AI Is Everywhere. But What Is It Actually Doing?
AI is everywhere in event technology. We hear about AI assistants, matchmaking, translation, search, summaries, recommendations, and an ever-growing list of other features that are now described as AI-powered.
Before we get into the different types of AI or decide where it belongs in an event, it helps to understand what the technology is actually doing.
Modern AI is different. It can learn patterns from examples or data and use those patterns to interpret, predict, recommend, or generate something when it encounters new information.
Traditional software might be told, if an attendee selects Exhibitor, show the exhibitor questions. A person created that rule, and the software follows it every time the same condition occurs.
AI can work differently. A model can be trained using large amounts of examples or information, learn patterns and relationships within them, and then apply what it learned to something new. Instead of relying only on a rule such as if this, then that, it can evaluate a new input and ask, in effect, based on what I learned, what is this most likely to mean?
The platform and the AI model are not always the same thing
When an event technology company adds an AI feature, it does not necessarily mean the company built its own AI model from the ground up. It may build its own model, connect to a model from another provider, or combine several technologies depending on the task.
What does that look like?
A microphone receives sound, not words. AI can use patterns learned from large amounts of speech to evaluate that sound and determine which words were most likely spoken. It is not simply matching the sound of a word to a word in a dictionary. It is using what it has learned about speech, language, and context to determine what was most likely said.
Search is changing too. Instead of choosing a few exact keywords, people can increasingly ask a question the way they would ask another person. AI can help interpret the words, context, and relationships in that question to determine what the person is most likely trying to find, then use that understanding to retrieve the relevant information.
“Where are we eating tonight?”
Awards Dinner | 7:00 PM | Grand Ballroom
There are almost no matching words, but AI can recognize that the attendee is probably asking about the dinner scheduled for that evening. It is not only searching for the word dinner. It is using what it has learned about language to interpret the attendee's meaning, then using that understanding to find the relevant event information.
The AI helped understand the question. The facts still came from the event information.
AI is not the event information
There can be two very different things happening when someone uses an AI-powered event tool.
Helps interpret what the person means, even when the wording does not match the event content.
Supplies event-specific facts such as the schedule, room, time, transportation, or other current information.
If an attendee asks, “What are we doing after the keynote?” AI may understand that the person wants the next scheduled activity. The platform can then look at the event information and find the reception.
But if the event data says the reception starts at 6:00 PM, that is the fact the system has to work with. If the reception was changed to 6:30 PM and the source information was never updated, AI cannot determine the new time simply by being smarter. It can make the connection between the attendee's question and the reception, but it cannot know an event-specific fact it was never given.
Imagine an attendee asks an event assistant: “I just finished the keynote. Where should I go next?”
AI might help interpret the question and understand that “where should I go next?” means the attendee wants to know what is scheduled after the keynote.
The event platform might then retrieve the current schedule and determine that networking lunch begins next in the Grand Ballroom.
AI might also turn those facts into a natural response: “The networking lunch begins at 12:30 PM in the Grand Ballroom. You are about a five-minute walk away.”
Traditional software, event data, search, programmed rules, and AI can all be working together inside the same feature.
That is the broad idea behind Powered by AI. Somewhere in the process, the technology is using a model that learned patterns from data and is applying what it learned to something new. The AI may be helping understand a question, recognize meaning, predict an outcome, or generate a response, while traditional software and event-specific data handle other parts of the same experience.
In the next Technology Bits, we'll look at why machine learning, generative AI, conversational AI, chatbots, AI search, and all the other terms seem to get mixed together.