August 6, 2020 Author: Matthew Renze

Based on the positive feedback we received from our discussion panel at AFWERX Fusion, Heather Wilde suggested that I elaborate on my responses and provide answers to the remaining questions that had to be cut for time. We thought it might be beneficial to my audience as well as the AFWERX Fusion audience.

How do you define the Base of the Future?

From my perspective, I think there are three key components to the Base of the Future: People, Processes, and Technology.

By people, I mean we want to create a culture of innovation. We want individuals to feel empowered to make observations, develop hypotheses, run experiments, iterate quickly, embrace smart failure, and continuously improve based on feedback.

By processes, I mean we want to use a Lean/Agile methodology. We want to utilize highly adaptable and iterative processes that are data-driven, promote experimentation, rapid prototyping, that fail fast, and allow us to pivot when necessary.

By technology, I mean we want to look at the entire base as a cyber-physical system. It’s an IoT platform that is sensing its environment and collecting data. It’s a data-science platform that allows for data-driven decision-making in real-time. And it’s an AI platform that allows for automation and improved operational efficiency.

How do you define innovation?

To me, “innovation” is a culture (or a way of thinking) that allows an organization to adapt rapidly to changing conditions. The key idea here is adaptability. How quickly can you respond to new information, how quickly can you test a new hypothesis, how quickly can you develop a new product or tool, how quickly can you evolve as an organization?

The opposite of innovation is stagnation — a lack of ability to adapt to changing conditions. So, to be innovative means to remove the barriers that prevent us from embracing change and adaptability.

How do you approach innovation?

As a data-science consultant and certified Agile coach, my job is to help organizations become more innovative with artificial intelligence, data science, and machine learning. We do this by making a series of changes to the culture of the organization. We encourage and incentivize a set of practices that create a more innovative culture. In addition, we discourage and disincentivize those practices that stifle innovation.

For example, we encourage individuals and interactions over processes and tools, we encourage working solutions over comprehensive documentation, we encourage whole-team collaboration over contract negotiation, and we encourage responding to change vs. blindly following a plan.

We also encourage hypothesis-driven experimentation, highly iterative development cycles, data-driven decision making, automation with AI, inverting information hierarchies, we embrace smart/fast failure, and we pivot when necessary.

Why is innovation so difficult?

The majority of all successful organizations over the past few hundred years (or so) were built on a foundation of a traditional approach to business. In the past, you would do some marketing research, you’d develop a product, you’d build the product, you’d launch the product, and you would still have several years to make a profit before your competitors caught up.

However, the world is now changing faster than traditional organizations can respond to these changes. Our technologies, markets, and consumer preferences are changing faster than we can learn to adapt to them using the old business model. Unfortunately, traditional organizations simply just can’t keep up anymore.

As a result, we have to throw away much of the old business model and embrace a completely new system of organizational thinking — one that’s highly flexible, adaptable, distributed, resilient, data-driven, and scientifically minded. Essentially we replace the traditional business practices of bureaucracy, process, and tradition with the modern business practices of communication, collaboration, and continuous learning.

However, this feels completely wrong and backward to most people because we’ve been taught this traditional business model our entire lives. Metaphorically, it feels like we’re trying to build the airplane while we’re flying it. Which, in essence, is exactly what we’re doing with Agile software development — and this is really scary when you first start working in this new way.

Convincing an organization to abandon their old way of thinking and embrace something completely new is really difficult. But, if you look at the most successful modern businesses like Google, Apple, Facebook, and Amazon, this is exactly what they are doing. They have been more successful than their competition precisely because they ditched the old business model and embraced this new way of thinking.

What are some real-life examples of innovation in today’s air force?

I can’t speak specifically to the Air Force, but I can speak to some real-world examples of innovation that are happening in the software industry that I’m really excited about right now.

First is self-supervised learning. In traditional machine learning, we have humans label the correct answer for every piece of data that we use to train the model. However, with self-supervised learning, we find something in the data that we can use to pre-train the model without human intervention.

For example, we might remove a pixel from and image and ask the model to guess what color it thinks the pixel should be. Or we remove a word from a sentence and ask it to predict the missing word. We use this approach to train the lower layers of a deep neural network. Then we use transfer learning with a much smaller set of labeled data to teach it how to perform specific tasks. It’s much more efficient and highly generalizable.

Second is GPT-3. This year, Open AI released a language-generation model called GPT-3, which is two orders of magnitude more powerful than their previous GPT-2 model (i.e. 1.5 billion parameters vs. 175 billion parameters). This model is so powerful that it can generate text that is almost indistinguishable from a human author. Even more impressive, though, is that it can also perform highly generalizable natural-language tasks with little or no explicit training.

For example, you can manually type in a task that you want it to perform — like translate phrases from English to Spanish or translate commands from English to a SQL query. You can also give it an example or two to help it understand more precisely what you’re looking for. Then you start typing commands in English, and it produces the correct Spanish translation or SQL query necessary to execute your command. It’s not perfect, but it’s pretty mind-blowing and getting better every year.

How do I spot opportunities for innovation?

I believe there are three main components: education, empowerment, and encouragement.

First, you need to educate everyone in your organization about innovation and technology. You can’t expect someone to come up with a data-driven solution if they don’t understand the basics of data science. In addition, you can’t expect someone to dream up an AI solution if they don’t understand the basics of machine learning. They don’t need to be experts, but they do need to know what is possible.

Second, you need to empower individuals to identify opportunities for innovation and propose solutions to these problems. You do this by giving everyone in the organization permission to observe their environment, identify problems, and propose a solution.

You also want to encourage or incentivize innovation by rewarding those that identify real-world problems and propose workable solutions. Over time, this creates a feedback loop in the organization that eventually leads to a culture where everyone feels comfortable enough to actively spot opportunities for innovation and propose solutions to these problems.

I think it’s the combination of these three components (i.e. Education, Empowerment, and Encouragement) that create organizations that are able to easily spot opportunities for innovation.

As an airman innovator, how can I get other people behind my idea?

Innovation is all about economics, and organizational leaders think in terms of economics. So, you want to communicate your ideas in a language that they will understand. Essentially, you want to think in terms of Return on Investment (or ROI).

Ask yourself, what is the cost to create and maintain this solution and what is the direct and long-term benefit that this solution will provide? If the cost is greater than the benefit, unfortunately, it’s not worth investing in your idea from a purely economic standpoint.

In addition, you want to look for low-hanging fruit to get quick-wins with your early innovation endeavors. Look for real-world problems that are low risk and high reward to focus on first.

I often recommend creating an “innovation backlog” where we list out problems that we’d like to solve and their potential solutions. Next, we assign a crude estimate of the cost to solve the problem and a benefit that the solution will provide. Then we rank all of these ideas based on their perceived ROI. Finally, we work on them one-at-a-time, starting with the highest ROI problem-solution pair and working our way down to the lowest ROI pairs. Once the ROI hits break-even or (eventually) goes negative, we find new a new set of problems to solve.

What role should senior leadership play in innovation?

After many years as a data science consultant, I’ve found that buy-in from leadership is the single biggest benefit or detriment to the successful adoption of any change in company culture — especially a culture of innovation.

If you have the senior leadership’s buy-in and support, then your job is essentially half done. However, if they are resistant or (worse) actively fighting against these changes, there is almost nothing you can do to ensure a smooth and successful transition.

This is why it’s critical that we educate senior leadership on the value of innovation and get their buy-in on this new way of thinking. It’s radically different than what most of us were previously taught and what had made traditional organizations successful, so it can be a tough sell.

However, once the leadership sees the big picture and understand why these modern Lean/Agile practices actually work, they will be much more likely to lead the organization in a new direction. But this requires explaining the “Why?” behind a culture of innovation in a way that resonates with them. This is something that many consultants can’t do because they themselves don’t understand the “why” or how to communicate it effectively.

After you have the buy-in of senior leadership, their role should be as follows:

  1. They should become advocates for changing the cultural norms of the organization.
  2. They should encourage or incentivize innovative behaviors.
  3. They should discourage or disincentivize stagnant behaviors.
  4. They should remove impediments to success.
  5. And most importantly, they should lead by example.

 

To learn more, you can watch the full discussion panel here:
AFWERX Fusion – Keeping Our Technological Edge: How to Innovate Effectively

Share this Article