical issues as we get closer to the possibility of AGI.

I will continue to monitor the progress of AI as much as possible just by trying to glean as much as I can from published articles. Hopefully blog readers will help on this. Even if we never get to the very frightening level of AGI or developing thinking computers smarter than us, we need to be monitoring the advances that are likely to explode with the introduction of chips like the IBM TrueNorth.

Artificial Intelligence Post Number 10

August 9, 2015

In my last update, I showed how IBM is investing billions of dollars that almost guarantees that a high level of artificial intelligence is going to happen, and it is going to happen soon. The IBM TrueNorth computer chip will give much faster computer speeds at much lower energy use, and it accomplishes this by mimicking much of the parallel processing of the human brain. And while the programming of this chip is evolving, downsized Watson (the computer that won on Jeopardy) clones are being expanded into many fields and being filled with broad knowledge bases. Once TrueNorth is fully operational, it will be relatively simple for IBM to marry the systems together.

To see the eventual effect of all this on the US economy, and on potential investments, we have to look at its broader effect. I already mentioned that fields like law will change dramatically since so much legal history will be so easily obtainable once that information is in a system like Watson. The need for lawyers will be greatly reduced. Doctors and teachers will be focusing on those relatively few far outside the norm, because the majority of patients and students will be well served by programs that are customized to their own needs, drawing from huge data bases. The interaction with the computer will be much different than most current computer programs, because the computer voice will not be discernably different than if the user were interacting with a live human. This indeed is the Turing Test criteria; that someone should not be able to tell if they are taking to a computer or a person! This will eliminate the need for many of the call centers often located in India!

This machine intelligence explosion will affect many products. Already mentioned are robots that are far easier to train and can interact more easily with multiple sensors. But perhaps less obvious is its effect on automobiles. Look at any major car manufacturer and how they are stressing their new driver-assist systems. Cars are becoming increasingly smart. Tesla has indicated that they will likely be making their own maps based on each of their vehicles communicating to each other and back to Tesla. This will make roads, road conditions, and traffic issues close to real time.

Apple is hiring people that have automotive backgrounds, and they may use some of their massive financial strength to enter the automotive business now that it is becoming so software and computer driven. Electric vehicles seem destined to become a bigger part of the industry since they are so easily controlled by electronics and are “greener” than their internal combustion engine competitors. Certainly there is a lot of speculation of Apple and Tesla somehow combining. Perhaps Tesla will supply the vehicle drive, batteries, and base structure to Apple; and Apple will design their own body and electronics for their own car. This would enable Apple to get into the business easier, give Tesla some needed cash, and better utilize and expand the charging infrastructure that Tesla has already started. Again, the growth of intelligent cars is making this whole thing relevant. And what will this do to manufacturers like GM, Chrysler, and Ford that are largely still designing vehicles like they did ten years ago? I don’t think that I would like to own their stock!

Already mentioned in earlier updates was China’s economy slowing as robotic manufacturing enables lower cost production of volume goods in the US. Recent economic data indicates that this is already happening. Sadly, as China slows they may become more combative, which also seems to be the case. Artificial intelligence gains will definitely hurt counties whose economies are driven by volume products produced by cheap labor working in poor working environments. The US economy is likely to continue to grow at its slow but consistent pace. And the stock market will reflect that growth, especially in companies like robotic manufacturers and those companies making smart systems for teaching, medical care and the legal profession. Most computer products are likely to gain from the greater efficiencies and downsizing that AI gains will enable. For example, a computer watch will only be limited on how data can be displayed efficiently, not in its computing power or ability to communicate.