Adaptive Intelligence

As I reflect more on the idea of resolutions and goals, and growth and development, I realize more and more how much effecting change really comes down to our ability to make a choice to do so.  Be it making a change in your own life, or a chance in the world around you, the first (and in many ways, most important) step is just to start.

The Key to Growth is Choice

this is Roy... Roy the killbot

Photo by Don Solo

It seems so simple, and yet, it’s something that so many of us struggle with.

A huge part of the struggle happens because we’ve had choices turn out badly in the past. If experience is the best teacher, then experience seems to teach us that making choices can potentially be a very good — or very bad — thing.

The result? Our ability to make choices hinges on how comfortable we are with our ability to accurately predict the outcome of our choices. If you are relatively confident that choice A will lead to outcome B, you will be more confident in making that choice. If you aren’t, then making the decision will seem much harder.

If you want to grow and change, eventually you have to master the decision-making process. And if part of the problem is that you’re not confident in your predictive abilities, then it would make sense that what you need to do first is work on the accuracy of your predictions. As you become more confident in your prediction-making-ability, you also become more confident in your ability to make the “right” decision — which will then make decision-making and growth come more easily.

Artificial Intelligence in Software

Our brains actually behave quite a bit like a piece of computer software: based on certain inputs (experiences), the software generates a set of outputs (predictions). If you give it insufficient data, you’ll get less accurate results. Even beyond simple input-output, our brains actually function like the “bad guys” in modern video games: with adaptive artificial intelligence.

In traditional video games, the artificial intelligence displayed by computer characters (NPCs: non-player characters) is very limited. If you shoot at an enemy, it will shoot back. If you run into a room where there are enemy characters, they will attack.

More modern (“cutting-edge”) video games feature an emerging technology known as adaptive artificial intelligence. In these games, an enemy sharp-shooter might adapt to your firing patterns, or they might adjust their tactics to take advantages of what they learn your weaknesses to be.

In developing this type of software, programmers and researchers are attempting to replicate what the human brain does naturally: use knowledge acquired from prior experiences to anticipate the outcome of future choices.

Predictive Patterns

Every time something happens in our lives, our brains file that information away for use in future situations.

You can see this most clearly by watching a child learn something new. When that child is learning how to walk, he or she may lean too far backward and fall down. The mind takes this information and makes a note not to lean so far backward in the future.

As we grow older, our brains have built up quite a large “database” of experiences from which to draw upon. As we continue to make choices, the results of those choices (both large and small, conscious and subconscious) are added to our knowledge bases, and are factored in when we make decisions in the future.

This “knowledge database” is the means by which our brains categorize and generalize experiences, and creates predictive patterns. Because the brain remembers the outcomes in general terms (you may not remember exactly when or how you learned to walk, but you still know how to keep your centre of gravity in the right place), it naturally identified commonalities and patterns.

As a result, we are not only are we able to predict outcomes based on “what happened before”, but based on what we expect will happen — our brains are natural simulators. It’s how you know not to put your hand into a pot of boiling water; even if you’ve never done it before, you expect that the outcome would be painful, and so you don’t do it.

Predictions and Expectations

In 1738, famous polymath Daniel Bernoulli suggested that we calculate expectations in the following way:

Expected value = Odds of gain × Value of gain

In general terms, this simplified formula forms a basis of statistics and probabilities. Say you were offered a bet: for a buy-in of $100, you get to flip one-thousand quarters, and you get to keep any of them that turn up heads.

Ignoring the fact that you would be incredibly tired of flipping coins by about coin 200, and that you probably wouldn’t want to be lugging around hundreds of quarters, according to Bernoulli’s formula we should take the bet. The odds of a coin turning up heads are 50%, or 0.5. The best possible potential gain would be if every quarter turned up heads, thereby giving you $250. So the expected value of the coin-flip game is:

Expected value = $125 = 0.5 × $250

Our expected value for the coin-flip game is greater than the initial bet, and so statistically, the bet is a good one to take.

We do the same thing for non-numeric decisions as well. Our mental simulators evaluate the odds of one outcome or another. Then, considering how much potential benefit the outcome(s) have, the simulators give us an emotional response: yes, this is a good decision and will likely turn out in our favor, or no, this is not likely to turn out positively.

When Expected and Actual Value Don’t Match

As we’ve all experienced, sometimes, our choices don’t turn out for the best. Going back to the coin-game example, this makes sense. While on average, we would win $125 from the game, it is possible that we would have bad luck and flip only a handful of quarters. In other words it’s possible that expected value and actual value just won’t match up.

A really common example of this is going to see a movie that you’re really excited about. Your brain evaluates the odds of completely enjoying the movie to be quite high — say 80%. But that still leaves a 20% chance that you won’t enjoy the movie. So while you were really excited about your decision to see the movie, you may still leave the theater saying “I just spent $15 to see that?!?

In this type of case, the expected value has been calculated exactly right, but that doesn’t guarantee the outcome — so the expectation doesn’t match up with reality. While we’re not thrilled with these outcomes, often times we’re “okay” with them, because somehow we subconsciously understand that we made the right choice, it just didn’t work out this time.

Even though it didn’t lead to the best outcome, the experience is not useless and our brains don’t throw the result away.  Remember that our brains calculate odds based on patterns stored in our knowledge bases (our memories and acquired knowledge). Both the choice and the result are added to our memories, and the next time we make a related choice, we will factor in the “new information.”

Need More Input

New information can only go so far, though. Because we are human, with limited experiences, we very often don’t have the ability to accurately predict the odds for our situation. Especially when we have limited information, our predictive powers can be “off”.

Consider the following example; I’m going to give you a set of numbers, and you’re going to predict the next three numbers in the series…

3, 5, 7

What’s the pattern? If you’re like most people, you’ll identify those numbers as “odd”, and your prediction will be 9, 11, 13. If you’re more mathematically inclined, you might think that I’m talking about prime numbers, and answer 11, 13, 17. But in either case, you would be wrong, because I’ve only given you limited information.

What if I now gave you the next three numbers as well?

3, 5, 7, 13, 15, 17

By taking in more information, you could correctly guess the next values to be 23, 25, 27. But the thing is, without that further information, you would have been hard-pressed to make an accurate prediction. Just like with those computer baddies, more data leads to better predictions.

Type Mismatch: Lots of Bad Information

Even when we’ve taken in a whole bunch of data, we’re still talking about predictions and possibilities. And as we all know, even when we spend a lot of time putting thought and care into making the “right choice”, it doesn’t always work out. For all of our thinking, our simulators don’t always come up with the optimal solution.Sometimes, the new information that has gone into our brain-powered pattern recognizer isn’t representative of actual fact.

Let’s apply this to a “real life” problem: winning the lottery.

Statistically, the odds of actually winning are so miniscule, you may as well be tearing your money up into bits and throwing it into the garbage. So why do people continue to buy lottery tickets? A big part of the problem is that their decision making process is based only on a partial data set. Think about it — how many times have you heard in the news about the latest “big winner”?

Now — how many times have you heard about all of the people who didn’t win? Of all the millions and millions of tickets that are bought week in, and week out, you only ever hear about the winners! That’s a lot of new information reinforcing the idea that you too could win. And if you were to sit down and listen to every person tell you the outcome of their lottery ticket purchas? You would spend years listening to people say “I didn’t win” before you’d hear a single person say “I won”.

The point is, if you only encounter one type of experience (positive or negative) that reinforce the same message, you’ll always get the same feedback — whether or not that feedback is accurate or not. And that feedback can skew your perceptions of reality pretty drastically.

Fine-Tuning Your Predictive Intelligence

I hope it’s becoming apparent that unless you continually give your brain new stimulus and information, you won’t be able to increase your predictive ability. You never know when you’ll experience something that won’t fit into your previous patterns — something that would hugely increase your predictive intelligence. On the flip side, if you want to become better at predicting outcomes, your only real option is to experience a wide variety of optins.

In other words, the only way to improve your ability to make the right choices is to make choices that could turn out badly.

“Oh gee, that sounds like fun,” I hear you saying sarcastically.

Here’s the thing: making the wrong choices is still better than just staying the same all the time (“not making a choice”). At least by making the choice, you will have learned something — and you’ll be more empowered to make a more informed choice the next time. I strongly believe that we do not make mistakes; we simply have opportunities to learn. If you look at it that way, then there really is no way to lose.

A common saying in my family is “this too shall pass”. The not-so-good outcomes? They’ll pass. The good outcomes? They’ll pass too. But the knowledge you’ve gained in either case will stay with you. Your predictive intelligence will adapt and develop as you continually feed it new inputs. And you’ll reap the rewards in the long term.

One thought on “Adaptive Intelligence

Leave a Reply

Your email address will not be published. Required fields are marked *