If you’re only using AI for chatbots, you’re falling behind

Business and tech leaders need to combine forces to convince skeptics that artificial intelligence should be doing more picking low hanging fruit.

Image: iStock/metamorworks

More about artificial intelligence

Chatbots and call center support are table stakes when it comes to implementing artificial intelligence, according to a new report. Cognizant’s Center for the Future of Work found that leaders are using the automated analysis for decision support, automated reasoning, text mining, and object and speech recognition.

The research includes a survey of 1,000 business leaders that measures how organizations view the potential of AI and their plans for deploying AI-enabled tools. The data modernization report describes a vicious cycle that keeps companies from doing more with AI.

A company has a low level of trust in the potential of artificial intelligence. This leads to a limited application of the technology, which in turn makes it hard to get the full potential of the technology. This perceived failure reinforces the low levels of trust.

Leaders are more likely than beginner or intermediate users to use AI for these activities:

Predicting outcomes
Understanding unstructured data
Problem-solving
Gathering real-time intelligence
Finding new insights

The survey found another differentiator among AI leaders: more sophisticated verification mechanisms. Leaders were significantly more likely to to run auto-regression tests, deploy generative adversarial networks and build explainable AI verification dashboards as compared with implementers and beginners. About 80% of leaders use those techniques compared with less than 40% of companies in the other two groups.

Ben Pring, vice president, head of thought leadership and director of Cognizant’s Center for the Future of Work, said sound verification mechanisms can help companies move from a vicious to a virtuous cycle.

“Since building trust plays a critical role in this process, good verification mechanisms are an important part of a consistent strategy to make the most of a company’s AI investment,” he said.

Pring said that in his experience the best way to build trust with AI is a use case in which both the outcome of using AI and not using it are immediately clear. He sees predicting outcomes as the most promising way to accomplish this.

“If the AI indicates: ‘If you implement change x, your sales will go up by z%,’ and you follow its suggestion and obtain an increase similar to what was predicted, then it is a straightforward conclusion that you were better off than had you not used the AI or followed its recommendations,” he said.

How to build human-centric AI

Another AI priority for leading companies is using AI to augment human decision-making and provide a second opinion. The report included a comment from a banking executive who said AI helps improve customer support due to the ability to examine data from multiple sources quickly. which leads to “lower costs, faster decision-making and less complexity.”

Pring said that “human-centric” describes how AI is deployed. There are two ways to do this. The first is to make human welfare and integrity a priority when developing and deploying AI.

“In this report, our focus is on the other definition, which is to use AI to empower humans to do their job better and/or to focus on areas in which they will add more value,” he said. “The point is that AI adds more value to decision-making by augmenting workers than by replacing them.”

The report used Chorus.ai as an example of this type of AI deployment.

“Their tool uses natural language processing and computational linguistics to identify important topics or pain points that emerge during sales calls,” Pring said. “Based on these insights, sales representatives can elaborate improved approaches for their next conversations.”

How to revamp your AI strategy

Cognizant recommends taking these steps to refine an overall AI strategy:

Develop internal AI champions across the company to counteract skeptics in the C-suite
Present case studies that highlight specific areas where AI can improve decision-making
Define decisions that need AI support
Go beyond low-hanging fruit
Focus on asking questions with a higher degree of complexity
Use AI to generate new ideas for humans to explore
Use AI to provide first-level analysis for human decision-making

Pring said that top executives from tech business units and other areas should present AI case studies together.

“This is based on survey data showing that executives from non-tech areas, though may be somewhat sceptical regarding the use of AI in general, tend to understand the benefits for their own areas of work,” he said.