Making Artificial Intelligence Chatbots Pay Off

Abstract: Artificial intelligence is infiltrating business operations. ChatGPT and similar programs can improve the efficiency of everything from customer service to long range business planning. Impressive as they are, AI-generated materials must be checked for accuracy and for conformance with company values.
Artificial intelligence is invading the business world. Companies of all sizes are experimenting with ChatGPT and similar programs to improve their performance in such areas as sales optimization, inventory management, business planning, customer service, and fraud detection.

“Artificial intelligence is enabling a new way of working, communicating, and sharing information throughout all business operations,” said Jake Maymar, VP of Innovation at the Glimpse Group, a technology company working with generative AI (theglimpsegroup.com). “It is a new age dawning that will extend the technological advances that began a few decades ago when the computer replaced the printing press.”

The benefits of AI accrue not just to large corporations. “Smaller businesses are often laser focused on selling, closing deals, and working productively with customers,” said Patrick Delaney, a software engineer specializing in artificial intelligence. “AI is well suited to enhance the efficiency and potency of those very business activities.”

Minding the store

While AI is a promising venture for most any enterprise, managers must ensure that unmonitored technology does not damage a company’s operations or tarnish its image. Especially in these early days, skepticism is as valid an attitude as enthusiasm.

Managers must keep their eyes open for several potential problems. The first is accuracy. Given the impressive size of AI’s resources, humans are apt to place too much trust in its output. In fact, AI often produces many factual errors which technologists call “hallucinations.” Experts warn that everything generated by the new technology must be validated by a human.

Unchecked errors, of course, can damage any company’s reputation. But a second problem is a lack of empathy that is all too evident in much of the output that emanates from AI programs. Customers and employees can be irritated when AI disrupts the personal, one-on-one human interactions that can create robust feelings of loyalty. “Over reliance on AI can lead to a missing out of human intuition,” said Delaney.

Third, no matter how “smart” the output from a chatbot program, it remains true that an experienced human being can develop an affinity with customers that can only be achieved through years of productive interactions. “Someone who has operated as a service professional for 20 years will have vastly more knowledge and customer savvy than an AI program that has been trained on three or four years of data,” said Piyush Tripathi, lead engineer at Square. “Businesses need to balance what is produced by the machine with what is produced by the human.”

Finally, managers must constantly address AI activity in light of the company’s current posture with employees and the public. Do the initiatives created by AI align with company values?

“While AI is objective in its analysis and conclusions, it is not aware of how to assess values and why they are important,” said Tripathi. “A project plan may be perfect in the eyes of a robot, but a human can look at it and instantly know it’s not going to be acceptable to the employees or the public.”

Improving performance

So where can an investment in AI return the most bang for the buck? Here are some areas of business operations for which chatbot programs seem especially suited:

• Customer service. Chatbot programs that handle customer problems are advancing far beyond the familiar, simple-minded web helpers of the past. Rather than accessing a limited database of information and spitting out canned responses to customer questions, the new programs handle inquiries in creative ways, using natural language. And their answers are informed by all of the information they have absorbed from the Internet.
• Advertising and marketing. Artificial intelligence can develop complete marketing campaigns, starting with themes and moving on to story boards, promotional text, and even artwork and videos.
  “An AI program can do the work of a creative director and even perform A/B testing,” said Maymar. “It can speed up such projects dramatically, completing in maybe an hour what unaided humans might require two weeks to accomplish.”
• Performance reviews. AI Programs can complete the foundational work of employee evaluations by comparing worker achievements against company metrics. They can also produce reports that reveal performance trends over the years.
• Project planning. “Thanks to its sophisticated algorithms, AI can create smart schedules that analyze the various parts of a project and identify the often-hidden dependencies that must be addressed to avoid unexpected delays,” said Tripathi. “Very often it’s hard to do all of that manually, because there are so many loose connections and variables that must be kept in mind.”
• Such programs can analyze worker and resource availabilities, equipment capabilities, and budgetary limitations when planning timelines for peak performance. The result is often a more objective plan, noted Tripathi. “These programs can help avoid the natural bias that can arise with the tendencies to go with what worked in the past or what is familiar, as opposed to what is logical.”
• Training programs. “AI programs can make the task of developing staff educational modules much easier,” said Phil Siegel, founder of CAPTRS, a nonprofit organization working with artificial intelligence (captrs.org). In the past, businesses would write static manuals that required constant updating over time. Modern AI programs can create dynamic, interactive modules built upon foundational performance parameters provided by company management. “Training initiatives can be made more specific than in the past, responding to the needs of people who are having trouble with certain parts of their jobs,” said Siegel.
• Supplier and customer evaluations. Will a prospective customer be able to pay their bills? Will a vendor close up shop before delivering critical materials? “Because AI can analyze data so deeply, it can figure out that something is likely to go wrong at an organization long before a human can spot problems,” said Tripathi. “This allows the business to take the appropriate action to avoid a hit to the bottom line.”
• Business plans. What is the best course of action for a business over the next three years? Into what new markets should it expand? “Artificial intelligence can lay out a potential business direction, along with an appropriate marketing plan,” said Maymar. “It will provide stimulating ideas and a clear path to follow, even if current management lacks an MBA or formal business education.”

Related: Strategies for Success in the Challenging Economic Landscape of 2024

Laying groundwork

While AI can increase company profitability by lending efficiencies to all of the above areas, the fact remains that a watchful eye is essential for ensuring that the activities of the new technology are accurate, empathic, informed by human experience, and aligned with company values. “We should not think of AI as some kind of magical process, because then we would lack the understanding of its internal operations necessary to criticize its performance,” said Delaney. “We need to know that it has limitations, that it can’t give all the answers, and that there’s still value in cross-checking everything it does.”

Experts advise using a two-step process in every area of company operations. That means letting AI do the foundational work, and then having a human check for errors of fact and nuance.

Some examples will make this clear. Despite their sophistication, AI-fueled customer service interactions can still seem tone-deaf. That’s why a human representative must usually step in after the AI program has completed such foundational work as performing an initial analysis and assessing metrics.
With promotional initiatives, too, experienced humans need to come in and check the work provided by artificial intelligence. “While AI programs may seem valid on their surface, very often they lack the human element that makes such initiatives work.,” said Maymar. “Without the human touch, they can seem soulless.”

In the area of performance reviews, employees will benefit from one-on-one conversations that discuss why targets were achieved in some areas and not in others. “The human manager needs to provide the encouragement that an AI program cannot,” said Maymar. “Personal interaction is necessary to transform an objective review into an opportunity to grow, coach, and share knowledge.”

Getting started

While AI holds promise for any business enterprise, it can be difficult to get the ball rolling. Experts advise starting with simple projects, encouraging everyone to participate. “Figure out a way to make AI a common part of daily operations throughout your business,” advises Delaney. “Schedule learning periods so the staff becomes familiar with the system.” Delaney compares the role of ChatGPT to that of Microsoft Word twenty years ago. What was a novelty then has become a routine part of the business world.

Experimentation will be necessary, and that can lead to some failures. Businesses need to keep in mind that if a certain operation doesn’t work particularly well with AI right now, it may do so in the future. “AI programs get smart really fast,” said Joe Karasin, an AI consultant and SEO specialist. “As you continue to train the program, it continues to improve, and gives you better, more targeted information.”

A major driver for the improvement of AI performance is the ability of the human operator to write effective questions. “AI learns from the corrections an individual makes to their prompts, and with repeated iterations does a better job of providing productive results,” said Karasin. “At the same time, the human operator learns how to fashion prompts that bring productive responses.”

While the AI learning curve may seem daunting, it’s wise for every business to get an early start, even at a basic level. “Managers need to become aware of how AI works, and how it can be used as a force multiplier for business capabilities,” said Maymar. “Everyone is going to be using AI pretty soon, and you don’t want to be the one left behind. The best thing to do is to ride the wave now instead of being buried in the crest as it goes over.”

About the author. Phillip M. Perry is an award-winning freelance writer based in New York City. His byline has appeared over 3,000 times in the nation’s business press. He can be reached at https://www.linkedin.com/in/phillipmperry/