Flying

Free Will

There are two lay theories of human behavior. Either it's determined strictly by cause and effect, like clockwork - or we do whatever we want to do, regardless of cause and effect. (Or some combination of these two.) Most people believe in free will because they know that they evaluate choices and make decisions. They live it every day. People know they make choices because, for example, they like the way fattening food tastes. "No one can tell me what to eat, no sirree - I have free will."

There are two threads in this debate that I want to explore. The first is how we confound external force with cause and effect and allow that to create a misunderstanding of what cause and effect really is. The second thread is how "free" relates to randomness. Both of these have a lot to say about the nature of artificial intelligence.

When we are forced to do something, we say we don't have free will. We don't want to be shot so we give the mugger our money. That's not a free choice, right? Wrong. It is a choice based on expected outcome. If the mugger held our hand and forced it into our pockets to get the money, then that would be true cause and effect. But if he just stands there with the gun and orders us - then it's our free choice that less money is better than less life, and so we act on it. This is no different than eating those fries despite being warned - at that moment we value the taste over the slight possibility that this one french fry will be the one that kills us. If one of these is free will, then both are. So what would non-free will really look like?

Non-free will might be like when an odor triggers activation of a brain state that corresponds to "french fry" and a memory trace of pleasure. Amongst all the competing memory activations - including the ones about how much you weigh and the effects of overeating - the hypothalamus reacts most strongly to the pleasure of the taste memory and triggers the desire to taste it again (actually, this IS the desire to taste it again) which triggers arm and hand movements to get that fry into your mouth. Does that sound like cause and effect?

Most people react to this last analysis as, "but that's not what it feels like! I know what I want and I choose it - my brain doesn't make me do it." To that my response is: do you also choose what you want? Or is what you "want" some function of your genetic makeup and your experiences? Isn't this still cause and effect? (I'm purposefully ignoring the mystical argument that we are not our brains. That just moves the problem one step away without actually solving anything.)

After many years (yes, years) of puzzling over this, I find that I really cannot even comprehend an alternative. What else would "free" will be, if not acting to achieve your goals without external coercion? The only other thing would be randomness. Either you do something for a reason (and that reason being some function of things you learned in the past) or you do them randomly with no reason. I don't think we really want to equate freedom with randomness - there's no meaning in that. So, freedom must mean that we act to achieve our goals, however ill-conceived they may be. And we may choose specific goals, but they are based on what we like or what we want, neither of which are under our control.

We now know that, with quantum uncertainty, there really is no such thing as mechanical cause and effect the way we used to think of it, because micro-things happen according to a probability distribution instead of at some exact predictable time. So there is a randomness to everything, even if not normally perceivable. That randomness can play a part in what we end up liking (by history or genetics,) or which brain state wins the debate and triggers which action amongst the top contenders. But this only adds variety, it doesn't change the core idea of "free" will.

We do what we want when free from coercion. We contemplate options and decide which ones we like better. But we are not free to decide what we like.

Emergent Consciousness

One meaning of emerge is to come into existence. This implies that something new exists where it didn't exist before. Man emerged from fish - if you believe in the most radical version of evolution. And fish emerged from... non-living stuff. Live emerged from non-life. How can this be? A lot of people think it cannot be. If something new comes into existence, then it must have been created by some intelligent entity. I'd like to side-step that issue for now and talk about what emergence really means, especially with respect to intelligence and consciousness. Can consciousness emerge from unconscious processes?

One of my favorite examples of emergence is the property of temperature. Temperature is a property of many molecules moving at different speeds and directions. The greater the disparity in their movement, the higher the temperature. If they are all moving in exactly the same direction and speed then there is absolute zero - which is still a temperature. No single molecule can have a temperature because that property emerges only when you consider a population of molecules. Put multiple molecules together and something new emerges that simply does not exist in a single molecule. 


Can consciousness work the same way? Put a bunch of unconscious stuff together and suddenly you get consciousness? No secret sauce, no divine spark, it just emerges. I personally think that this is true, but who cares what I think. However, as a working hypothesis subject to being proven false, it could be a great start. Actually, you have to reverse the hypothesis and presume that consciousness cannot simply emerge, and then demonstrate a single instance of it actually happening, to make scientific progress here. This is a lot different than artificial intelligence. Computers can mimic intelligent behavior without being conscious, without "knowing" what they are doing, but how can you know if it's indeed conscious? I don't think we have a good enough definition yet to be able to measure it scientifically, but I do think we can grasp enough of an idea of it that we can make progress.


One idea I've read is that intelligence can emerge from crowd behavior. If you have dozen's of people guessing at something, like the weight of an elephant, the crowed average is frequently amazingly close to the right answer. However, there is no crowd entity that is "above" the crowd and suddenly becomes alive on it's own. No, this - if it happens reliably and not just randomly - it would be emergent intelligence, not emergent consciousness.


At the core of consciousness is the idea of self-knowing. The idea that one knows what they know; meta-knowledge. It's not enough to react, one has to contemplate options and chose their action before we might consider them conscious. So who is it that is doing the inner knowing? The old idea of a homunculus arises - the idea that each of us really exists as an inner person who witnesses what the outer person is doing and thus controls the outer person. But ... then who is the homunculus of the homunculus? Removing the mystery one level does not answer it. There must be a beginning, or something that consciousness emerges from, to break the cycle and avoid infinite regression.


Marvin Minsky, in his seminal The Society of Mind, 1982, disputed the very idea of consciousness by positing that our minds are merely a collection of little experts that act in concert to achieve higher levels of thinking. One homunculus wasn't enough for him, he wanted us to have thousands. And with thousands of inner experts - there was nothing left to emerge except intelligent behavior. I never found his work compelling. In fact, that whole direction of expert systems as a model of artificial intelligence has failed to live up to the hype. This doesn't stop new classes of students from trying it again - the MIT media lab project Concept Net is essentially a rehash of Doug Lenat's failed Cyc project from the 1980s. (That fact that Lenat is still trying to get it out of the lab and into the real world doesn't mean it hasn't failed.)


Not wanting this to become a book...I'll stop here for now. But I hope I've given you something to think about. Whatever that means.

Online Collaboration

"Social" media is also used for work - hence the rise in blogs such as this one. However, it's all push. Collaboration is mainly when people butt in and add their own content. Synergy only happens when people are occasionally inspired to new ideas by what they've read and integrated with their own knowledge before they add new content back into the ether. So where's the pull? Where's the actual productivity enhancement? Can we do better than just creating additive knowledge?

I'm a big fan of David Allen's books on "Getting Things Done." It's a complete system of work productivity based on a simple idea: think about each new task long enough to figure out what the next couple of steps are, and write those down in the appropriate place so that you will be reminded to do them when the time is right. The implications are huge - you mind is freed from worrying over things that cannot be done right now so your free to be more creative and - dare I say happier - in your daily work life. However, the actual mechanisms for finding "the appropriate place" to record those next steps so that you are "automatically reminded" are cumbersome and clumsy. I know - I've tried everything from an online To-Do list to the  Outlook plugin for Allen's method. Unless you're crazy motivated to keep things organized or unless the reward is immediate, you won't. And if your a natural procrastinator - the kind who could really benefit from Allen's methods - you will quickly find the task of staying on top of your to-do lists to be just another chore that you keep putting off.

Here's my thinking. What if these to-do lists were collaborative? What if you didn't write all these next steps down in a private tool that only you know about - what if they were pooled in a common area for everyone in your work team? What if each of your team's next-steps were freely available for any team member to move it forward - or to remind you when the opportunity was ripe? Often a teammate can think of an easier "next step" if they just know what your goals are. Often it's easier to solve someone else's problem - and if someone else solves yours, we all win. Can social media be enhanced to support this kind of productivity environment? Consider figuring that out to be my first public to-do pull request.

An apparent downside to this is how much privacy you give up when you tell the world (or even just your work team) all the things you need to do. Suddenly everyone knows your business and will hold you accountable when you don't meet all your obligations. Instead of a few people only knowing about one or two things that you need to do for them, suddenly everyone knows what you owe everyone else. I have two possible solutions for this de-motivator.

First, people change. "My generation" is so worried about how much privacy younger people are just giving away online. However, the younger people themselves are just fine with it. The whole idea of embarrassment seems strangely out of date in the online world. So maybe this "problem" will simply fade away. Maybe.

Secondly, the to-do lists do not have to be personal. Instead of writing down what you need to do - and hence what you will be perceived as failing at if you don't get them done - simply write down what needs to be done, anyone will do, just as long as it does get done. If you do this at your work-team level, the tasks become team goals. Then, instead of tracking what you don't get done, you simply track what each person accomplishes out of the pool of team goals. Everyone in the team can add their own goals. You can even rate the goals by difficulty and have people get achievement points by some function of quantity and difficulty for the tasks they accomplish. Hmm, sounds like a Facebook game. Must add this to my list...

Getting Started

Finally. Ok, I'm a procrastinator. Age 54, it's 2010, and my first blog.  Here we go. I want to ruminate on the intersection of artificial intelligence, gaming, social media, online classrooms, and transforming work productivity via Web 2.0+. My current role at Laureate Education, Inc. is a culmination of all these interests that has me finally excited enough to start blogging. Your call if it's worth it. See my profile for more about my background - too much to list it all here.

I'm calling this blog Virtual Intelligence as an obvious play on Artificial Intelligence. The problem with artificial intelligence is the artificial - which it is. Given enough computing power, time, and interest - computers can be programmed to mimic many aspects of intelligence. But if the manner in which it does so cannot self-generalize to other problems and evolve or adapt on its own, then it is simply not interesting as a model of what humans do. I have a lot to say about the parallel progress-blocking dichotomies that exist between neural networks and symbolic logic, calculus and set theory, behaviorism and cognitive psychology - expect to see blogs on those topics coming soon.

But I'm currently interested in how we can use lessons learned in computer gaming and social media to enhance online classrooms and, ultimately, transform what we mean by "work." There are many voices now moving in the same direction, as any TED follower is well aware. The computer game industry, and to a lesser extent the social media industry, have been forced to master the art of making work into fun. Think about it - what do you do in an online game? You go on quests and side-quests, amass virtual fortunes in various materials, learn new skills, lead or join teams to conquer obstacles and other teams... and that is what work is, right? Yet - with no external reward, millions of people spend money to do this work online - for more hours than their day job. Why? And how do we expose and leverage that motivation and environment to enrich our work experience and productivity? And once we do - and we most definitely will - what will our society become?

Big questions, and I'm hardly up to answering them. But I can certainly ruminate.