As artificial intelligence infiltrates almost every professional industry, companies are increasingly tasked with the decision of whether to incorporate it – and how.
Most every company has a website, but not every company’s website features an AI chatbot, which begs the question: Are AI chatsbots a necessary investment as the technology evolves?
“I don’t think there’s any particular industry that couldn’t benefit from AI enablement in their workflow,” said Max McCoy, executive vice president at Hartland-based website design and development firm Lanex.
With 16 years of tech-focused experience – from software, code testing and bug fixes to serving as the company’s head engineer – McCoy has seen firsthand the dramatic advances in AI technology and the ways that companies are realizing its potential. Still, many are not yet sure what to do with it, he said.
Several other website design and development companies as well as AI developers agree that it is best to invest in AI now with the rate at which the technology is growing and evolving.
Deciding to add an AI chatbot
AI chatbots are most prolific in the world of e-commerce. They are often used to replace or assist with the duties of a customer service representative. Within the past seven to eight years, however, both B2B and B2C companies are showing increased interest in implementing AI chatbots in their websites, according to Lori Highby, founder and owner of Milwaukee-based digital marketing firm Keystone Click.
Highby launched Keystone Click in 2008 and has built a multidisciplinary team that offers research driven web development and AI-enabled marketing strategies. Her experience in the field, however, dates back 20 years.
“The technology and the expectation of the customer has definitely evolved,” Highby said.
Keystone Click dedicates roughly 40% of its business to website design and the remaining 60% to marketing. Among its website design clients, roughly 25% are interested in implementing AI chatbots on their websites, Highby said.
Once a company decides to add an AI chatbot to its website, it must then decide what function it will serve. There are two primary forms of AI chatbots to consider – rules-based and automation-based.
Rules-based chatbots will respond in an “if, then” manner, which is mostly used on e-commerce websites like Amazon. Automation-based chatbots can be given a personality and more often resemble a model like ChatGPT or another large language model.
Keystone Click primarily services manufacturing and construction companies. Leveraging AI chatbots on their websites has been successful in giving customers the instant gratification they seek, Highby said.
“People make buying decisions based on how fast someone responds to them and the (confidence) that the right answer is being provided,” she added.
The ability to field frequently asked questions and provide immediate answers to clients has enticed several of Keystone Click’s clientele to use AI chatbots.
Vetting chat architecture
After deciding whether to add a rules-based chatbot or an automation-based chatbot, a company must decide how deeply the system will operate within the company. The choices can be narrowed down to two models, one hosted by a local server and one hosted by a cloud server or a third-party source. Each comes with their own set of risks and benefits.
Companies that choose to operate their own open-source LLM (large language model) on a local server will have more control over the system but will likely be tasked with more upkeep. Maintenance of the system will require the company to update not only the LLM, but also the system’s knowledge of the company’s data.
“It’s kind of like your brain in a jar on your shelf,” McCoy said. “It gives you a lot more control and specificity when you’re deliberately restraining it to focus on a particular set of data, even if it’s a very vast set of data.”
Companies that choose to leverage the same kind of model on a cloud-based server operated by a third-party system may have less control of their data but will not be required to update the system on their own.
Companies must also vet the model options for security liabilities, customer needs, return on investment and desired business outcome, McCoy said.
Determining what kind of chatbot the company will feature can affect its implementation, the risk mitigation process and the cost. Law firms might use a chatbot to answer a client’s preliminary consultative questions while chatbots for manufacturers may process documents or mine historical data for the purposes of strategic business analysis, McCoy said.
More involved models that house proprietary information will require more upkeep and cost to maintain.
“You don’t want a bot to hallucinate bad legal advice,” McCoy said.
Maintaining an AI chatbot
As AI evolves and new features are developed and adopted frequently, companies utilizing its capabilities will need to keep up with the maintenance of their systems.
Regardless of the technology a company uses, leaders must walk in the customers’ shoes and test the infrastructure regularly for glitches and tweaks, said Highby.
“Whenever you’re leveraging this type of AI technology and it is customer facing, you should have a consistent cadence of testing it to make sure you’re satisfied with the outcome that it’s creating,” she said. “Don’t just set it and forget it.”
The future of AI chatbots and human intelligence
Among the many concerns employees and the public may raise about AI, the lack of human interaction often makes its way to the top of the list. Employees worry about their jobs being displaced by the capabilities of AI, while consumers worry about the quality of support they receive from a company using AI chatbots.
To quell some of the fears of those who are skeptical, companies need to be transparent about how they are storing information and what they intend to do with it, Highby said.
“If you’re trying to hide the fact that you’re using AI, that’s going to hurt you more than help you,” she added.
Companies also need to consider ways that AI can augment their business and displace work rather than disrupt workflow or human efficiency.
“If these services can provide a significant cost savings for your organization, think about ways that you can reinvest in your employees and reconfigure your business,” McCoy said. “I think it could be a great opportunity for positive reinvention.”
One of the key opportunities of the evolution of AI chatbots is the ability to program them with deep company knowledge beyond the decision-tree model that many e-commerce companies are currently using.
“It is a big leap forward in terms of customer service,” McCoy said.
