Beware confirmation bias — it’s clouding your strategic vision

Strategic Vision

Few things in business are as important as your strategic vision. The better defined, the more actionable, the more incisive that vision, the better your prospects as a company are — it’s that simple. If I fear for the future of a given firm, it’s almost always because I distrust their strategic vision. Whether that vision is unobtainable, built on faulty logic or false assumptions, or just generally weak, the less I trust your plan for the future the less I trust in that future. I mean, that seems pretty self explanatory, I know; but, it is worth discussing explicitly because it’s often the simpler problems that are foundational issues, and those are the ones that lead to long-term insolvency.

One of the biggest dangers when formulating a strategic vision — whether that’s for your entire firm or a new product therein — is confirmation bias. In simple terms, confirmation bias is the tendency to interpret new evidence as confirmation of one’s existing beliefs or theories. Basically, you have a narrative already in your head, so you distort any new information or finding to adhere to that narrative instead of objectively considering every new piece of information on its own accord and then adjusting your beliefs or narrative to fit the evidence instead of the other way around.

Confirmation Bias (x) Humans

Humans as a species are predisposed toward confirmation bias. Unless we make an active effort to objectify our thinking, we’re more likely to confirm our own beliefs or narratives than shape those beliefs and narratives around new and contradictory information. You see this in corporate thinking all the time — “that’s just not the way we do things here” or “but we’ve always done things this way, and things have been good until now.” You see it in purportedly scientific fields like medicine all the time — many doctors make medical assessments and treatment plans based on their experience to date instead of what the science and/or data may say about a particular combination of symptoms.

This is a predictable and repeated offense from humans. One of the most fascinating developments from the world of artificial intelligence comes from DeepMind (née Google) and its wünder-A.I. Alpha Zero. Alpha Go is the purpose-built learning machine designed to take on the world’s best Go player (which it did and dominated). The next phase of development from DeepMind came with the Alpha Zero.

Instead of teaching Zero to play by studying human games from the past, Zero was taught how to play by another A.I. system. That means the biases humans have baked into Go strategy over generations weren’t present. Instead of learning from a human master that taught it a certain way of playing because that way is believed to be superior, Zero was free to experiment with both the entire bounds of the game and near-limitless strategic creativity.

AlphaGo, which dominated the best Go player alive, was decimated by Alpha Zero in just three (3) days of learning from the Zero. It went from not understanding Go to beating the best Go “player” ever designed in three days.

Alpha Zero was recently applied to chess, another insanely complex game of strategy, and the results are, well, weird. Logic and strategy taken as givens for generations are being upended in favor of gameplay described by experts as “alien”. From the MIT Technology Review:

“It doesn’t play like a human, and it doesn’t play like a program,” [Demis] Hassabis said at the Neural Information Processing Systems (NIPS) conference in Long Beach. “It plays in a third, almost alien, way.”

Hassabis is the Founder and CEO of DeepMind. He continued to MIT Technology Review:

…It sometimes makes seemingly crazy sacrifices, like offering up a bishop and queen to exploit a positional advantage that led to victory. Such sacrifices of high-value pieces are normally rare. In another case the program moved its queen to the corner of the board, a very bizarre trick with a surprising positional value. “It’s like chess from another dimension,” Hassabis said.

Hassabis speculates that because Alpha Zero teaches itself, it benefits from not following the usual approach of assigning value to pieces and trying to minimize losses. “Maybe our conception of chess has been too limited,” he said. “It could be an important moment for chess. We can graft it into our own play.”

A.I. that learns from scratch isn’t burdened with preconceived notions or confirmation bias. It has the full gamut of strategy to try, fail, learn from and perfect. It’s upending the entire way we strategically consider a game that’s thousands of years old. And that’s just what Alpha Zero has done this year; just imagine what it could do in the near future.

This application is immensely instructive for business. We cannot be subject and prey to our confirmation biases. We have to think objectively and strategically when outlining our respective corporate visions for the future. The more we can consider facts, market conditions, our respective advantages and disadvantages, relative opportunity, etc. when determining our short-, medium- and long-term visions, the better off we’ll each be.

Alpha Zero remade Go and Chess by doing so — maybe we can remake our industries similarly.