How AI Will Change Training, for Better or for Worse
A forward-thinking conversation about leveraging AI in hospitality training
Moderator: Rachael Nemeth
CEO, Opus Training
Director of Training, Bonchon Korean Fried Chicken
CTO, Opus Training
Artificial Intelligence (AI) has been making headlines for quite some time now, and yet it remains somewhat of a mystery to many. While it is not a new technology, it has recently reached a turning point where it has become accessible and helpful to business professionals. This newfound usability has prompted many hospitality trainers to consider how AI can be utilized to streamline their content creation and optimize their operations.
Rachael: Why now? Why does this feel like this is the first time AI is hitting professionals like us?
Ashley: This is the first time it has felt like AI is useful. Though tools have been out there, it would have taken a lot of time to learn them, and a lot of us didn’t have the time. It’s also scary! Tech fear is real. But now that it is widely accessible, we can use it and see results immediately.
Jeff: Each year, there are so many advancements in technology. We see it with our phones and laptops, and we have gotten to a point where computers are incredibly high-performance. That has unlocked new possibilities that we could only dream of before because our computers are so fast and so powerful.
At the forefront of AI is a company called OpenAI. The news headlines were spurred by their release of technology called GPT-3, which is an AI model that has been made available to the public. GPT stands for generative, pretrained, and transformer. First, it generates; it creates things. But the important part is in the middle – pretrained.
It was trained on the largest data set that an AI model has ever been trained on: 175 billion parameters. What that pretraining does is creates this thing that understands the world through all of the text that was on the internet before 2019. It takes all of that data and generates relationships and connections, and goes back to the first part of generating to answer questions and create content.
The T in GPT3 stands for transformer, which is the architecture that it runs on. Three means it is the third version. And now, with GPT-4’s release, it is 10 times more advanced than its predecessor.
Rachael: What is one word you would use to describe AI in training?
Rachael: Ashley, you sent me a text a couple of months ago about AI. Do you remember why you were so excited about the use cases for AI at that time?
Ashley: Yes! I have long believed restaurants are equal parts art and science. And the science, we mostly have nailed, but the art part is the hardest to train. With AI, it feels like for the first time there is technology that helps us create an engaging experience while we execute. And that AI could also help trainers deal with topics that change quickly, like de-escalation training, which is inherently hard to do from a distance. It can now create custom content that is specific to our regions and operational issues.
Rachael: What are some other practical examples? What are a few ways we can deliberately and thoughtfully use AI for training?
Ashley: A focus for us at Bonchon is to bridge literacy and language barriers. In branching or role play scenarios, that is really difficult for people with language barriers. To open up that channel is amazing for us. We can data mine keywords to see what they are actually taking away so we can see where we are missing the mark. We can pull keywords to track employee and business trends, so we can make better training.
Jeff: AI can help us build accessible content. AI can look at your course content and tell you what reading level you are at to make sure the course is targeted to your audience. I’m also excited about image matching. Let’s say you need to learn how to make a Bonchon dish. If you had an accurate image of it, AI can automatically write the training for that, and then validate that the employee did it correctly.
Ashley: I love that. With so much off-premises focus, we have training on how to check a bag to see if the order is correct. If we have an image of what’s correct, there is software now that could validate it. We spend so much money refunding missing items, if we could frontload it and validate it, which would be a game-changer for us.
Rachael: One thing I have been thinking about is AI helping trainers build frameworks. Let’s say AI generates a basic outline and then I can fill in the expert content. If my prompt is “Write a seven-day course for servers who have not been in the industry,” AI could generate the basic outline, and then I can then fill in the subject matter expertise. This could save us time, but does not replace the gut work.
Rachael: We keep hearing the term adaptive learning. How can AI help tailor the learning experience to each individual?
Ashley: Currently, it’s hard to figure out where a trainee needs to start. If we are in person, we can ask them questions about what they know. But if we are not face-to-face, AI can help speed through items they are proficient at, and slow down on those they are not. It helps us make the best use of their time, which translates into dollars saved. We can also get immediate, real-time feedback from them (that is not a basic 1-10 scale survey) about what’s working and what isn’t, and we can make adjustments quickly.
Rachael: I have also heard of trainers using Murf AI on their web-based training to add narrations, and examples of comparing and contrasting training they already wrote with AI’s outline versus their own, to see what they are missing.
Ashley: As trainers, when we make one change to one area, it can now be carried through to the rest of the content, without sorting through it all yourself. I put all of our programs into AI and asked it if it saw anything major I was missing. Having that speed to sort through information saved me hours of time.
Rachael: You’ve got me thinking. Let’s say you have to re-up training once per year, which could be compliance or even general company information. You could now send it out in a different and fresh way. Perhaps if we provided a new version each year – the same content, but delivered differently – that could positively impact learning outcomes.
There were so many times that I thought “Technology could do this.” -Rachael Nemeth
Rachael: How do we leverage our front-line managers? How can we leverage the subject matter experts who are out there delivering the knowledge verbally? Your wine directors, sous chefs, shift leaders. Can AI can help those people become creators as well? Can the library of content we would have had to purchase now be developed internally?
Ashley: We tell managers you can’t manage from an office, that it is about people engagement, but then we bog them down with data work: to forecast for scheduling, ordering, equipment fire-up times. The data is there, and if we could help automate that, these front-line managers can get back to what they are good at and what they love to do. So, yes.
Rachael: It makes me think back to when I was a people operations manager. My only job was to eliminate administrative work for our managers. There were so many times that I thought “Technology could do this.”
Rachael: Jeff, how can AI-assist content help hospitality trainers?
Jeff: We are investing in a course builder and template catalog to develop quality courses quickly. It is hard for an off-the-shelf template to capture the art that Ashley previously mentioned: how you as a company and you as a leader talk about your specific brand and business issues. What we have found is that our old templates were good for structure, but they did not match a company’s culture and did not feel like they belonged to the company.
Now, we use AI to understand source material – voice, tone, and intention – to generate better training content. When we upload a file with the course information, with those source inputs, we can create content that feels just like the company wrote it. You proofread and make sure it flows correctly, and then it’s good to go. You can also test content that you have written to make sure it is cohesive.
Rachael: Is there a dark side to AI?
Ashley: There are always downsides. There is no user manual, and we are all learning together. What I would say to skeptics is that it is not all or nothing. You can test it and see what works for you. If you have the time to do things the long way, great, but most of us don’t.
AI is a tool. We have to know what to tell it. For instance, for us in franchising, we have joint-employer liability issues, so we have to make sure that we tell AI to actively check for new rules. It doesn’t just run itself.
Jeff: This is not a driverless car. It is a turboboost tool that can help you be more efficient and faster, but you need to be the operator. Sometimes AI gets it wrong: incorrect facts, issues with tone that don’t sound like part of your brand, etc. You need to give content a proofread to make sure it is of high quality.
Rachael: Will AI lead to job loss?
Ashley: For some people, it may. Trainers who adapt to it, utilize it, embrace it, and work with it will be valuable. As things get faster, we will be asked to do more and more. I have been a one-person training department, and what I am asked to do is enormous. This helps me do it quicker and adds to the value I bring to the table at this point.
Jeff: We will have AI in our day-to-day regardless of what job we are doing. People with proficiency on their resume will stand out. You can use AI to set yourself apart in recruiting and screening.
Rachel: This reminds me that back in the day, you could highlight on your resume that you could type 90 words per minute. Then it grew to listing your word processing technology experience. Then it grew to Google Suite, and now LMS proficiency. AI is just another layer. If you become an expert in leveraging AI for training purposes, you could go further and be a higher earner in your career.
Rachael: We’ve talked a lot about how AI affects individuals, but how does it affect the business?
Jeff: AI can recommend training based on data points. Say you have a new menu item, and it is a total hit everywhere, but there are three locations that are not performing. AI can automatically look at those data points, and say re-training needs to happen. We can send that training out and then see the sales results. We can automate the cycle.
Ashley: Turnover happens. We could hate it, we could try to combat it (which we do), or we could accept it. We look at our current team members as future guests, and we want our training to be engaging and cutting-edge and congruent with what our guests experience. AI can help.
Rachel: What about big tech stealing our data?
Jeff: When you use these models, and you are agreeing they can take your data and give you an answer. Now that data can help answer someone else’s question. So that data now contributes to the parameters. It is an important conversation to have.
Rachel: What are free AI tools trainers use now?
Ashley: I use ChatGPT and DALL·E 2 to get image suggestions.
Rachel: There is also software to layer on top of these tools. I use Writesonic to give my blogs a kickstart by giving me an outline of how I could write something.
This conversation concluded that AI is transforming our world as trainers. We can work together to become experts together. We can use this incredible technology to co-create a better world for our employees, for our businesses, and for our own careers.