[Ryan Weber] Welcome to 10-Minute Tech Comm. This is Ryan Weber at the University of Alabama in Huntsville. I’ve seen a lot of articles recently about how disruptive AI will be. How it’s going to take all of our jobs and totally change the future. I wanted to get a reliable prediction about how artificial intelligence will change the world of work, so I brought in an expert. Seth Earley is the Founder and CEO of Earley Information Science. He’s got twenty plus years’ experience with knowledge strategy, data and information architecture, search based applications, and information findability solutions. He’s also written about technical communication and artificial intelligence in the past. So, I thought he might be able to bring us some insight about how Al’s going to change technical writing and how technical communicators can prepare themselves for a future working with robots.
[Robot Voice] Stay tuned for the conversation with Seth Earley.
[Weber] Welcome to the podcast Seth. Thanks so much for joining us today to talk about AI in technical communication and I know AI is a big topic sort of just across the board lately. Can you tell us how AI is currently being used in technical communication? What impact you think it’s having on the field?
[Seth Earley] There’s a lot of potential and in terms of the actual execution and impact, I don’t think it’s as great as it will be. I think we’re still at the very early stages but we’re starting to see organizations leverage technical content in chat box support and call deflection, but the problem is that the technology and maturity of the industry and the technology is not very far along. So it’s not very mature and so it’s not going to displace a lot of technical support organizations, but it’s going to augment those and there’s a lot of different ways to do that. But we’re starting to see an increasing realization and awareness that technical content is very important to chat box and AI initiatives and cognitive initiatives, cognitive computing initiatives. And so we’re starting to see people take kind of take notice and unfortunately because more organizations don’t understand the nuances and the mechanics behind how these things work, they’re setting up entirely new organizations to build AI content. I’ve seen that more in more than one organization or they’re handing things over to global integrators who are telling them that they need to do things very differently and not necessarily rely on the skills that they have in house.
[Weber] So what you’re saying is that a lot of organizations are sort of outsourcing all of the content that’s written for automation and they really don’t need to, is that right?
[Earley] I’m seeing that to some degree. I’m seeing the phenomenon of the whole idea of content and knowledge being confiscated by this term of art that’s used by the-the industry and it’s called training, training the AI. “Why do you need more budget? We have to train it. Why isn’t it ready yet? We haven’t finished training yet.” But then you have to say, “Well what do you train AI with?” You train it with content, data, and knowledge assets, and in many cases that-those have to be highly curated assets. There’s this misconception that AI will solve your problems by you know looking at messy content and you know large amounts of unstructured content and make sense of it and learn from it and figure it out and nothing can be further from the truth. There are some large-scale data approaches to handling things like support calls and building content and that’s again to machine translation. So machine translation uses large amounts of data by comparing languages, by comparing things that are written in English, and things that are written in French, or Spanish, or whatever the other languages are, and then it comes up with the rules and the correlations to do the translation. And the same thing can happen if you have large amounts of human chat content and you have large numbers of chat log, large chat logs, and large numbers of instances and sessions and you’re essentially taking questions and you’re translating those questions into answers.Just like you would translate English into French or Spanish or whatever other language you’re talking about. Those are called training sets and so training data can be considered from different perspectives. But when we’ re building a chat bot to handle very specific support types of conversations or transaction conversations or advisory conversations that content cannot be derived from large numbers of harder chat sessions. It’s just not a feasible way when you need to give your customer exactly the answer that they need. So there’s a lot of implications around what’s happening in this space but a key factor here is that this process and this capability needs to be owned by the organization and the organization really does need to look at the internal communications and technical documentation organization for content.
[Weber] So is there opportunity for technical communicators to kind of step in and argue for their worth in this process or should they be hands-off, or should they be proactive into arguing for their worth here?
[Earley] I think the later. I think you-we really do need to rebrand and reposition yourselves in terms of technical communication. The organization doesn’t necessarily-they look at technical communications, and tech docs, and tech pubs, as a necessary evil, where “Okay this is documentation that has to go into the product that nobody reads,” that’s one type of documentation. There’s another type of documentation that says, “Well this is really important and critical content that is needed to support this product. However, the way that content is written and the way it’s chunked and the way it’s managed has to be thought through a little bit differently.” You know just like we’re chunking content for translation, you know we have to chunk content for retrieval and so we need to think, “Well what is-what exactly is the user going to need at a particular point in time,” so that means we need to build personas and user journeys and we need large numbers of use cases and then, “How can I serve that content up in that context,” and think of a chat bot as a channel. That’s what a chat bot is, it’s a channel. People think of it as its own thing or that it is AI, it’s a channel. It’s a channel to content data and knowledge. There’s lots of different technologies that you can layer in. There’s lots of different ways of solving the problems, but in and of itself, you still need the content, the data, and you still need the knowledge assets and those have to be curated and structured in such a way that they can be retrieved as the user requires them.
[Weber] Right now you’re saying that this technology is kind of still in its infancy, a lot of companies may not be doing this as well as they could, but let’s look say ten years, fifteen years down the road, where do you think that AI is in terms of technical communication at that point?
[Earley] We are going to be in a different world. We are at an inflection point in human history and we’re going to look back at this and go, “Remember when we didn’t have bots? Remember when you had to actually type in your questions? Remember when you couldn’t talk to your computer? Or remember how dumb Siri was or Alexa?” Can’t say Alexa too loud, she’s listening in the other room, but the point here is that it’s-this stuff is going to get better and it’s going to get better really, really quickly and you know the large organizations: the Google’s and the Facebook’s, and the Microsoft’s, and the Amazon’s of the world are investing in the heavy lifting of the algorithms and the mathematics to do certain things and it’s going to be up to organizations to build and curate their knowledge depositories and their assets. And organizations are going to compete based on the value of their algorithms and the way they apply them and their content and the way they understand their customers. So, when you think about doing business with an organization, think of the best experience that you ever had with say a sales organization, and somebody who really knew you and solved your problem, and anticipated what you needed, were honest and they were out to help, and of course they benefited because they got your business. But they knew you, they knew their products, they knew their offerings, they knew their services, and they knew how to satisfy your needs. That’s what we’re building in bots. That’s what we’re building in website experiences and again because we’re talking about a channel and a channel to content, that content has to exist. That knowledge has to exist. That understanding of the customer has to exist. We need to have a very deep understanding of the needs of our customers. We have to understand the in great detail and depth because we need to solve their problems with our tools, and our products, and our services, and our solutions, and then we need to know when to service that to them.
[Weber] Yeah, if I’m understanding right, because you know there’s a lot of anxiety about AI, that it’s going to take jobs and you know do everything that people can do, but if I’m understanding your prediction correctly, both now and in the future AI works best when people are giving it very carefully curated content and a lot of very carefully researched data about the users who need that content. And so that’s kind of what the humans will be doing on the other end, the technical communicators, is that right?
[Earley] Yes, there’s no doubt that it’s going to impact labor and employment and jobs, we’re going to remove the mundane, the repetitive, the boring stuff. You know we’re going to get rid of a lot of support calls that are just rote, and the humans are going to be engaging with other humans. They’re going to be creative. They’re going to be continuing to build knowledge of the organization and we’ll be able to do personalized interaction at a scale that we can’t do now. And that’s what it’s going to do. It’s going to allow for a much greater depth and richness of communication and interaction with whatever organization you need to do business with and hopefully it won’t just be Amazon, Google, or you know App-Amazon Google.
[Weber] One giant conglomerate, yeah.
[Earley] Hopefully we won’t get to that point, but we will get greater fluency with these things, they will become more capable. They will disrupt jobs just like there’s jobs that were disrupted by every technical revolution, every change in technology. There used to be a person whose job it was to read to factory workers because they would be bored. They were put out of work by radios. There are no more bowling pin setters anymore. There’s lots of things that have changed over time, but we couldn’t imagine the jobs that we have now back then. And you know Queen Elizabeth in you know the fifteenth century refused to grant a patent for a knitting machine because it was going to put too many knitters out of work.
[Weber] So you’re saying we’ve seen this before in history we’re going to see it again and we don’t really know exactly what’s going to happen in the future, but we know things are going to change.
[Earley] Where it’s going to go it’s hard to tell but it’s going to be a very different world in ten or fifteen years.
[Weber] Well that’s interesting. So, if you are technical communicator now, what can you do to prepare for this very different world?
[Earley] Well I think you need to understand content and you’ll need to look at chat bot frameworks and you need to understand how natural language-processing an actual language, understanding work, and honing your skills in information architecture, knowledge engineering, any of those areas that you can learn and build stuff. You know see what you can understand form that perspective and demonstrate the capability.
[Weber] Very cool. Well hey thank you so much for giving us your insight on this. It’s really interesting and hey we’ll check back in fifteen years and we’ll see how the predictions panned out.
[Earley] (chuckle) Very nice.
[Robot Voice] Thanks for listening to 10-Minute Tech Comm..