Below are all of the questions that were submitted by attendees and answered by Adam. We hope you find them valuable!
As artificial intelligence transforms how tourism organizations engage visitors and streamline operations, destination marketers and tourism professionals face crucial questions about adopting these powerful tools safely and effectively. This comprehensive resource addresses the most common concerns about AI implementation in the tourism industry, from preventing chatbot hallucinations and protecting sensitive visitor data to maintaining authentic human connections while leveraging automation. Whether you’re exploring AI-powered trip planning assistants, seeking budget-friendly tools for small tourism offices, or wondering how to train your staff on emerging AI technologies, you’ll find practical, actionable answers that balance innovation with responsibility.
Learn how leading destination marketing organizations are using AI for personalized itinerary creation, multilingual visitor support, and enhanced accessibility while preserving the creative storytelling and local expertise that make your destination unique. These evidence-based strategies will help you confidently integrate AI tools like ChatGPT, Claude, and Microsoft Copilot into your marketing workflows, improve operational efficiency, and deliver exceptional visitor experiences without replacing the irreplaceable human touch that defines hospitality and tourism.
You can ensure that a chatbot (ChatGPT, Claude, Gemini, Meta and other available models) is accurate by educating yourself on how to flag when the chatbot is hallucinating. If the information it’s giving you seems a little too correct or organized, that’s worth questioning.
Working more with any chatbot gives you the experience to flag when that chatbot is going off the rails, making things up, or hallucinating. It’s very important to have stop gaps to fact-check these models, whether it’s a basic Google search without AI, using one or two other models to see if the information lines up, or best case scenario, having a human domain expert to bounce the information off of and see how accurate it is.
If you’re building your own chatbot, you’ll most likely be using some sort of Retrieval-Augmented Generation (RAG) or similar process to train your chatbot. In this case, it’s mission critical that all resources and data are accurate AND organized. Organization of the data is very important.
You wouldn’t necessarily need an AI agent per se to help you automate responses and calendar invites. “Agents” is very much a hype term right now.
What you would most likely be doing is building a pretty technical architecture that employs AI at specific points, using the connectors (technically called Application Programming Interface or APIs) from your calendar as well as connectors (APIs) to some chat or email interface where the people you want to communicate with are, whether it’s client-facing or internal. The goal is to create relevant and accurate responses based on information from your calendar and then deliver that to your stakeholders or clientele.
I would not focus on agents, but rather on building an efficient system/architecture that employs AI only when needed. More critically, build in touch points for human-in-the-loop (HITL) to check that the system is running properly and the AI is doing what it’s supposed to do and not hallucinating. Look at tools like Gumloop (more user-friendly) or n8n (more technical but can do more).
This issue most likely arises when an answer optimization engine (AEO) finds information that is often vague and formulates some sort of answer as an amalgamation of the different information it has found which is technically a hallucination. The potential root of this problem is based on a lack of schema and metadata that is formatted for AI scraping bots to read properly.
It’s paramount to start implementing schema and metadata and other formatting for all client-facing content that may be surfaced by search bots like ChatGPT, Perplexity or Google/Gemini. This process right now is labeled under a few different acronyms but AEO and GEO (Generative Engine Optimized) seem to be widely used.
So, if your content is formatted with proper schema and metadata, along with short FAQs around the content, then the chatbot that is surfacing the content to potential clients should be accurate. It’s not guaranteed, but that’s the best way to tackle this problem.
Do not chase shiny objects in the AI space, because you’ll never get good at using a tool. My recommendation would be to start with ChatGPT and use it as much as possible. Soon after you start getting used to implementing ChatGPT, start leveraging Gemini or Claude and copying and pasting the same prompts from ChatGPT into one of these two models. Slowly start building out your AI toolset.
These tools are improving at a rapid pace, so any tool that says it specializes in XYZ business task, there’s a good chance that one of the frontier models like ChatGPT, Claude, or Gemini will be building out features that do the exact same thing. I have seen this play out over the past three years where startups come in and say that they can analyze a PDF back when ChatGPT couldn’t do that, and two months later ChatGPT rolled out that feature and many, many startups were wiped out.
The leaders of the frontier models have openly said: do not build a business around specific features because they will continue to roll these features out and wipe out any startups that are doing this. So again, focus on getting good at using one of the three frontier models, and as you get better at it and more comfortable, you will start to be able to use these tools for more and more things and lean on the features built into these frontier model suites.
To harness AI and utilize its potential, it all needs to start with leadership from the top, whether that’s the CEO or the head of the department, championing the use of AI properly and ethically and educating the employees on how to use AI ethically and properly. Simultaneously, you need to be building out Standard Operating Procedures or AI SOPs that are a go-to for all employees to look at when they’re not sure if, when, or how to use the AI tool that the company has approved.
This is the first part of onboarding AI to any business that wants to make it a success.
With this education coming from the top down, you can instruct employees on what data they can and CANNOT enter into a prompt box or upload to a model. Critical thinking is a big part of this, but having an AI-SOP that states this specifically is best.
The technical side of this question is a bit more difficult. That has to do with discussing what frontier models to use, whether to use an off-the-shelf model, or whether to build your own local or native model on top of an open-source model that you have total control of training and safeguards to keep sensitive data inside your servers. This is a much more nuanced and technical question that has way more variables.
There are some very large companies using frontier models across their organization and it is assumed that there are safeguards in place so that sensitive data remains private and secure. I would do some research on this.
But if you’re asking this question, I would assume that you still need to focus on the human aspect of this and make sure that the company from top to bottom is aligned on the use of AI tools and AI roll out.
This is a good one. This can be easily addressed by letting your clients or stakeholders know that the content was created with AI.
If this question is asking about potential copyright collateral being integrated into content that was created with AI, that has to do with how the user creating this collateral prompted the model. So that goes down to educating the end user on how to ethically use the model.
I do not use specific artists, writers, or individuals in the prompt to get a similar style as an output, that is wrong and unethical. I have not seen a case where it is okay to use copyrighted material in a prompt where the output looks similar to the copyrighted material. That is unethical and should not be done.
This comes down to education and experience. Start by using the tool and experimenting with it. Think about everyday tasks where AI might be able to help you, and then try implementing AI to assist with those tasks.
Really, it’s about getting familiar with the tools, understanding their capabilities and limits, and then utilizing them wherever you see fit. If you’re already doing good work, there’s every reason to believe you can implement these tools to augment what you do and become even more valuable. This is actually one of the best ways to stay ahead and reduce concerns about AI impacting your role, by learning to utilize it rather than be afraid of it.
The easiest transformation I could see in tourism is custom-trained chatbots being the focal point of the website. Someone could ask specific questions in any language and get timely, very accurate responses in a cordial, personal way. This would be better than the current messaging interfaces seen on a lot of tourism business websites, you know, the button down in the bottom right that are pretty janky and unreliable.
I envision what I’m talking about being the main interface of the website, where pictures can be featured in the chat, videos can be shown to the user, and the dialogue is very friendly and playful as well as accurate. This chatbot would have a personality that represents the branding and messaging of the tourist business, whether it’s a tour company, a hotel, or a restaurant. Basically, like a very high-end digital concierge.
That’s just one idea. There are a couple other ones that will probably be implemented in the next three years that lean on AI.
Enter the below prompt into ChatGPT:
What are practical examples of AI applications that enhance visitor engagement and support the tourism industry? Include specific use cases for chatbots, personalization, virtual tours, operational improvements and other aspects of the tourism industry.
Give me the use case and the problem it solves. Give me the exact tools needed to build the solution and a rating systems for easy, medium or hard for the build.
Provide the output in a subheading and bullet point layout.
Avoid being verbose and giving me extra information and ideas. Stay focused and only give me what I have requested!
The best and safest way is not to gather AI-generated content but to create it yourself. Again, I’m going to lean back on educating yourself on best practices and the most ethical ways to utilize AI. Make sure you’re not using copyrighted content in your prompts, but you are using your brand messaging, and aligning that with the goals of the content. You can also use your brand colors and reference graphics if you are creating media with AI.
Focus on becoming very good at leaning on specific AI tools to create content that is in line with your campaign and messaging.
To help train and educate your staff to integrate AI tools, you need an AI leader who is the go-to person within your organization. That person is not only leading the charge of implementing AI but also developing AI SOPs, AI standard operating procedures, that can serve as a guide for all of your employees who are using AI.
This will help make sure that your employees are using AI ethically and properly, as well as maintaining data security to ensure that sensitive information is not uploaded by accident. Making this a top-down initiative is the best scenario.
1) Suitability screen
2) Data & compliance readiness
3) Impact & economics
4) Variability, risk, and controls
5) Integration & operations
6) Measurement & pilot
Prioritize
Generally avoid automating
Human intuition, thinking critically and creativity are extremely valuable assets when comparing our work to what AI can do. Again, I’m going to go back to education and learning what these tools are capable of and what their limits are. That’s the best way to understand where they fit in the human workflow.
Augmenting your work with these tools and bringing more value to what you do is the way to make sure they don’t replace humans or intuition, which would be a big mistake. These tools are also capable of enhancing creativity, not replacing them. Period. Full Stop.
The below are just ideas and I have not personally used AI for any of these uses cases though I do believe that the tech does exist and AI can most likely aid in all of the recommendations below.
Real-time translation, captions & voice: translate signs/menus/conversations; speech-to-text & text-to-speech; multilingual scripts.
Personalized accessibility planning & step-free routing: recommend accessible routes/hotels/attractions; elevators/transit; tailored itineraries.
Visual & alternative-format assistance: computer vision describes scenes/art; obstacle detection; alt text, audio descriptions, easy-read guides.
Conversational assistants (chat/voice/SMS/WhatsApp): 24/7 multi-language support, screen-reader friendly, bookings & info.
On-site navigation aids (AR & landmark recognition): wayfinding cues for low-vision; safe route guidance.
Staff copilot: live translation and inclusive-language tips for front-of-house teams.
Affordability & access: automation lowers costs, widening access to planning/support tools.
Yes, the frontier models like ChatGPT, Gemini, and Claude all have a free tier before you have to pay anything. I would use all of those free tiers as much as possible, push them as hard as you can until you start needing access to more prompting and better models. You’ll naturally get better at prompting and using the models, as well as knowing when you need to upgrade to a paid plan.
The best tools would be to start on the free tier of ChatGPT, Gemini, and Claude and to use those to their full potential on the free plans. If you’re going to roll this out with multiple staff members, you should come up with an AI SOP (Standard Operating Procedure) so that your staff members know how to use the AI tools and how not to use the AI tools.
Most importantly, this comes down to the employees of your organization using AI tools properly and ethically and knowing what they can and cannot put into a prompt and upload to an AI model.
I think the largest risk would be based on how sensitive the data is of your visitors. If there are no names, addresses, phone numbers, or general identifying data points, the risk is lower. If you’re concerned about security around using models, then you should consider either working directly as a partner with a frontier model company and developing guardrails for whatever model you’re using, or develop your own solution on an open-source model within the confines of your company’s servers to make sure that no sensitive data goes outside of your servers.
The easiest way to ensure transparency when using AI-generated materials is to let your readers know that content was created with AI. This can include a footer below all images that were created with AI, as well as content that was created with the help of AI. You can also put a disclosure on your website stating that you and your company use AI tools when creating content.
AI can be used in many different ways to streamline administrative work. The key is to look at tasks that are repeatable and have set parameters going in and going out, and don’t have a lot of nuances or decision-making involved.
That can include general organization of documents based on client documents that were filled out and having that information updated in a spreadsheet. It can also mean calendar information being updated based on bookings. It can be replies or communications that are standard when dealing with clients, so creating a chatbot that is trained on a lot of previous interactions with clients and being able to either directly connect the chatbot to the client-facing messaging service or email, or manually copy and pasting questions into a chatbot that has been trained on the best ways to respond to those questions.
I’m sure there are other tasks and use cases that can apply to administrative work for a small tourism organization.
To maintain the human touch in marketing while using AI tools, it is critical to educate yourself on best practices and ethical ways to use AI. The key is to go in knowing that you are using AI to augment your work, not replace it, and that AI tools will make mistakes. It’s your job to catch those mistakes and to always stand by work that is created with AI.
If you’re not proud of the work that was created, then you shouldn’t publish it or distribute it. This goes hand in hand with always assuming that the AI is just making a draft and that you, the human, are always going to put your own creativity and personality into the content, whether it’s written, visual, or audio. The human touch is what makes it special.
I did some high level research on this topic as I am not an expert in this field. I hope some of these examples can give you a starting point on what is possible for your own organization.
– Destination management organizations (DMOs) and tourism boards using AI-powered chatbots as virtual visitor staff to provide 24/7 customer support and travel information. This enhances visitor interaction and convenience. Link
– AI trip planners by companies like Booking.com and Tripadvisor that create personalized travel itineraries and accommodation recommendations using natural language processing and large data analysis. Link
– AI-driven tools for sentiment analysis, customer segmentation, and automated service recommendations are widely applied to optimize operations and marketing for tourism businesses. Link
– Here is a list of some of the most well-known travel brands I and how they’re integrating AI into their businesses. Link
Honestly, it’s just getting started with the tools and understanding how they operate and how to prompt them to get the output you need based on the task you’re working on. This is holistic and applies to all industries. Having a high level understanding of how to utilize different AI tools and pick up new AI tools without having to spend a lot of time learning how they work will be the most valuable skill.
This means prompting in natural language, uploading documents and prompting to analyze those documents, pulling out different data points, and comparing and contrasting data points. This could also include creating multimedia content utilizing different tools, as well as, from a decision-making standpoint, being able to brainstorm and ideate with AI tools to get as much insight as possible into making important decisions.
I think it all comes down to making sure you stay relevant and competitive.
For someone completely new to AI, they should start with ChatGPT, the free model, and start entering basic questions that you would Google into ChatGPT. So, using ChatGPT as your new search and research tool is the easiest way to start.
An easy prompting technique is to describe your business or what you do with your business and ask ChatGPT to give you seven ways that ChatGPT can help you make your business or tasks more efficient and save you time. (Copy and paste the prompt below to get started. Make this process your own and iterate on this prompt to get better insights!)
See where that takes you and just start ideating. As you think of problems around your business that you need to solve, or you think of things that AI might be able to help you with, type it into the prompt box in ChatGPT and see what comes out. Start having a back-and-forth conversation and make it very natural and organic.
PROMPT: “I run or work for a [describe your business/role]. Can you give me seven ways that ChatGPT can help me make my business or role more efficient and save me time?”
EXAMPLE “I run a small tour company in Boston that specializes in historical tourism. Can you give me seven ways that ChatGPT can help me make my business more efficient and save me time?”
Soon within this process, you’ll want to have a cloud document platform like Google Drive or Microsoft SharePoint where you can save valuable outputs that ChatGPT gives you so that you can go back to them later on and build on top of those outputs, similar to the way I worked through the tier two example in the webinar.
To vet AI-generated information for public-facing materials, always treat AI outputs as a first draft that requires verification. Fact-check any claims using reliable sources through a basic Google search, cross-reference the information with one or two other AI models to see if the outputs align, and ideally have a human domain expert review the content for accuracy.
Never publish AI-generated content without reading through it carefully yourself, checking for hallucinations or information that seems too polished or organized, and ensuring it aligns with your brand voice and messaging. If you’re not confident standing behind the content or proud of the final output, don’t publish it, the human review and approval process is essential to maintaining credibility.
