Saturday, October 1, 2022
Bringing the Latest in News Straight to Your Screen


Finding The Data: How To Avoid AI And Analytics Project Failures

By News Creatives Authors , in Small Business , at October 15, 2021

Ajith Sankaran, Senior Vice President, Course5 Intelligence.

The pace of digital transformation has accelerated, and there is no longer any question as to the massive impact rapid digitization driven by AI and data analytics has had on our world.

While many organizations are reaping the benefits of successful AI and analytics implementations, the reality is that many of these projects continue to under-deliver or even fail. This is despite all the advances in supporting technologies and the ever-increasing investments poured into the industry. Back in 2018, Gartner made a widely shared prediction that 85% of AI projects would eventually fail. Fast forward to the current day and these estimates have remained; in fact, analysts and industry leaders have continued to make similar predictions along the way.

Failures In Data

These failures and challenges have been attributed to reasons such as lack of a cohesive AI strategy, application of AI and analytics to the wrong projects, poor organizational alignments and lack of continued C-suite commitments. However, I believe that the single biggest reason for the failure of AI and analytics projects involves data. I see data failures manifesting in the following ways:

• Absence Of Data Strategy And Data Governance: While companies develop grand AI and analytics road maps, some fail to build a data strategy and set up data governance systems. If this foundation is not laid, it inevitably leads to multiple failure points.

• Challenges With Data Availability: Data silos are a reality, and there will always be multiple levels of dependencies when it comes to organizing data. Challenges emerge when there are delays in accessing and connecting with different data sources. Even if some of the cloud-native systems ensure easy data availability, integrating with legacy systems is a challenge that organizations continue to struggle with.

• Poor Data Quality: While the adage “garbage in garbage out” has become cliched, it remains a very real fact when it comes to challenges faced by organizations implementing AI and analytics. Clean, machine learning-ready data is a prerequisite for AI and analytics projects. Organizations often realize that the data they have is not clean only after they start ambitious programs.

• Lack Of Data Readiness: Even if the data is clean, it still needs to be ready or labeled for use by analytics, machine learning and AI algorithms. Unfortunately, without a sound data strategy, labeled data is often not available, and this affects the projects. In many cases, data scientists end up wasting their time and companies’ resources labeling and annotating data.

• Low Levels Of Data Literacy: With increased usage and data democratization, everyone in the organization, not just the data scientists and data engineers, must understand and use data. In other words, they need to be data literate. According to a recent Harvard Business Review article, while 90% of business leaders cite data literacy as a key to company success, only 25% of employees feel confident in their data skills. A lack of data literacy can lead to data silos, data security challenges and low levels of inclusivity.

The Way Ahead

While these challenges remain, there is no question about the growing promise of AI and analytics projects. Overall, I see that success stories in this industry all have one common theme: a well-thought-out data strategy with robust data governance systems.

According to Gartner, within three years, “75% of organizations will shift from piloting to operationalizing AI.” Many organizations have already realized the importance of a sound data strategy as well as the need for ensuring the availably of clean, analytics-ready data. Within your own operations, data engineering needs to be considered as necessary as data science and as a foundational element to the success of your AI and analytics programs.

I recommend organizations adopt data operations (DataOps) to drive enterprise-wide programs, and data engineering should play an integral role alongside data science. Utilizing DataOps can help enable better collaboration among data scientists, data engineers and other key stakeholders. This will ensure that every team is working in sync to use data more appropriately and more efficiently, which in turn leads to more accurate analysis, better insights and improved business strategies.


Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?


Comments


Leave a Reply


Your email address will not be published.