Ben: Artificial intelligence, yeah?
Liam: Yeah, that and machine learning probably. The coupling of those two as a discrete entity. I need to do the AI.
Marley: A lot of businesses seem to wrap up that they need AI and machine learning when it’s really just about a bad process issue.
Liam: The amount of times we have companies that have a ridiculous amount of data, we've got so much data we don't know roughly how to use it. And again, they look at some of the technologies that people are doing around machine learning models and sort of the idea of artificial intelligence... again because there are a lot of groups publishing it. When realistically what they really need to do is just start performing data transformation over the top of it and standard data science pieces on it.
Not every solution needs to be in the machine-learning model. A lot of the data points in terms of what they're actually looking to address and find in terms of the data and about their audience, or looking at some of their outcomes in terms of what they're not doing in terms of their product or in their marketing or in their pitch...you don't need to necessarily put that through copious amounts of machine learning or even the idea of creating an AI engine or compute layer to do that. They are probably the two words that get misused the most if we're honest.
Ben: It's not the future everyone thinks it is. Everyone thinks that you're going to ask AI "How do I..." I don't know, "talk to me in a natural way." And every chatbot that I've ever seen is, is utterly terrible. They're not very intuitive. And everyone's going, "Oh, we want that fantastic chatbot experience," but what they really should be aiming for in AI is something that's, you know, solving the customer problem. And yeah, I think it's just been oversold. It's really, really oversold.
Allan: Yeah, but it's rapidly changing. I mean, I think, I think we're rapidly moving towards a place where the Turing test is going to hold. And I think that is, we're talking about 44% of job replacement by, by AI in the next 20 years across Australia.
Liam: That’s more around robotics and machine learning…
Allan: Actually, no. It’s actually not.
Allan: That's not the physical, it's not the white collar, sorry, it's not the blue collar workers that are going to get hit by AI role-replacement first, it's going to be all the white collar. It’s going to be banking, finance in general, and legal.
Liam: What, accountants, really?
Allan: Yes, effectively. And software developers. Software development as the processes today is going to change rapidly. We deal a lot with the chat services that would apply at a call center or contact center. And I think you're right. It does need to solve the customer problem first, and I think that's where most companies are really behind is the detailed workflows, you know, consistent form that AI would need in order to be effective. And that's going to take some time just to map it out. To be fair, it's like training your replacement. People are going to resist that change. But yeah, in the background it's really that evolution as well of the humanization of those technologies. I can imagine a time, and it's inside our lifetimes, that you're going to pick up the phone and speak to someone and think you're speaking to somebody, and if it doesn't solve the problem it's moot. So yeah, I completely agree.
Ben: The classic example is I know a guy whose name is Paris, and the chatbot says, "What's your name?" He types in "Paris" and then it goes, "Oh, you want to go to France?" And he's like, "No, my name is Paris. You just asked me that." They can’t even do that right so I’m very skeptical.
Allan: It’s got a way to go.
Liam: I think that’s more an implementation issue there. In terms of the context of what it’s capturing or otherwise.
Ben: Ah, maybe. Maybe...
Liam: If you take Alexa and Google Home, I think there’s an element of their chatbot integration, and again some of their machine learning aspects in terms of NLP space. It has done reasonably well, not to say they’re like the be-all end-all.
Ben: Just saying, I’m yet to see any chatbot that impresses me. And I’ve tried them all.