Deep Learning for Humans

What can computers teach us about being human? A version of the world's most advanced neural network lives inside your skull. As Artificial Intelligence (AI) marches on in a perfect storm of big data and deep learning algorithms, we have an opportunity to consider what perfect storm may help us become better humans. It is likely that your most important, perplexing problems are crowded out by the less important but more urgent noise. I offer a challenge to turn your efforts intentionally towards your greatest challenges and a method for progress and persistence to stay the course to solutions and new levels of understanding- in a sense, your own singularity. 

Technical vs. Adaptive Problems

First, let's distinguish two general types of problems. Technical problems can be solved by applying a set of technical skills learned from a defined body of knowledge. Computers are very good at these, doing exactly what they are programmed to do repetitively at blazing speeds. Adaptive problems, however, do not have textbook answers. This is where deep learning attempts to make circuits and software mimic what it can of the human brain's incredible capabilities. Odds are that your greatest unmet challenges are adaptive problems with many factors that nobody but yourself can guide you through. But the odds are also that you have all the hardware you need for these challenges in between your ears. The challenge is using the right software. You cannot lead yourself or others through adaptive challenges with a technical approach. You are going to need to follow your intuition to gather more data and will probably make some mistakes in the process that challenge your fundamental assumptions.  

The hallmark of progress is discomfort. Humans do not like not knowing things. The graph above illustrates how the threshold of learning is in a relative state of disequilibrium.  There is, however, a productive range of distress, a sweet spot above the threshold of learning and below the limit of tolerance.   If we do not embrace this discomfort in an adaptive challenge, we fall into a state of work avoidance that, while less stressful for the moment, leaves us hanging in a state of relative disequilibrium.  A growth mindset is a tremendous asset for tolerating this discomfort for as long as necessary to finish our own deep learning process. In this time of discomfort we need to stretch just beyond the edge of ourselves and all the knowledge available to us in order to learn through a series of experiments. 

Growth vs. Fixed Mindset | Carol  Dweck

Experiments

The scientific method helps us understand the physical world in which we live, and can also be applied to understanding and improving our own lives. In basic terms, this method involves making a hypothesis about what will happen if you do something, designing an experiment to test that hypothesis with controls, conducting the experiment, observing what happens, and then analyzing the results afterwards to then draw some testable and verifiable conclusions. Each experiment helps us gain insight into the true nature of things until we condense that knowledge into a set of theories that can be verified by reproducing the results with other experiments. 

Hypothesis

There is something in your life you want to change, but are not sure how to change or what will happen if you even try. Einstein once famously said that "we can't solve problems with the same kind of thinking we used when we created them." An experiment is an opportunity to test a new way of doing things, or verify that assumptions you may have are true or not. The goal is to go into the experiment with a hypothesis that says "If I do X, then Y will happen." Let Y be the change you are seeking, and have courage to take some risks in defining X.  

Designing an Experiment

What is your next best opportunity for the growth you know needs to happen? Odds are you don't know unless you've done some serious reflection recently. So take some time to reflect on your life and think of a situation where you could do a controlled experiment and see if a certain course of action (X) leads to the results you think it might (Y). As a leadership coach helping Georgia Tech students design good leadership experiments for themselves through the Leading Edge, I can attest this is not easy! Designing a good experiment requires deep thought looking into yourself, being aware of your present opportunities for growth, and committing to discomfort in the future in order to learn and grow. As a coach I ask questions that facilitate the internal thought process like: what is your ideal outcome in this situation? What are five things you could do to solve this problem? What does that make others feel like? By the end of every coaching session we have an experiment defined and committed to that takes the form of a Goal, Action Plan, and way to measure Progress (GAP). Before finalizing, really ask yourself if you know that by doing X, Y will happen. If you are fairly certain Y will happen you are testing an assumption. If you really want to grow, find a Y that you do not have a clear X for and stretch your mind to find an X that you think with bring about the Y you are looking for. 

Conducting An Experiment

Have courage. As you approach the situation for the experiment remember all the thought you have put into the actions you will take and be open to learn as you take those actions. Try to execute the full experiment in as controlled an environment as possible (i.e. minimal distractions or external factors that could effect the outcomes your experiment is testing). 

Making Observations

Don't black out while you conduct an experiment. Take detailed mental notes of all that is happening as you take action. The more detailed your observations, the richer the analysis of the data can be after the experiment is complete. Take special note if things do not go according to plan.  

Analyzing Results

Here we unleash the power of reflection. Hopefully you carried out your experiment and thoroughly tested your hypothesis. Now you can see, did Y actually happen? Just by being open and aware you should have a rich set of observations that you can learn from while reflecting. Write everything down while it is fresh on your mind. Replay the experiment in your head. Pour over the data until you see themes or things you had not considered before. After this analysis you should have a set of verified or falsified assumptions and some new information that you can form other experiments around. 

GT Leading Edge's Leadership Experiments Model

Iteration

If you did your first experiment right, there was probably a rush of anticipation, fear, courage, emotion, awareness, and learning. Wow, that was hard! But you've discovered a truly virtuous cycle: reflecting, planning, experimenting, reflecting and growing.  In machine learning, computers learn by iterating on cycles like this. As humans we need to keep iterating as well. The challenge is to stay curious about yourself and what is possible. Take charge of your own singularity by looking at your brain and asking, "Is this thing on?" 

---

Special thanks to Leading Edge program at Georgia Tech for investing in helping Leadership Fellows coach students one-on-one on their leadership opportunities. This post is based on my experiences with incredible growth as a leadership coach in this program.