Browse Our Archives

AI: Opening Pandora's Chatbot

AI: Opening Pandora's Chatbot
  • Push to The Latest: No
  • Show in The Latest?: Yes
  • Not Available Online: hide

Sep23 ai 01

Artificial Intelligence is your new best friend, and it has the potential to be your new worst enemy. 

With the relatively recent release of several AI services, such as Google’s Vertex AI and ChatGPT, AI can automate tasks, analyze vast data sets, simulate human cognition, optimize processes, and assist in decision-making. It can turbo-charge automated customer support through sophisticated chatbots, analyze medical data for patrollers, simulate weather patterns for snowmaking, offer visitation predictions, optimize transportation and grooming machine routes, and even assist in financial planning.

It’s boundless, exciting, overwhelming, and frightening at the same time. It’s a new frontier complete with good guys and bad guys. For every positive application of AI, someone out there is conceiving a subversive application, like the annihilation of all humankind, for instance.

While it’s easy to see immediate benefits, there is a galaxy of risks that are near impossible to foresee. So, proceed carefully to explore the benefits while avoiding potential pitfalls. 

WHAT IS AI?

Let’s ask AI. (Note: We used ChatGPT for this exercise, but it isn’t the only service of its kind. Simply put, where once there was “an app for that,” there will soon be an AI tool for nearly any task. Microsoft, Google, Amazon, IBM, and Salesforce have similar “language models”—essentially, systems of generating human language using algorithms and language data analysis to identify patterns and predict coherent text.)

I typed the question, “What is AI?” into the ChatGPT interface. 

In a second, it returned an answer: “AI refers to the development of computer systems that can perform tasks that typically require human intelligence and encompasses a wide range of technologies and techniques that enable machines to learn from data, reason, and make decisions or predictions. These systems are designed to mimic human cognitive abilities such as speech recognition, problem-solving, pattern recognition, and decision-making.”

What it left out: any potential downsides to its existence. Many humans in and out of the AI business acknowledge the huge potential for mischief; more than 1,000 AI leaders and developers have signed a letter that proclaims AI systems with “human-competitive intelligence” pose profound risks to humanity. “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war,” the letter warns. That has led many experts to ask governments around the world to regulate AI ASAP.  

AI IN THE SKI INDUSTRY

With all that it can do, what AI can’t do (yet) is identify what resort leaders should and should not be thinking about as they head into this brave new world. For that advice, SAM spoke with Gregg Blanchard, VP of marketing at Inntopia; Dave Gibson, founder and president of Propeller Media Works, a digital marketing agency; Jamie Ippolito, owner of resort software developer Dirigo Valley Systems (formerly Dirigo Design & Development); Mike Lannen, founder of website and digital marketing firm Eternity; and ski industry marketing veteran Dave Tragethon, who recently launched his own firm, Tragethon Consulting.   

THE BENEFITS

There are boundless ways to integrate AI. On the most basic end, users—managers, communications staff, IT techs—can simply ask it questions or have it write content. It can produce code, brainstorm ideas, and more. On a more advanced level, developers can integrate AI services as an Application Programming Interface (API) into a company’s existing applications, products, or services, which would expose information mined by AI to other digital services or software programs used by a resort, across multiple channels. 

Data analysis. Lannen, Ippolito, and Gibson touched on that point at an educational session on AI at the 2023 Ski Vermont annual meeting in April, where they highlighted how the technology can help process and analyze data. 

“My general high-level take on how resorts can use AI is that they’re ultimately going to be leveraging tools and apps that are coming from vendors or suppliers that they already use,” says Ippolito. “This includes e-commerce suites, customer relation management systems, cloud storage, those types of things.”

Other possible applications, Lannen says, include analyzing past and present weather and snowmaking data, visitation data, financial data, maintenance records, and even incident reports. 

“Anything that you could put inside a spreadsheet, you can have a conversation with. Maybe you export every single e-mail communication with your guests, or all the comments on your social media posts,” Lannen adds. The difference between AI and, say, a software program? “AI can then generate a result of whether people are generally happy, or how they are feeling about your brand.” 

Learning machines. With human input, AI models adjust, becoming “smarter” or more accurate and capable over time—as in days and weeks. By the time you read this article, the website chatbot you interacted with just a month ago will be as outdated as a Sony Walkman. 

Early version chatbots worked more like interactive Frequently Asked Question (FAQ) pages, Ippolito explains. The user typed in his or her question, and if it hit on certain keywords, the chatbot would provide the pre-programmed answer. Today, he says, AI is taking over the conversation and providing more specific (and relevant) answers.

AI-powered chatbots enable users to interact just as they would with a human. Some models can even detect and respond to human emotions. They can provide the user with an answer or recommendation to virtually any question, and point to websites and other resources. With AI, the chat feature “learns” with every guest interaction. What’s more, it can also learn through any past interactions captured digitally. 

“That’s the big news: this learning model understands language, common language, and you don’t have to know how to write code,” Tragethon says. “It can follow every one of those individual conversations and find the commonalities.”

(For more about AI-led chat platforms, see “How Can I Help You?” SAM, January 2023.)

Tedious tasks. AI can remove the tedious part of many tasks, including answering the morning snow phone a few hundred times, and it does so without complaining, a coffee break, or falling on ice in the base area and filing a workers’ comp claim. 

“Just to be able to answer those questions at scale is going to be a really powerful thing for something as complex as a resort,” says Blanchard. “But there’s work that needs to be done before you turn that on, because if it’s not giving you really good information, there’s some potential for clumsiness and awkwardness before it [begins to work correctly].”

Garbage in, garbage out. Like SEO, there are ways resorts can optimize how AI pulls data from their website or elsewhere, including how data is assembled. “Structured data” follows a standardized format, has a well-defined structure, complies to a data model, and is easily accessible by humans and programs (think: banking/transaction information or customer names and email addresses). It refers to organized and formatted information presented in a consistent and predictable manner, which allows for easy analysis, retrieval, and interpretation.

Ippolito says AI needs to have enough data that’s structured, labeled, and modeled for it to be able to intelligently provide the guest with, say, a list of the top five places to dine.

“What AI is really good at is taking tons and tons of structured data that is defined in some way in the language model and providing results back quickly. If you feed it enough data and you structure it and you label it and you define it, it can be really powerful,” he says.

Human touch. The problem with most AI tools today is that this structure means their outputs are still pretty generic. That will change with time, but for now, Gibson says, the caution with AI is to not leave your guests feeling like your ski area lacks the human touch. 

“I’d be very careful about how AI is used when it comes to direct conversations with guests,” he says. “When you’re chatting live with a guest, the risk factor increases, but so does the opportunity factor. There’s a risk of missing an opportunity to lend an empathetic ear or upsell a vacation or pass product.” 

Gibson says he envisions a hybrid chatbot solution that utilizes AI to answer basic, repetitive questions, but with a human monitoring the conversation and “looking for value-add opportunities to close the deal and answer those more sensitive questions.

“It’s still important to hear that the mom on the other side is nervous and only needs to be assured to enable them to take that next step,” he says. 

That said, AI is becoming ever more human-like. It can “train” on your company’s specific tone, language, and style to produce content that resembles a brand voice, or even a local accent. Ippolito says his team has developed a prototype that can autogenerate both the written and verbal snow report in whatever voice the user chooses. 

“You can feed AI data on your brand voice, and even ask it to be enthusiastic but not too crazy. Anything that matches your company’s tone,” he says.   

THE RISKS

AI feeds on data—some of which you might not want to share. Maintaining a secure network and protecting data has been a primary concern for Inntopia, which provides reservation and booking solutions for the travel industry, since well before AI came along. Blanchard says that to stay on top of the latest phishing and hacking trends and other risks, the entire staff receives security training every month. AI has made that an even bigger priority.

Data security. “We’re super vigilant about this stuff, but the big asterisk is that we let clients take their data and do other things with it, and that’s the stuff that worries me,” he says. “‘I think it’s important to have a policy that everyone’s aware of—and understands—[regarding] where data can and can’t go. There are different layers to AI, and once the data leaves your world, you have no idea how someone else might access and use it.” 

With ChatGPT there is an option to not share your data for AI training purposes, but that requires the user to put a lot of trust into what is a mostly unknown universe at this point. For that reason, Inntopia has established an internal committee and policies around AI.

Others also urge caution. “It’s kind of a slippery slope,” says Lannen. “I wouldn’t recommend putting straight-up financial documents inside ChatGPT or other services like it, as it stands right now.” 

Of course, nothing stands still in the world of technology. Lannen says solutions are being developed and coming online that are more specifically designed to handle secure data analysis.

“For anybody touching data—managers that have to analyze revenue, budgets, staff scheduling—there are and will continually grow to be better tools for these standard functions that can handle large volumes of data,” says Gibson. 

Unknown sources. In addition to protecting financial and guest data, Tragethon points to copyright infringement as another potential AI risk worth building a plan around. 

“You can never take [generated] responses at face value, because the mystery of AI—and AI engineers will tell you this—is that they don’t know specifically how it formulated the response, or where it got the information,” he says. “You always need to check it from a marketing standpoint. It (AI) could have stolen the copy from a competitor right next door. You don’t know where that content came from.”   

IMPLEMENTING AI

Right now, how well your company integrates AI largely depends on how well human intelligence is applied in the process. There are potential benefits for resorts of all sizes.

Leveling the playing field. Gibson sees an advantage for smaller independent ski areas. “Think about giving everybody on your staff an assistant who is a world-renowned expert in anything, whom you can ask any question and not ever be self-conscious about how nitpicky or repetitive the questions are,” he says. 

He recommends putting together an introduction to AI informational packet for managers and providing education that’s relevant to their department. 

Cultivate AI champions. Tragethon’s advice is to invite the key members of the team to a roundtable discussion and discuss tasks or areas where AI can help. “I’d want my people to get comfortable with AI. There should be some champions even at a small resort, someone that can develop an expertise and get really comfortable and familiar with the technology,” he says. 

“In the ski industry, a lot more of that is going to land in the administrative offices, because AI isn’t going to go out and shovel the snow for you,” he adds.

Talk to vendors. The key for ski area managers, says Ippolito, is to train your staff, but also start conversations with your existing vendors. 

“Look at those vendors that are already taking advantage of AI and the ones that are in front of it. That’s how you’re going to incorporate it into your operation,” he says. “More so than having to learn it and figure it out all by yourself, you’re still going to rely on those same types of vendors.” 

Know your goal. “It’s an exciting new technology, and it’s going to be helpful right away. But make sure you know the desired result you want before you start playing around. Ask the why,” says Blanchard. 

“And as for the next few years,” he adds, “strap in.” 

As the already popular saying goes, AI won’t replace you—but the (inspired, imaginative, intrepid) person using AI just might.

 

FIVE TIPS FOR IMPLEMENTING AN AI GAME PLAN

1. Develop a data security plan. Put security at the top of your AI list. If the person who oversees your data and network security doesn’t know what AI stands for, you may want to shop for a new vendor. 

2. Start the conversation with vendors. AI is so fast-moving, the challenge for ski area managers is simply keeping pace. But vendors have been living and breathing AI for the past couple of years. Let them be your AI Yoda.

3. Form an AI team. Include company managers and department heads, but do not leave out other thought leaders, including your AI champions—and be open to new ideas. 

4. Identify desired results/goals. AI presents endless possibilities, so it’s important to calculate the potential risks and rewards of the task you are trying to accomplish and ask yourself, “why?”

5. Start small and work your way up. Just like with the internet, social media, and texting, it’s better to dip your toe before committing to a deep plunge into new automation. It could save you the pain and embarrassment of an AI belly flop.