The National Geographic Daily News (http://news.nationalgeographic.com/news/) recently published an article: "Watson Wins Jeopardy!—6 Artificial Intelligence Milestones" that has some very interesting observations. These observations are important to those of us here at Discovery Machine, Inc. for many reasons. The example milestones offer a nice historical style glimpse at the history of artificial intelligence (AI) and contextual opinions on the potential long term influence and impact. Discovery Machine, Inc. applauds the growing attention and realization of the potential uses of AI.  We have taken a disciplined approach to implementing useful AI based solutions from our very earliest beginnings at Georgia Tech. Discovery Machine sees much of the true long term value of AI to be the enhancement of people’s expertise in many and varied computer based settings. The core product for this is our Discovery Machine Modeler, a knowledge articulation and leverage development and deployment environment. If needed the Discovery Machine Modeler  can be customized to speak the language of a particular expert’s world. These domain specific Custom Consoles offer an environment where all types of people can amplify their own expertise and the expertise of others to create all kinds of "Wins" in the commercial and government arenas. "Wins" that are not for a computer, but for the people who will greatly benefit from AI based knowledge leverage. These benefits are evident in such varied fields as medicine, education, national defense, space exploration, computer gaming, financial services, etc. Winning at Jeopardy is a great way to help cast products like the Discovery Machine Modeler in a new and welcome light, thanks Watson! The National Geographic Daily News article: " Watson Wins Jeopardy!—6 Artificial Intelligence Milestones"  is linked at: http://news.nationalgeographic.com/news/2011/02/pictures/110217-watson-win-jeopardy-ibm-computer-humans-science-tech-artificial-intelligence-ai/#/jeopardy-watson-computer-winning-gary-kasparov-deep-blue-chess_32328_600x450.jpg

Have you ever noticed how you can look forward to something for so long, and then all of a sudden it’s upon you, and then it’s gone? I think that should teach us that life is to be lived one day at a time, and in the present. My friend Spencer Johnson had a great message years ago in his book The Precious Present. He said that we need to learn from the past, but not live there. Plan the future, but don’t live there. Because we are at our happiest when we’re living life in the present - one day at a time. So if you look forward to something for a long time, and then it’s gone, now you’re back to your regular life - what are you going to do about that? It’s interesting and powerful to recognize that life should be lived in the present. Time flies. We’re here for such a short period of time. So enjoy every single day. And reach out and tell someone you love them and you care about them, because when all is said and done, as I’ve said many times, the only thing that counts is who you love and who loves you.

Unless you’ve been hiding under a rock you have probably heard the news that IBM’s Watson beat the two best Jeopardy! players of all time, Ken Jennings and Brad Rutter.   It is important to see this achievement for what it is and for what it portends for the future.   Yes, Jeopardy! is a game, and yes it has quirky rules, and yes it is limited to questions and answers.   But, it is also unconstrained.  The questions can address any topic and rely on the full panoply of human experiences from jokes and puns to complex knowledge based relationships.  This is something new! Artificial Intelligence (AI) has made numerous unkept promises over the years.  As early as the 1950’s researchers made claims that computers would "soon" match or exceed human beings in their ability to reason.   This misplaced enthusiasm emanated from successes in games such as checkers and then chess with a series of wins culminating in IBM’s Deep Blue beating the reigning grand master, Gary Kasparov, in 1997.   In specialized disciplines, expert systems demonstrated great aptitude for decision making in narrow areas.   In the 1970’s and 80’s it was presumed that these successes would be combined into a general intelligence.  Again, this promise went unkept.  Why?  Efforts in combining these systems through interchange languages and common-sense ontologies proved to be as hard as the AI problem itself.   So through the 1980s and 90s, AI was left with a set of practical systems (some quite successful) that were expensive to build, limited in their application, and completely incompatible with each other. In the 1990’s other paradigms of intelligence emerged, such as case-based and model-based reasoning.  These paradigms challenged the notion that heuristic rule-based search (the underlying engine of most expert systems) form the basis of general intelligence.   The role of experience, the retrieval of stored memories, and the adaptation of these memories to present problems proved useful to additional classes of problems.   Still, experience-based systems were limited to the range of experiences provided to them.   As AI researchers like to say, all these systems were "domain specific." What Watson now shows is that an artificial intelligence system can succeed in a "domain independent" problem space covering wide ranging questions of subtlety and nuance. Given the computational power of 2,880 processors and some 14 terabytes of memory (not disk space mind you), Watson can answer questions in under three seconds that stagger the imagination. How does Watson do this?   Aside from the raw computational power, Watson uses many different strategies all at once to come up with the answers and selects the best answer from those uncovered.  This multi-strategy approach is one that we at Discovery Machine Corporation have embraced at a smaller scale for use on your desktop machine.   Intelligence does not come from a single strategy such as heuristic search or CBR, but rather from a combination of strategies.   The strategy used for puns is not the same as that used for determining historical dates.   At Discovery Machine we capture the many and varied strategies of experts and combine them to form intelligent behaviors.   What Watson portends, is that as computational power increases, domain independent AI will be available on your desktop or even your phone in the not too distant future.

There is a Positive Psychology course that Margie and I have been taking that is really interesting. We ran into a guy named Nathaniel Branden, who wrote about the six pillars of self-confidence. His big theme is nobody’s coming. If you are thinking about someone who is going to get you out of a situation, and you’re waiting for them to take all the action, the reality is that people can do things, but nobody is really coming. What are you going to do?  One thing that’s interesting is the difference between passive victims—people who are in a situation and immediately go to self-pity—"This is really tough." Then they want to point fingers and blame other people. This leads to frustration, and eventually anger, and things kind of spiral down that way. This is the passive victim that somehow thinks their fate is in somebody else’s hands, versus the active agent who takes action—"Okay, this is tough, but what am I going to do? What can I do in my area? What ideas do I have?"  They are willing to take responsibility, which is being able to respond, and give suggestions that will help. They have a feeling of confidence—"Somehow we’re going to make it through this thing together." This leads to hope and optimism. We all need to take action—what can we do to help?  Let’s work on responsibility. I have confidence and hope. What is it that makes some people be able to pull out of tough times? It’s all about resiliency. So remember—we’re all responsible somewhat for the condition we’re in. So be an active agent, not a passive victim. Life is a very special occasion. Don’t miss it with a lot of negative energy.

Last week, Todd discussed IBM’s new Jeopardy playing super computer, Watson. The rise of machines that continue to evolve capabilities once reserved for humans brings up the notion of a technological singularity. That is, a point in time where life on earth qualitatively changes due to technological development. There are a number of mechanisms by which such a change could occur. For instance, if medical technology develops to the point at which human lives can be extended indefinitely, this would result in a qualitative change in the way in which we go about living those lives. However, for the purpose on this discussion, we’ll talk more about the type of singularity which would occur through creation of a superintelligence which exceeds the capabilities of the human intelligence creating it. Once such an event occurs, some significant shift would take place in the way we live as humans. Such an intelligence would be able to improve itself, and do so faster than humans would be able to. What’s more, humans would likely not be able to predict such changes. A malevolent AI with such capabilities would represent an existential risk to humanity. Indeed, even an AI which is simply indifferent to humanity could represent a huge risk, because as Eliezer Yudowsky points out, "The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else." Thus, Yudowsky suggests the need for "Friendly Artificial Intelligence" or FAI. In some ways, the notion of FAI relates to Asimov’s idea of the "Three Laws" of robotics, which ultimately attempt to protect humanity by hard-coding a set of rules into all artificial intelligence. However, scientists like Yudowsky suggest that such a simple set of rules is insufficient, as an entity with greater-than-human levels of creativity and resources would be able to come up with a way to circumvent or rationalize away those rules in any situation. This is, after all, what intelligent beings do; we find ways to circumvent roadblocks, and achieve our desired goals. Given that, the idea of FAI suggests that the way by which humanity can protect itself from a malevolent or indifferent AI is to make sure that the ultimate goals of that entity are in line with those of humanity. In this situation, while a friendly AI would be free to harm humans, it would choose not to because the idea of harming other intelligent beings would go against its core goals. This then brings up the question of what those goals should be. If we are to create a new kind of artificially sentient being, and we are the ones who will define its sense of right and wrong, and to what ends it should direct its efforts, this will require us to come up with answers to some very difficult questions. Interestingly, those questions will likely be the same ones which have puzzled our species since we were first able to engage in complex abstract thought. Only the stakes will suddenly become much higher. Whatever our decision about what constitutes a good and just course of action, the entities we create will carry those beliefs onward, possibly with such great efficiency and resolve that we will no longer be able to alter their course after the fact. So we may only get one shot at it, and we’ll need to make it a good one.

Jim Steffen was a graduate student of ours at U. Mass. He wrote a book called Aligned Thinking and has been helping us work through some of his way of thinking. The most important concept Jim talks about is MIN—which is "Most Important Now." He says that the way you really enjoy life the most is to decide:  What am I going to do right now? What is this hour about? How can I focus my energy so what I’m doing right now is the most important thing I can do, so I’m not in the midst of one thing and thinking about doing something else? You know, my mind is all over the place. I know a lot of us like to multitask and all those kinds of things. But it’s great if you can get in the MIN attitude and think, okay, I’m going into this particular meeting. Where does this fit into my life? What am I trying to do? How can I get into the mindset that this is the most important moment right now, and really focus in on it? It’s the same way with people. Try to just focus in on people for three or four minutes and just be there for them. This is the most important thing right now - this is a MIN relationship. So what I’m trying to do, first thing in the morning, is to look at the day and see how I can plan what I’m supposed to be doing and how I can get my mind set on the most important thing I ought to do. It’s a wonderful little concept and thought about focus in life. So I thought I’d throw it out for you today.

The Department of Labor just released the jobs report and at first glance it looks like great news.  The unemployment rate fell to 8.9%, nearly the lowest rate in two years.   Private employers added 222,000 jobs in February, near a 12 month high.  Everyone is excited about this news.  Even Rush Limbaugh focused on it during his show today (of course he claims the Department of Labor fabricated the number since no president in history was re-elected when unemployment was over 9%). So, does this news mean it is safe to start spending my savings now, maybe buy some stocks?  If you listened to the current hype, that would be a likely conclusion.  This is the type of question that a Discovery Machine model would be good at answering - or at least help you answer it. Simple rule-based AI systems would collect statistics like unemployment rate, jobs added per month, percent of underemployed people, etc , and analyze them with a deterministic algorithm.  The algorithm may filter or scale each statistic, compare them to pre-determined thresholds and then compute a weighted sum of these results to generate a final recommendation.  After some tuning, this system would likely work well for a small set of circumstances. A dangerous thing here is that statistics such as those above are only valid under certain sets of conditions.  Mean and standard deviation don’t need much if the data you are measuring doesn’t follow a normal distribution.  A system like this can produce misleading results with very high confidence. A Discovery Machine model would work somewhat different.  First our knowledge capture methodology would greatly broaden the set of applicable circumstances.  It would consider many other factors like what is the current time of year, what has been past performance after a set of similar statistics was attained, what percent of the recent jobs added are in retail or farm jobs and so on.  This domain specific knowledge is gathered from a subject matter expert. The decisions in the Discovery Machine model could be rule based, but as more and more conditions are considered, these overall decisions become more nuanced.  The result is that the model produces recommendations that apply over a larger set of conditions.  Sure, we could still use the same statistics, but would only draw reasonable conclusions if the assumptions behind the statistics are valid.  This type of knowledge embedded in the model is one way to reduce that chance that the system would make a stupid AI mistake (like some that Watson did).  

Here are pictures of Easter 2010, also known as The Kissam’s Jewish Easter.  No, The Kissams aren’t Jewish, but it seems more than half the people invited every year are.  This is probably my tenth or eleventh year celebrating Easter together.  This year Amy joked that it may as well be Fourth of July, you’d never know it was Easter-except there always is a honey-baked ham.  Anyway, it’s all about being together, drinking wine and champagne, sometimes doing our nails, and just generally having fun. The morning started with Beaner in her mismatched jammies, digging into her Easter basket (thank you, Nana!) Auntie Jinx came to town from NYC, and it was fabulous having Team Roberts in its entirety: Jinx brought the bunny paraphernalia, and Winky couldn’t wait to get her hands on all it! Here’s Lex’s god-father Sumner, hanging with Stu and Clara: I’ve been patiently waiting for Clara to grow into this dress from Grammie: Coco posing the Bunny Ears on Clara, in the background you can see Amy, Sybil’s daughter. Aren’t these little bunnies adorable?  Lex, Coco and Kimmy: Clara and her Godmother: Tough girl on Easter Sunday: It was the perfect day-warm and sunshiny in Kim’s backyard: Loving her Aunties: "Trot Trot to Boston . . ." with Sybil: Once again, we failed to get a sister, family , or group shot with all the Easter guests.  Argh!! As you can see Clara had lots of outfit changes-mostly we kept her in just the little white bloomers-it was so hot and what is cuter than a naked little baby girl?

I just heard a very interesting theory about the Golden Rule, which is in almost every faith-you know, "Do unto others as you would have them do unto you." It’s about loving your neighbor as much as you love yourself. This theory was that you can’t really love your neighbor if you don’t love yourself. If you don’t feel positive about yourself, then it’s pretty hard for you to reach out and be positive to other people. Mahatma Gandhi said, "Be the change you want to see in the world." There was a story about a woman who journeyed for miles with her son to have an audience with Gandhi. She said, "Would you help my son? He eats too much sugar." And Gandhi told her to come back in a week. She couldn’t quite understand that, but they trekked all the way home and came back the next week. They then sat with Gandhi and he told her son to stop eating so much sugar. She said, "Why couldn’t you have told him that a week ago?" And Gandhi said, "Because I was eating too much sugar myself at that time."  Ha! The other thing that’s really interesting is that if you feel good about yourself, it makes other people around you feel good. And if they feel good, they send those vibes back to you and they kind of multiply. Norman Vincent Peale said, "Every day you have a choice. You can feel good about yourself or you can feel lousy. Why would you want to choose the latter?"  If you feel good about yourself, then you’re able to reach out and help others. Helping others is about happiness. The more we reach out and help other people, the happier we get. In fact, most of the time helping other people makes you feel better than if you were doing something for yourself. So take care of yourself. If you do that, then you can take care of other people. It all starts at home. Confucius said, "It’s self, family, neighborhood, state." If you want to create a great nation, a great state, you’ve got to start with yourself. So when you’re discouraged, remember that the change we want to see in the world has to begin with ourselves. Be good to yourself.

I have been looking at the increase in the knowledge loss from Baby Boomers, as this generation starts to turn 65 in 2011.  The question is always "what is the value of this knowledge"?  A 1/3 of Americas workers are baby boomers. Born between 1946 and 1964.  I am a boomer. We generate half of the America’s consumption spending and we lead most of the companies.  But what is the value of our knowledge? What is the mistake, when people keeping telling you to capture this knowledge but you don’t heed the call? Remember a long time ago when the IT guy told you to back up to floppy disk, then CDs, then an external hard-drive.  We all have a story where we forgot at the wrong time and then we got a blue screen.  For me it was an important work proposal and a master’s paper.  I had to stay up for 2 days to trying to re-create something close.  I didn’t heed the call There have been a number of different attempts to create a price estimate for knowledge.  Professor Baruch Lev, of New York University, released a paper in 2000 that in the US knowledge assets account for 6 out of every 7 dollars of corporate market value.  A story released on ManagerNewZ in 2006 showed the turnover costs of a specialized high level employee to 400% of their annual salaries. In their study they included an estimated lost expertise cost.  My approach to value of this knowledge is simple.  Take your corporate net worth and delete your physical asset values and any liabilities. Then you are left with the value of knowledge. At Discovery Machine we use artificial intelligence to automate knowledge that we have captured using our patented methodology.  Companies need to capture critical knowledge before it walks out the door.  Discovery Machine can help you build an actionable model of your top employee knowledge. Now you would have the data/information to train the generation X and Millennial’s that will replace the boomers.  What is your knowledge capture plan for the wave of knowledge that is about to walk out the door?  What is it worth to you?  Think of the person who sits to your left and right.  Are they boomers?  When will they retire?  Do you have a knowledge capture plan for them in place?  Do you want an award winning DARPA funded AI way to help you do it?  Give Discovery Machine a call.