Practical Facts About the Current State of Machine Learning and Artificial Intelligence
- Posted by Marketing Daitan
- On August 11, 2018
- AI, Data, Digital Transformation, Machine Learning
Where is There Actual Traction?
Recently Forbes Magazine published a great article on the subject of Machine Learning (ML) and Artificial Intelligence (AI) based on research done by McKinsey Group Institute. It covers interesting data points about investment levels and rate of adoption by industry. It’s a helpful piece because it highlights the state of the business and market, and provides perspective based on interviews with business executives.
Here are couple points that caught my attention:
- “The current AI wave is poised to finally breakthrough. Investment is growing at a high rate, but adoption remains low.” The data point behind this finding shows that the total investment in AI ranged from $26B to $39B, with large tech giants share being $20B to $30B. Start-ups represent the delta amount. What is interesting about the tech giants’ investments is that it was largely spent on their research and development (90% according to McKinsey) and the balance on acquisitions. Given the far reaching impact AI is expected to have, they invest to build AI as a core competency in anticipation of strengthening their future.
- “AI adoption is greatest in sectors that are already strong digital adopters.” The report shows that early adopters include High Tech, Telecom, Automotive, and Financial Services and the slower segments include Healthcare, Education and Tourism. It’s no surprise that early adopters are factoring their AI/ML innovations into their digital transformation initiatives. McKinsey discusses this and our experience has been the same.
How to Think About a Machine Learning/Artificial Intelligence Project
Daitan’s client base falls well within the identified early adopters, which aligns with the types of conversations we have with them, and work our Innovation team is pursuing. We believe that even with all the hype, practically speaking, fundamental best practices still apply and I’d like to touch on them:
- Start with a realistic business initiative before embarking on development efforts. Whatever the AI strategy, recognize that it should incorporate into your digital transformation initiatives. Establish an agreed-upon business metric that you want to improve and how to calculate it. As a reference, Netflix could demonstrate a clear business metric of saving $1B in cancelled subscriptions based on their ML development in Search that now delivers better results to users.
- When you have scoped an ML/AI project, it will be important to build the right team in order to operationalize it—in other words, it takes a team to support the advanced work of Data Scientists and Data Engineers so the business ends up with a viable product and result. Daitan blogged on the subject of roles and skills needed for data science.
- The key to successfully implementing is data and training the algorithm, in order to get meaningful business-relevant results. Speaking at a recent ComTech Forum event, Daitan presented that in the context of ML/AI development, there are important considerations around data and training. We talked about our experiences building high volume, high velocity data pipelines, securing them and the complexity involved when accessing disparate data siloes on legacy and/or third-party infrastructures.
- Use the right algorithm to accomplish your objective. Looking at the McKinsey data we see that the majority of investment has been in Machine Learning, albeit in a graphic that also shows a blending of technologies and applications. From Daitan’s perspective, we think of Artificial Intelligence and Machine Learning as having been around for decades; and Deep Learning (DL)—recently emerged and capturing lots of interest—is really just a subset of Machine Learning. So, as you try to determine how to use these algorithms, keep in mind that an experienced Data Scientist will be able to identify which type of algorithm will yield more accurate results, while minimizing model training costs.
What’s Been Daitan’s Experience?
Over the past 2 years, Daitan has helped a number of our clients implement advanced technology solutions, as well as provided Data Science, Data Engineering, Customer Experience and DevOps resources to successfully implement it. In our experience, companies are taking measured steps to take advantage of Machine Learning and Artificial Intelligence technologies based on clear business objectives that intelligently further growth. It’s easy to get swept away by all the hype and media coverage, it’s better to plan smart moves capable of realizing a tangible Return on Investment.