November 29, 2017 | Written by: Rob Thomas
Categorized: Artificial Intelligence | Data Science
Share this post:
Using and exploiting artificial intelligence (AI), is a goal for many enterprises around the world. Of course, before you can begin working with the cognitive technology, a number steps have to be taken. For starters, AI requires machine learning and machine learning, requires analytics. And to work with analytics effectively, you need a simple, elegant data, or, information architecture (IA). In other words, there is no AI without IA.
But as laborious and complicated as these steps seem, they are not the biggest impediments to embracing AI. Real success in using AI comes down to an organization’s ability to adopt a culture that is centered on data.
Don’t discount culture. It can be one of a company’s most powerful assets, or its biggest obstruction. In this case, having met with thousands of enterprises over the years, the reality is, that most enterprises do not have a data culture.
In fact, many do not even know they need one. Sometimes, ironically, current cultures can cloud an enterprise’s ability to see the need (i.e., they can’t see forest for the trees). In other cases, organizations are aware of the need, but paralyzed by the perceived, or real, complexities.
In either case, this must change. Ben Thompson, the business and technology analyst and author, once wrote, “Culture is not something that begets success, rather, it is a product of it.” If an enterprise has not had visible/material success with data, how could anyone possibly expect it to form a data culture?
Our mission is to make data simple and accessible to the world. We are enabling companies to sow the seeds of a data culture, with a practical approach to achieve a successful outcome. Said another way, we are enabling organizations to do data science faster. When that occurs, the results are tangible, the benefits are clear and the power of data is unleashed.
I once heard that the difference between a data science project and a software engineering project is that with the former, you have no idea if it will actually work. Even if you are a staunch ‘fail fast’ supporter, that is too much of an unknown for many. Most organizations that make an investment want some understanding of how they will generate a return on that investment. I understand that is not the Silicon Valley mantra, but most enterprises are held to a different standard of ROI than Silicon Valley. It’s not right or wrong, it’s just different. High certainty and modest returns is preferred by many enterprises over a more aggressive approach. In economics, we call this tolerance for risk adjusted returns.
My observation is that most of the time invested in building and deploying machine learning is not spent on algorithms and models. Instead, it is spent on the most mundane of tasks: data preparation, data movement, feature extraction, etc. These are a necessary evil and the place where most risk in a data science project resides; garbage-in/garbage-out leads to low certainty.
AI is fundamentally about using machine learning and deep learning techniques to enable applications that are built on data. Every organization that aspires to a data culture has to pick a place to start. Deep learning will make data accessible that previously was not; if that will create momentum, with a high chance of success, that is where you should start. For other organizations, better predictions and automation will beget a data culture. Regardless of which path you choose, the objective is the same: do data science faster.
For more from Rob Thomas, visit https://www.robdthomas.com/ and follow him on Twitter @robdthomas.