By Seth Earley
Artificial intelligence (AI) and chatbots are the current technologies du jour. Just like big data that preceded them and the rage around predictive analytics, we are once again enjoying a period of heightened expectations and overhyped reality.
The greatest challenge in actually using any of these emerging technologies is still the same—understanding and clearly defining the business need, and having the right data and content to fuel the technology engine. Progress is being made, despite the atmosphere of hype. More prospective users are looking carefully at the purpose of the AI or chatbot, and considering how to support its operation. Even in the past several months, I’ve seen an evolution in understanding.
For some time, the AI vendors claimed that all you had to do was “train the AI,” but never defined exactly what that meant. “We need more training,” and “We need the right learning content,” were the often-repeated statements from developers that accompanied change order requests and additional funding requirements. Now,many companies are aware that chatbots are channels to content and data and realize that AI is actually a content delivery platform—that the magic is in how the content is structured and rationalized.
However, they still have no idea how to decide what to publish or how to structure it. Anything on a website that is customer-facing would be fair game, but what about a policy and procedure manual for entering employee payroll data? And what about legacy content?
Most companies have never done a content inventory, and there is likely no Director of Content Experience who has any insight into which silos contain what content. Where does a company even begin?
Your company may produce all kinds of content—technical product documentation, instructional content, and support information—but is now being faced with investigating machine-based content delivery, such as chatbots.
Are you intrigued and excited, while also feeling panicked and terrified? Are you afraid for your job and overwhelmed at being asked to do one more thing in addition to an already overloaded day job?
Relax, take a deep breath. Pause. Count to 10. OK, feel better? Good!
Now that that’s over, let’s take a clear look at the situation. The reality is that most organizations have checked the box when it comes to content. Regulatory requirements have forced them to create and organize content. The products they sell are now well documented, and various content functions, such as content marketing and technical publications, fulfill that requirement.
However, companies do not always scrutinize the value of the content beyond those requirements. I once worked with a Medicare administrative contractor (a “MAC,” which is an insurance company that processes Medicare claims) that employed 20 writers to produce enormous amounts of content. In a workshop to structure and tag content, I held up a piece of information and asked, “Who uses this? What’s the purpose? Why should someone read it?“
I was answered with blank stares from across the room. No one had asked those questions before. They couldn’t tell me why someone needed to read the content. They created because they were supposed to create content. That was their job. A lot of other content may not be a regulatory requirement, but it is needed for customer service and that content, in contrast, is not well organized or structured to facilitate its use.
Why do we create content in the first place (other than to comply with a regulation)? The answer should be, “To solve a problem.” Why do people have problems? Usually because something is not designed correctly. For example, customers call support reps because something is broken. We need the help desk because we can’t complete a task. We look things up because we need an answer. And what we get instead is a list of documents—tens, hundreds, thousands of results, some of which are hundreds of pages long—when what we really want is an answer.
Ideally, chatbots help provide these answers. Chatbots are channels to content and information to solve specific problems. They are not magic, and they require that we look at the needs of the user in a new light. We have to understand their journey in detail. We need to know what they’re trying to accomplish and what gets in their way.
Organizations are realizing this the hard way while they’re spending millions of dollars on AI technologies that are fated to fail, because the business has not thought through these fundamental questions. Moreover, they are hearing from vendors who say they can figure it out.
The vendors have the ear of executives and say, “Give us six months, $2 million dollars, and all of your content, and we will give you something that will solve your problems.” I’ve seen this approach fail miserably, over and over, because executives don’t realize that they already have, in their own technical writing organization, the talent to solve these problems.
One aspect of AI that is often overlooked is the role of the technical writer. Amidst the appeal of new technology, technical writers serve a vital role in development of AI applications, including chatbots, by being the subject experts most equipped to address the content requirements of chatbots. I’m giving a keynote at the Information Architecture Summit coming up in March in Chicago. In that talk I’m going to discuss the role of technical writing and the role of information architecture. These skill sets are going to be in greater demand today than they ever have been, and will continue to be as these technologies evolve.
Don’t get fooled by the hype and inflated expectations. At the end of the day, this is about using content to understand customers and meet their needs. Finding the right words to do this will often depend on the professional technical writer. These individuals, whose significance is often obscured by the latest technology, are important now and will be even more so in the future.
To get started, technical writers should begin by working with the marketing organization to:
- Understand and map out the customer lifecycle;
- Identify the points in the lifecycle where the customer needs to accomplish an objective;
- Deconstruct this objective into a finer-grained journey that illustrates the customer’s wants, needs, thoughts, emotional state, and tasks along with the content and information the customer requires to get to the next step to achieve their objective;
- Identify the sources of content and data and how they will be accessed;
- Determine the best channel to that content—whether human, a website, a mobile device, etc.;
- Evaluate the state of the content and restructure it to answer specific, common, high-value questions;
- Map out the possible questions that the customer may have at each step and structure dialog to prompt the correct question;
- Test on actual customers;
- Use human-bot hybrid learning to train the bot;
- Provide points of human escalation; and
- Measure, manage, govern, and improve.
Obviously, there’s much more detail than I can cover here, but this is the broad framework. From there, it’s a matter of committing the resources to complete every step and understanding how to implement each of the elements above by working with technical and customer experience teams. The role of the technical writer is increasingly vital to successful implementation of emerging AI and chatbot technologies.
For more information on the role of Information Architecture, see “There’s No AI without IA,” https://www.infoq.com/articles/artificial-intelligence and “The Problem with AI,” https://www.infoq.com/articles/problem-with-ai from IT Professional Magazine.
SETH EARLEY (seth@earley.com) is CEO of Earley Information Science (EIS). He also serves as editor and data analytics for IT Professional Magazine from the IEEE. Seth has a long history of industry education and research in emerging fields. His current work covers cognitive computing, knowledge engineering, data management systems, taxonomy, ontology, and metadata governance strategies. Seth has worked with a diverse roster of Fortune 1000 companies, helping them to achieve higher levels of operating performance by making information more findable, usable, and valuable through integrated enterprise architectures supporting analytics, e-commerce, and customer experience applications.
