The Expert Conundrum

What is an expert? If we take the definition of the Cambridge Dictionary, an expert is “a person with a high level of knowledge or skill relating to a particular subject or activity”. No problem with this definition, right? So… a machine cannot be an expert? Well, in this post I will try to elaborate a little around these issues and how they relate to Innovation.

We can all agree that an expert is someone who has experience, a lot of experience. With this experience the expert is able to know (or do) what is right or wrong in an specific field, although it is probably more precise to say what it works and what it does not. Why? Because they have already lived that situation and they remember, or because their experience allows them to make the proper connections to answer a question. So, my next question would be if experts are important for Innovation.

If you listen or read innovation literature, you would probably find statements not very favorable to experts, actually they are sometimes reflected as conservative ignorants that hamper innovative developments; the embodiment of the status quo we, “the innovators”, need to battle with. Well, I would not be so radical, as some experts are the first ones willing to open and renew their fields, but there is some truth to it as expertise can be considered an inertia that can make radical movements more difficult. We can make the metaphor of seeing the expert as an anchor, it certainly offers resistance to movement, but if we wanted to stay for the night, the anchor is quite useful! On the other hand, let’s not completely jump to the other side of the fence, I mean, believing that anything an expert deems irrelevant would then be a great innovation in the making. Of course not, because not all changes are always for the better.

Thus, do we need experts for innovation? Experts are very useful in their field, specially knowing what works and what has not worked before. However, when there is a change of perspective, we could say that experts are not useful anymore, but that would be unfair and untrue. The problem is that when a new thing comes around, there are actually no experts on the matter because it has never happened in the past. So, my recommendation is to always take expert opinion into account as they may have seen it fail before. You know what they say, sometimes you win and sometimes you learn.

And can a machine be an expert? Well, for some time we have had “expert systems”, which basically are machines where experts dump their expertise, normally in the form of if-then rules. This way we can ask questions to our experts without them ever getting tired. Could one of these expert systems be an asset for innovation? Again my answer would be quite similar than above, good to know what happened in the past, but no so much for figuring out different futures.

But there is lately a new creature in our environment of expertise which we call Machine Learning. There are different implementations of Machine Learning (supervised, unsupervised, reinforced, …) and I am not going to go into them in this post (you can check a previous post I published on the matter of Deep Learning https://innovationpapers.net/2017/09/27/deep-learning-i/), but for the purpose of this discussion on experts we can say that the machine is able to “learn” from data, meaning that it is able to “perform a specific task without using explicit instructions, relying on patterns and inference instead” (you can find more information at https://en.wikipedia.org/wiki/Machine_learning). The machine is able to identify patterns in data and make predictions from them. In a moment, the machine is able to extract that “knowledge” and put it to work (well not in a moment, but close to one if we compare it with the years a human needs to acquire its expertise). The tricky part normally is having enough “clean” data to carry it out.

The question is, if we had enough data, can we say that a Machine Learning system is an expert? If so, do we need experts anymore? Is expertise encapsulated in data? Difficult questions and I will not be adamant in my answers, especially because technology evolves quite rapidly and we are seeing machines producing works of art or beating the Go champion. But my personal opinion is that machine learning is very good finding patterns in past data and assuming the pattern will repeat again, but that is not “understanding”, just being very useful. On top of that, you need to be careful if assimilating data correlation with truth, or you may think that eating ice-cream will increase your chances of drowning just because most drowning accidents happen during the period of time when more ice-cream is consumed. So, if the pattern has not happened before or there is something outside the data (being summer in my example), we humans are still (and I say still) ahead, because we can make unexpected or unforeseen connections.

To conclude, and I will be happy to read your comments, I believe we need experts, even for innovation activities. And “artificial experts” are not yet a substitute for what a human expert can provide. Having a lot of data cannot replace experts who are able to interpret it, but it can surely help them… the Augmented Expert!

One comment

  1. Very interesting post!
    A good morning brain stimulator, thank you! 🙂

    Yes, I have heard similar things, yes. And possibly it’s a right statement in some cases, like when “experience” it’s combined with “risk aversion”.
    But… in many others, experience is just another lever to guide innovation, or even to potentiate.
    Would Jobs be Jobs without Wozniak?

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s