Exploring Machine Learning And Events

Ladies and gentlemen, it’s time to dive into the topic of machine learning and events. We are currently witnessing one of the biggest technological revolutions of all time. Right now, in real time, terms like AI and machine learning are being thrown everywhere. And while it may seem like we’re far from all the crazy, futuristic-like concepts we see in the movies, we’re not. Technologies like this will start making a huge impact in industries throughout the world. And of course, the event industry is among them.

But because we’re still taking the very first steps, it’s easy to get confused. What is, after all, machine learning? What is AI? How do they differ from each other? And most importantly, how can you leverage them successfully? If any of these questions have popped into your mind lately, today’s Even Tech Podcast is for you!

Joining our hosts Brandt Krueger and Will Curran is Chuck Elias. Chuck is the founder and CEO of Sciensio, the former business transformation, and growth executive for GE, Home Depot, super value, and fortune brands and the graduate of Harvard Business School. This week, they will take us on an exciting journey into the world of machine learning and events. Are you ready to become a part of the revolution? Press play!

Click here for the full audio transcription.

What’s AI And What’s Machine Learning

First, let’s understand what is what. Because the two terms are usually thrown together, it gets tricky to understand the differences between them. So, Chuck explains that “AI is basically nothing more than computers. The computers kind of finding information or taking large data sets and delivering insights, providing information automatically. So fundamentally it’s smart computers that are saying, “Wait, I asked you a question, and I got an answer back.”

AI has two main components. The first one is natural language processing – think Siri and Alexa. The second one “is really around, deep learning, machine learning, neural networks. You can call it a bunch of different things, but really what it’s about is going out and looking deep into data and trying to find insights and really actions that you could take that you wouldn’t get otherwise”.

How Much AI Is In This?

“If you look at some of the conference bots that are done through Alexa or … there’s a little bit of AI, but really what you’re doing is you’re just scripting a whole bunch of conversations and there isn’t a lot of depth behind them”, explains Chuck. “Even some of the early stages work that was done with Siri and those types of things, yes there’s AI, but it’s very kind of narrow. Very kind of limited”. So yes, AI is there, but not as much as we’re led to believe.

Natural Language Processing

Brandt advances that “you don’t have to be the person that knows the exact right search terms anymore in order to get what you want. You just ask the question and then it answers it and it’s either right or it’s wrong and hopefully more times than not it’s right”. Chuc agrees and takes it a step further: “the goal here is to have a natural conversation. It really is the way that you speak and that’s where you … If the bot doesn’t have a personality, then it’s not really doing much for you. Siri has a personality. Alexa has a little bit of a personality and you want to make it fun and interesting. So that’s really important”.

Chuck has picked up a couple of interesting things in regards to the learning process. First, “it’s really important that you understand is the bot delivering a high correct response rate”. And second, “it’s really important that you know a bot, it’s a bot, it’s not a human. Because you actually interact differently with a bot, than you will a human”. In reality, “people are much more accommodating of technology of bots than they are humans”.

Machine Learning Training

Unsupervised Training

“It goes back to many, many years ago when we’re just using algorithms and trying to do data mining in that way”, begins Chuck. “There’s two forms of training. They’re supervised and unsupervised. The unsupervised is what you hear in the press where, I think it was Google or one of the AI, they started talking themselves and created their own language (…) It’s going out and is trying to find connections and trying to basically solve for what is the answer? What could be the answer without any input at all, just large data sets”.

“It’s all around looking at large, large, massive data sets and trying to say, oh, it turns out that if you have this one condition and you also have this, then you might have this there. These connections that you would never think about in advance”.

Supervised Training

“Unsupervised is, I don’t know, I don’t even have a hypothesis, I’m just going to throw it in and see what comes out. That’s fundamentally the difference between these things that are really simple. So it doesn’t matter whether you like science or not. What we’re really doing when we do pure machine learning is we’re throwing stuff in and saying, okay, create all the variations that you can think of for this”.

What’s The Point?

If you’re wondering where this is going, and what it all comes down to, Chuck has the perfect response: “As you start to aggregate that data and that real feedback about what’s happening, then you start to say, what would I do differently? How do I message differently? How do I communicate differently? What kind of signs do I use? What things are people interested in? And that’s really the kind of where all of this is going, and the holy grail here is, it’s about a one to one conversation that’s going to create raving fans. That’s all we’re trying to do. The only reason we’re opening the channel and we’re using text is because everybody texts, and if I can open up that channel and actively listen and respond to your needs, you’re going to connect more with me and we’re going to build a business and then we’re going to keep getting deeper and deeper. That’s really what this is all about”.

Machine Learning And Events

Brandt ties it together with the event industry: “Well, and that’s goes back again to the learning, and the never forgetting aspect of it is that every time you’re doing your event, unless you have the exact same staff every time you’re able to lean on these technologies to keep that experience consistent that you’re able to not have to completely retrain a staff every time with, okay, the bathroom’s down there, here’s the most general session runs from here to here, and you’re not just handing them a piece of paper and asking them to memorize it”.

“The other thing that is impressive to me is the ability to compile it. As we were talking about before, if you start to see questions that the bot wasn’t trained for”, he continues. “At best you’re pulling that together at the end of the day or in a post-con, or maybe even not even immune until the pre-con the next year. The other thing that I really like about this type of technology is you’re able to start to put that together in real time. (…) Someone else asked the bot about that, and then we can formulate that answer and then we don’t have to answer that question again”.

What’s Next?

What excited Chris the most is learning to engage with people by leveraging machine learning. “How do you pull this data in and how do you begin to segment one to one? With a solution like this, with a tool like this where I can reach down and send you something that you will read. 98% of text are read, most are read in under five seconds, then you’re going to read it. Well, that’s a pretty high bar for marketers. Therefore you better send something they care about, or they’re going to shut you off. So that’s really to me what gets exciting. I start to change the engagement paradigm, one-to-one, I make it lower friction”.

Moving Past The Distrust

Many event planners and attendees are still suspicious of these technologies. There’s resistance from both sides, even though it presents so many exciting possibilities. Chris tells us a bit about how he tries to move past this: “If people don’t know how to access it, they’re not going to access it. So what we do, we tie into registration systems and once you’ve registered, we’ll send a text to you. (…) So I’m the bot that’s running, I’m here to help. That’s all it says. I’m here to help. If you want to connect, say hi. We’ll send a second message just before the event and if you didn’t connect before, we’ll say, “Hey, I’m still here to connect.”

“It’s all about the value. Does the person perceive that is valuable and that it can be trusted. (…) And so what we find is, you’ll see, generally we get about 70% of the people will give cell phone numbers. 60 to 70% of those will engage with the bot, at some time during the conversation.”

“What we are really always talking with planners about is, think beyond that just the question and answer. Think about how you’re going to interact. What do you trying to achieve, and you’re not trying to achieve”, he concludes.

‘What The Future Hold: Machine Learning And Events

“Events are always under pressure, to try to figure out how to do something more. And fundamentally it’s about taking the data that they already have and making it usable and making it an actionable”, says Chris. “Because the reality is, if you’re an event planner, your days are pretty much taken up with just trying to orchestrate the event and trying to get you to learn new technology and it’s just not going to work. So it’s not intuitive and simple. There’s not a whole lot of value in it for you. So, the goal is always to find these actionable tools that you can actually go and deploy specifically for your event”.

“I love what we do, and I love technology, but I’m not in this because the technology, so when I approach it, I go, “Is it solving a business problem and is it creatively solving it and is using technology to do that?” That’s the question. So, if it’s really working the way it should, it scales”.

Clay Christensen from Harvard also gave Chris some solid advice: “Look, fundamentally, we hire companies, we hire products, we hire tools to do specific jobs, that’s what we do. So if you’re trying to say, well, what will this technology do for my client or for my end user, for my attendee? Then you simply have to ask the question, well, what other tools could they hire to do that job?”.

Conclusion

And that does it for this week’s Event Tech Podcast! As always we’d love to hear your thoughts. Are you excited about the prospect of machine learning and events? Have you employed the technology in any way yet? How far do you think machine learning can take the event industry? Let us know in the comment section below!