Despite teething problems, artificial intelligence (AI) has become mainstream. In fact, it is more than mainstream. It's inevitable. That is to say, no matter how enterprises set up their technology infrastructure, it seems unlikely they will remain competitive without AI. A recent report, IBM’s 2021 Global AI Adoption Index, underlines this.

Based on a survey of 5,501 businesses globally, the report shows that one-third of companies are currently using AI in some way, while 43% are exploring it. However, problems remain. While recent advances are making AI more accessible than ever, the survey found that a lack of AI skills and increasing data complexity are top challenges. There are five take-aways from the research:

  • Business adoption of AI was basically flat, but companies are planning significant investment.
  • COVID-19 accelerated how businesses are using automation.
  • Trustworthy and explainable AI is critical to business.
  • The ability to access data anywhere is key for increasing AI adoption.
  • Natural language processing is at the forefront of recent adoption.

A large majority of investments continue to be focused on the three key capabilities that define AI for business: automating IT and processes, building trust in AI outcomes, and understanding the language of business, the research showed.

Artificial Intelligence in the Mainstream

Other research indicates just how far AI has come in the past few years. Just this week, research conducted by San Mateo, Calif.-based Freshworks looked at the state of IT service management (ITSM) and IT operations management (ITOM) and found that AI technology has hit the mainstream, with 93 percent of businesses currently exploring or deploying some level of AI in ITSM.

That research showed that most organizations expect AI to be deeply integrated within ITSM tools instead of it being an add-on that requires additional effort to engage employees. Among the findings are:

  • Virtually all IT managers (93 percent) are currently exploring or deploying some level of AI technology for ITSM/ITOM modernization.
  • Nearly 70 percent of IT managers say AI is either critical or very important for upgrading and modernizing service desk capabilities.
  • Among six AI use cases, ITSM chatbots are the clear leader in planned or actual AI deployments.

In terms of what organizations hope to achieve, the principal objectives are:

  1. Speed of implementation (40 percent).
  2. Integration with legacy systems (40 percent).
  3. The overall cost of implementation (38 percent).
  4. Training AI bots to return the most accurate response (39 percent).

Related Article: Why Artificial Intelligence Won't Replace the Human Workforce

AI Is Already Baked Into Business

Not only has the use of AI in the enterprise reached the mainstream, it also seems that more enterprises are starting to depend on it. Wayne Butterfield, director of ISG Automation, a unit of Stamford, Conn.-based technology research and advisory firm ISG, said it should be no surprise to see organizations adopting or experimenting with AI.

With so many AI components, ranging from machine learning to natural language processing, and lots of AI use cases in between, it's likely that actual AI usage is even higher than current surveys suggest. AI, in one or more forms, is already built into many of the widely used enterprise platforms, Butterfield said. He cited the example of chatbots using natural language processing in platform tools like SAP, Salesforce and Workday as examples.

“Natural language processing, machine vision and machine learning are just a few of the ways AI is so prevalent in the enterprise today, and why AI will continue to be important moving forward across all industries," Butterfield said. "Reading and responding to emails, conversing on WhatsApp, extracting a clause from a contract or even assisting in the picking of ripe fruit are all examples of AI currently in use across the enterprise.”

Related Article: Where Automation Is Being Used in the Digital Workplace 

Learning Opportunities

The Role of AIOps in AI Deployment

At the heart of any AI deployment is artificial intelligence for IT operations (AIOps). AIOps uses big data, analytics and machine learning to enhance IT operations and is inevitable for forward-thing organizations. Most modern enterprise IT environments consist of a complex mix of on-premise and cloud environments which run a wide variety of dynamic workloads that require frequent reconfiguration and scaling up and down, said Atul Varshneya, vice president for AI at Santa Clara, Calif.-based Tavant.

These applications and other IT systems generate a massive amount of data, and the data volume keeps increasing as IT environments evolve. Applying analytics and machine learning can help companies extract information from this data to make smarter decisions. For example, enhanced visibility into performance and dependencies across all environments can provide insight into significant events related to slow-downs or outages and automatically alert IT teams about problems and their root causes.

“Through intelligence enabled by analytics and ML techniques, rich information is extracted from the data generated by applications, and systems including various monitoring mechanisms," Varshneya said. This results in several benefits:

  • Proactive management of potential issues: AIOps can predict problems well in advance, enabling IT personnel to resolve them proactively.
  • Faster resolution of identified issues: AIOps can also provide rich information about problems through the explainability of its prediction to help identifying root causes and get to resolutions faster.
  • Efficient IT operations: With theses capabilities, alerts that require an urgent response can be reduced significantly, leading to more uptime and overall higher performing IT operations.

4 Steps to Make the Most of Artificial Intelligence

So how should enterprises proceed? There are four things enterprises need to consider on their way to AI adoption, said Sam Babic, senior vice president and CIO at Westlake, Ohio-based Hyland.

  1. Start small, build momentum: Look for a high value, low complexity problem to solve or decision to make with AI as a starting point. This is also true when tackling projects at the organizational level. Demonstrate small, tangible wins to gain momentum for AI initiatives and then build momentum from there.
  2. Create an AI/data center of excellence: In the formative stages of AI adoption, it is good to set up an AI center of excellence where subject matter experts either report directly or through the dotted line. This center of excellence provides focus and dedication to the topic and allows a centralized approach to patterns and practices derived through learning. Likewise, the purchase of tools and other decisions can be consolidated. As it grows, the center can expand into a community of practice with stakeholders throughout the organization.
  3. Understand the outcomes you want: Oftentimes, organizations focus on understanding the opportunities AI can unlock and then map them to organizational goals. Instead, start with the organizational goal first and then map how AI can help. This seems like a nuance, but the latter approach enables the organization to more quickly focus on requirements necessary to accomplish the goal vs. getting lost in a sea of possibilities.
  4. Be careful of bias: "Garbage in, garbage out" is a long-standing term that is even more important when leveraging AI. Operating with bad data, whether it is stale, incorrect or skewed will yield bad decisions. Training a machine learning algorithm is like training a child. Teach them bad habits and they will execute those bad habits. Closely analyze and clean data to ensure human bias is removed from training.

Proof of Value Replacing Proof of Concept

Artificial intelligence is more than just a nice addition to the technology stack, it's essential if companies want to survive, said Bruce Orcutt, vice president of product marketing at Milpitas, Calif.-based ABBYY. And it's getting more sophisticated.

“COVID was definitely an accelerant but also the developer skills shortage is a contributing factor,” he said.

Orcutt pointed to the example of document processing. AI technologies like optical character recognition and machine learning have been used to intelligently capture documents and send content to enterprise applications for years, but they required significant training and IT resources from IT. Now, more advanced AI makes that same legacy document processing technology easy for business analysts to use in the form of cloud-based, no-code platforms that can process any type of content they are working with.

“For AI to see rapid adoption, it needs to be user friendly, deployed quickly and return value immediately," Orcutt said. "The days of 'proof of concept' are gone, enterprise leaders want proof of value now."