I write columns on news related to bots, specially in the categories of Artificial Intelligence, bot startup, bot funding.I am also interested in recent developments in the fields of data science, machine learning and natural language processing ...
I write columns on news related to bots, specially in the categories of Artificial Intelligence, bot startup, bot funding.I am also interested in recent developments in the fields of data science, machine learning and natural language processing
ServiceNow survey results shed light on the state of machine learning in business.
Earlier this month ServiceNow released the results of a survey it ran to better understand the state of machine learning in businesses today. The survey questioned 500 CIOs across 11 countries and 25 industries, providing a broad brush on the opinions of machine learning (ML). Let's take a look at some of the more interesting findings from the survey.
Eighty-nine percent of respondents stated they are using machine learning in their organization. Drilling down on that number, 40% are in the research and planning phase, 26% are piloting, 20% are using it in some areas of the business, and 3% are using machine learning across the business. Although the majority are still in the research or piloting phase, it's certainty a positive that so many IT leaders consider ML a core part of their go-forward strategy.
What CIOs are doing with machine learning is quite varied. Sixty-eight percent are using it to automate repetitive tasks, which is an ideal "dip your toe in the water" use case for it, as it's easy to measure the before and after. Fifty-four percent are looking at having machine learning make complex decisions, and 40% to recognize data patterns. Also, 32% are using it to establish links between events, 32% for supervised learning, and 31% for making predictions. Some examples of where to apply machine learning are approving loan documents, tracking contract expiration dates, and creating individual health plans.
To no surprise, security operations is the area where machine learning is most highly penetrated and expected to see the largest amount of growth. The below graphic shows currently 82% of decisions are either completely automated or automated with some human intervention. Within three years, that will grow to 99%. The chart also compares the use of machine learning in customer management, call centers, operation management, finance, and sales and marketing. Security is where almost every business is feeling the pain today and manual processes are too slow, which is why the penetration rate is so high.
Looking at the barriers to adoption, 51% of respondents cited insufficient quality of data, 48% outdated processes, 47% a lack of human skills, 41% a lack of budget, and 32% cite a lack of complex decision-making abilities by machines. These results surprised me, as I thought "lack of human skills" would be, by far, the top response. My assumption is that businesses got a taste of bad data in the piloting or researching phase. It's my belief that future success for it is based on having data and the machine learning algorithms to analyze it. However, bad data will lead to bad insights and bad decisions -- getting the data house in order is absolutely mandatory for companies embarking in ML.
With respect to organizational adaptions, the survey showed that 48% cited they needed to make organizational changes to accommodate machine learning. I suspect that as the technology is applied to a broader set of use cases, this will increase. Thirty-nine percent will need to redefine job descriptions to focus on working with machines, as 27% indicated having to recruit employees with new skill sets. One big surprise is that only 18% developed policies to ensure the accuracy of data. I would have expected this to be higher based on the fact that bad data was the No. 1 inhibitor. Again, I expect to see this number grow as the use of machine learning matures. I also think the "hire new talent" number is low based on where companies are in the adoption of machine learning. Anyone looking to pivot their career should seek expertise in this field as it will pay big dividends in the future.
For the report, ServiceNow cut the data into an isolated a group it called "first movers," which are the 10% that were ahead of their industry peers in spending on machine learning. Companies getting into machine learning now should learn some lessons from this group by understanding where they differ significantly from the larger respondent base. The main areas of differentiation are:
Talent -- 76% of the first movers have redefined jobs to focus on machine learning versus only 35% of others surveyed. Also, 30% of first movers have put in place plans for workforce size related to machine learning compared to 15% of others.
Better business processes -- 72% of the first movers have developed a roadmap of future process change versus only 33% of others. They accomplished this by developing methods of monitoring mistakes made by machines and implemented policies to ensure the accuracy of the data.
Forward looking -- The first movers are more focused on innovation than others (70% versus 54%). Also, 50% are automating routing processes and 46% are digitizing processes, compared to 33% and 27% respectively.
Strong technology -- Across the board, the first movers were way ahead on the use of new technology. The data shows:
Analytics -- 93% versus 62%
Cloud -- 96% versus 53%
Mobile -- 70% versus 50%
IoT -- 70% versus 31%
Modernizing infrastructure is clearly a key to machine learning success
High expectations -- 87% of first movers expect machine learning to grow the top line versus only 67% of others. The lesson here is to set the goals high and work toward them.
Overall the survey was certainly a positive view into the use of machine learning. If you're reading this and your company isn't a first mover, no worries, as only 10% of the audience of the survey was. However, the market is coming and coming fast, so take a lesson from that group: Upgrade your technology, make sure you have the right talent, clean up the data, and take advantage of machine learning.