The road to AI adoption is far more complex than one can imagine. Building data science models and testing them is only one piece of the puzzle. To understand the roadblocks and best practices, Wayne Eckerson invited Nir Kaldero in our latest episode to learn why organizations need to start paying more attention to people, culture and processes to make data science projects a success and how democratization skills pays off in the long run.

Nir Kaldero is the Head of Data Science, Vice President at Galvanize Inc. and the creator of the GalvanizeU Master’s of Science in Data Science program. A tireless advocate for transforming education and reshaping the field of data science, his vision and mission is to make an impact on a wide variety of communities through education, science, and technology. In addition to his work at some of the world’s largest international corporations, Kaldero serves as a Google expert/mentor and has been named an IBM Analytics Champion 2017 & 2018, a prestigious honor given to leaders in the field of science, technology, engineering, and math (STEM).

Key Takeaways:

• Biggest challenges that organizations face for AI adoption are people, culture and processes.
• If we want to make the AI revolution work in the near future, we need to make sure that we democratize the skills to everyone in the organization
• The business executives should know how data science workflow functions.
• 60% of AI/Data science projects are not being implemented because the executives are not thinking about the steps to implement them.
• Business leaders need to be educated on how to operationalize AI and Data Science projects.
• Executives should know how to evaluate models and decide if the model is going in the right direction.
• Cross-pollination between experienced data scientists and inexperienced business executives (in technical terms) and vice-versa pays off on data science projects
• To bridge the gap between technical and non-technical teams, introduce new personas who will act as Data translators.
• Data translators aren’t necessarily data engineers or business analysts. They should have knowledge about top-level business goals and should know how to code in Python. They make the data ready for production.
• Data scientists should be responsible for creating the data product and data engineers should be responsible for scaling those products
• To close the skills gap, allocate budget to out-scale the talent within the organization instead of hiring new talent.
• Uber democratized machine learning successfully on their platform
• Data science automation is good but we still need data scientists to tune the models

Below is one question and answer from the podcast

Wayne Eckerson: What are the biggest barriers for adoption of AI and data science in most organizations?

Nir Kaldero: It’s a very good question. When I think about AI and specifically the AI adoption, there are three different principles. The first one is technology, the structure, and the capabilities. The second one is all about the data and the third one is the bucket that I typically advocate- People, cultures, and process. Having worked with Fortune 500 companies, I would argue that people face the biggest issues when it comes to people, culture and process. Data Science and data engineering are a top-down initiative. The biggest gap we see is around executive knowledge. If you go ask them in plain English, ‘What is Machine learning?’, ‘What’s the purpose of AI or how AI can revolutionize their business?’, they will typically have a hard time to answer the question. When we talk about the necessary changes that should be made to the cultural aspect, it should definitely come from the top.

The second is how to bridge the gap between technical and non-technical teams to work seamlessly together. The business side think of themselves as marketers, think themselves as account people or operations people. On the other side, we have the data scientists and the data engineers. These two different personas don’t necessarily don’t know how to communicate seamlessly. We, the business leaders that are leading the strategy and the growth of our organization need to figure out how to breach those gaps between the two. When the time comes to operationalizing projects in AI, Data Science and Data Engineering, the business can react seamlessly and the two personas can work together in order to make this model work better.

The third one is the skills gap. Everyone is talking about the skills gap but I think beyond we need to look beyond that. Currently, there are a lot of data scientists and data engineers that have a Ph.D. or come from a hardcore stem degree. If we want to make this AI revolution work in the near future, we need to make sure that we democratize the skills to everyone in the organization. So everyone can actually benefit that come from leveraging the data and creating modeling techniques and understanding the interesting predictions to enhance the decision-making process.

We need to start thinking about how we make these skills relevant for everyone.

Learn more about Nir Kaldero here and more about Data Science for Executives here, and connect on Twitter (@nirkaldero) and LinkedIn (www.linkedin.com/in/nirkaldero/).