Artificial intelligence (AI) had a big year in 2016, with massive strides forward in deep machine learning, IoT, predictive forecasting, more intelligent business processes and smart personal assistants. While technology seems to move at an ever-increasing rate, AI is seemingly everywhere and may prove to be one of the fastest-growing technology segments in 2017.
That’s why organizations need to determine how exactly they need to prepare to foster AI in the year ahead. For this purpose, there are several key aspects businesses should consider to understand how this new technology will disrupt traditional business models and how to integrate AI in their strategic plans.
1. Clean data: If you feed AI bad data, it can’t learn or make good decisions (there's too much noise to find the signal).
Machine learning algorithms are the backbone of AI and require data to learn from. Today, many organizations are producing more data than they know how to use, but that data often sits in separate silos or is unstructured and difficult to analyze. In order to get insights from this data, AI analyzes the data sets to find predictable patterns. The problem, though, is that if these machine learning algorithms are fed bad data, these systems can find incorrect patterns, create bad models, or often end up useless because the system can’t find the value in the data.
A study by Xplenty highlights that 30% of business intelligence professionals spend at least half of their time cleansing data. While AI-powered tools can help analysts to cleanse data with algorithms written to clean up, sort, and categorize it, some companies (such as IBM and Amazon) still use humans to do data labeling that software can't handle. This means that it’s too early to eliminate humans from the processes of providing AI tools with the right data at the right time.
AI is no silver bullet. You have to get a handle on your data before AI can be beneficial.
2. Integrated data/data warehousing: Connect the data from different silos to enrich AI.
Businesses that can’t connect critical data sources are losing huge opportunities that have a direct impact on the bottom line. According to EMC (with research and analysis by IDC), "by 2020 the digital universe -- the data we create and copy annually -- will reach 44 zettabytes, or 44 trillion gigabytes."
Business leaders need to make sure that all the valuable customer data that comes from various systems is centralized and integrated. That’s why big data systems like Hadoop and Amazon Redshift, which are capable of uniting massive volumes of data, continue to grow rapidly. These solutions are typically built with powerful APIs or data integration capabilities, especially SaaS solutions like Amazon Redshift, which help organizations integrate data from various enterprise transactional, sales, marketing and service systems.
More unified data powers better organizational analytics that drive better manual decision making and analytical reporting, but are also critical to educating machine learning systems and creating more well-rounded AI. Integrated data means holistic analysis, and most importantly, well-rounded recommendations.
3. Identify the starting point: Use cases are great, but think about deploying AI strategically.
The operational deployment of AI is an essential step for any organization, and like any major digital initiative will require a well-considered strategy. This means evaluating your IT infrastructure, the business processes that can be impacted by AI, and most importantly, where AI will be able to help you impact the business most.
The strategy must also align with other organizational priorities and strategies to reduce disruption and improve adoption quickly. AI must be implemented seamlessly without negatively impacting customers – whether that means marketing, sales, orders, customer care, technical support or in-person operations. In addition, the strategy should be able to work in concert with all of the company's interaction channels, including call center, web or mobile self-service, as well as sales and service partners, in order to ensure all channels are delivering a consistent experience.
For example, if data from one source or channel is critical to AI making better decisions or recommendations, you’ll want to make sure that the data from these touchpoints can be consumed quickly enough to inform your AI initiative and not disrupt the experience. You wouldn’t want your system making “intelligent” recommendations to do something that the customer has already done.
In summary, AI offers tremendous promise for improving the efficiency and productivity of teams that interact with customers, whether this is automating routine tasks for sales and service teams or recommending better content or target segments for marketing. Using these three key considerations, AI can make it possible for business users to find the signal in the data without being as dependent on BI teams for help.
This article originally written in Forbes by Katherine Kostereva