Follow Datanami:
July 9, 2019

Why You Don’t Need AI

Tim Hall

(Production Perig/Shutterstock)

Talk of Artificial Intelligence (AI) is everywhere, and the rush to implement AI is unquestioned. Studies such as those by Gartner and the McKinsey Global Institute, report that over the past four years, AI implementation has grown by as much as 270 percent. And by 2022, the AI market is estimated to be valued at $6.14 billion.

If that weren’t enough, the U.S. government launched the American AI Initiative in February of this year — “a concerted effort to promote and protect national AI technology and innovation” — which includes education and training opportunities to prepare the American workforce for AI.

But all this talk of growth and implementation doesn’t address the one critical question every business should be asking: Is AI the right solution for my current business problems?

AI, Up Close and Personal

While AI is all the rage, the fact is that the vast majority of folks simply don’t need AI to solve a bulk of the challenges they face. Besides, investing in AI isn’t as simple as implementing plug-and-play hardware or software. Before making any concrete decisions, consider the following drawbacks to adopting an AI solution:

  • Lack of AI skills – In the Gartner study referenced above, close to 54 percent of the respondents said the skills gap was their greatest challenge.


  • Shortage in AI specialists – More than 20 percent of the respondents in a Deloitte report said there was a shortage in positions such as AI software developers, data scientists, user-experience designers, change-management experts, project managers, business leaders and subject-matter experts.
  • Fuzzy AI strategy – To succeed with AI, you need a solid AI strategy based on your core business objectives, priorities and goals. You should also have an understanding of the core problems you’re looking to solve as most automation technologies that exist to specialize in just one area.

Problem Solving Without AI

Despite AI’s potential, not everyone needs AI technology to solve their day-to-day business challenges. Like buying the latest gadget with more features than you’ll ever use, bringing in AI can be overkill. Instead of jumping on the AI bandwagon, a more pragmatic approach needs to be taken. Take a step back and examine the problem from a business perspective and see what needs to be accomplished. Then, determine the kinds of metrics and events that are required to address questions as they arise or head off problems before they occur.

By simply closing holes that exist in the ways in which we observe the hardware and software stacks, sensors, and systems within organizations, much can be done with existing tools and technology — in some cases, traditional methods are actually better suited than today’s AI solutions. With time series data, for example, the vast majority of it can be effectively analyzed using the Holt-Winters algorithm — a straight-forward method to forecast and predict outcomes. And many traditional solutions don’t require the expertise and specialized knowledge that are necessary to build an AI solution, which is a critical factor given the scarcity of AI engineers and the difficulty many companies have in attracting that talent.

The Holt-Winters algorithm is useful for making prediction using time series data (courtesy: Gregor Trubetskoy)

The type of business also needs to be considered. For example, during a CIO council meeting, one of the leaders there put everything in perspective when he said he ran a grocery store and not a software design business. His employees were not, nor would they ever be, people who could understand, design and implement AI solutions. When the time came for AI, it would need to be something that could be easily implemented and used by his employees.

Your Future in AI

There’s a real danger in adopting AI without a solid business strategy or without considering the long-term impact of AI on the business or its customers. Just because you have a lot of data doesn’t mean you need to adopt AI. It may be that all that data is just a lot of useless metrics.

There’s no question that the promise of AI sounds compelling and may very well have a role to play in many industries. But for now, it exists as a nascent and emerging technology requiring specialized resources that are capable of dealing with complex technical challenges and have sufficient business domain experience to understand where it can be best applied to achieve a positive business outcome. Like many “new” technologies, there are likely to be many failed AI projects along the path to maturity. The most significant and positive near-term benefits that organizations can take advantage of as a result of the AI trend is to take a step back and analyze what metrics and events they have access to today — along with what else they believe they need to collect to answer questions that they have.

This preparatory work can lead to near-term and practical benefits while the tools and technologies required to make AI more approachable occur. Adopting AI for the hype or because it’s inevitable are not good reasons to invest right away. If you jump too soon, you might miss the boat.

About the author: Tim Hall is vice president of products at InfluxData, developer of InfluxDB, an open source time-series database. Previously he worked at Hortonworks.

Related Items:

Why Deep Learning May Not Be So ‘Deep’ After All

Why You Need to Trust Your Data

No Time Like the Present for AI: The Journey to a Successful Deployment