It goes without saying that technical writers create knowledge base articles for internal teams and customers. The knowledge base article is available in a wide variety of flavors, including user manuals, technical spec documents, configuration guides, procedures, and software product features. These knowledge base articles are read by humans, who then act!
This paradigm shift is taking place in terms of how knowledge base articles are produced, used, and some actions are carried out. The following technological trends and consumer behavior are responsible for this change.
- Consumers who want things done quickly will favor automating low-value, routine tasks with bots.
- Customers want rough answers to the “right” questions so they can get things done more quickly.
- Rise of Generative Artificial Intelligence (AI) makes enterprise applications more accessible to non-technical users by streamlining knowledge discovery and content curation.
- an increase in people’s faith in AI systems
In this blog, we’ll examine how changes in consumer behavior will affect the creation of knowledge base articles and how people will use them.
Change in Technical Writer Role
Technical writers and knowledge base providers are still determining how the rapid technological change following the release of ChatGPT-4 will affect their respective job roles and technological platforms. However, as shown in the table below, there is a change in consumer consumption trends for knowledge creation, discovery, and consumption.
|Audience of your knowledge base||Humans||Artificial Intelligence bot|
|Content discovery engine||Search engines such as Google, Bing, and so on||Generative AI such as ChatGPT, Bard, and so on|
|Content discovery optimization||Search Engine Optimization (SEO)||Semantics content structuring|
|Content discovery mechanisms||Search keywords||Prompt|
|Outcomes of knowledge consumption||Humans doing work||Artificial Intelligence bot doing work with minimal human intervention|
Technical writers need to change the way they think when creating knowledge base content so that it is understandable to AI bots and humans alike. Technical writers have traditionally produced content primarily for humans, so this is a radical change in how they do it now. Technical writers will create new knowledge for AI bots in the future, particularly for Generative AI technologies. However, as shown in the table below, the characteristics of writing for human consumption and how to create content for AI bots are very different.
|Characteristics||Writing for humans||Writing for an AI bot|
|Content length||Concise and precise||As explanatory as possible|
|Use of visuals||Short videos and gifs||Media content needs rich annotation|
|Content tone||Can be conversational, monotone||Generic tone|
|Content output||Fixed||Highly customizable|
Writing for Humans
Given the short attention span of people, a technical writer must write knowledge base articles with concise content. The traits are determined by a person’s biology, psychology, and cognitive abilities. This is equivalent to
- The majority of people have shorter attention spans, which restricts the content’s word count and reading time.
- For quick learning, most people prefer visuals and video. Because of this, technical writers create rich multimedia content like screenshots, videos, and animated gifs.
- Since most people prefer simple language, technical writers are constrained to use a limited number of words and vocabulary to convey complex ideas in straightforward language.
- Humans are motivated by emotions and ideologies; therefore, the content must be inclusive and open to all people. The information cannot be biased.
Humans engage in specific actions as described in the knowledge base article once they have read and understood it. For instance, if a software user reads about a feature of a product in software documentation, they may configure that software.
Writing for an AI bot
Technical writers must modify their approach when creating new knowledge base articles for AI bots. An AI bot can interact with both humans and other AI bots and has limitless computing power to synthesize new knowledge and limitless storage to store massive amounts of data!
Given the traits of the AI bot, writing for an AI bot or Generative AI requires a different approach. This is equivalent to
- Create content that is as illustrative and structured as you can. Generative AI is capable of condensing lengthy content into briefs and can simplify complex concepts.
- More conversation-like scenarios with examples are needed to make the content easier to consume by generative AI technologies.
- Semantic metadata addition: Increasing the amount of metadata used to annotate content enhances and expands the textual content.
- Include FAQs – By creating FAQs with a series of prompts (questions) and responses (answers), you can train generative AI.
- Keep a glossary of business terms that includes definitions, nuances, assumptions, metrics, and other information.
- adding metadata to textual information In order for Generative AI to decide whether to seek human intervention when actions are automated, it is helpful to add metadata indicating whether a specific action is an input to or an output from an interface.
- Related content a list of articles that, from the perspective of a technical writer, are related to the current articles
If technical writers create content that is AI-friendly, then a customer can ask Generative AI what goal they want to achieve. Precise steps can be provided by generative AI technologies like ChatGPT, Bard, and others. Bot technologies like RPA and intelligent automation through APIs will probably ask a customer if they want to carry out those actions on their behalf.
How can we make sure that our current publicly available knowledge base contents are prepared for generative AI models to ingest and provide accurate answers based on our customers’ prompts, given that technical writers will produce content for AI bots? A sizable corpus of text taken from the public internet is used to train large language models (LLMs), such as ChatGPT. Your knowledge base content must be used by the LLMs, so those language models must be adjusted. Numerous examples of prompts and responses must be provided in order to fine-tune the LLMs. This can be accomplished by creating numerous FAQs with questions and detailed responses. The steps for technical writers to adapt their current knowledge base to be AI-friendly are listed below.
- Add more information and clarifications to all of your knowledge base articles.
- More metadata should be added to text and multimedia elements.
- Create a glossary of all business terms and use them consistently throughout your knowledge base.
- Create numerous FAQs with numerous questions and laboratory responses.
- To create a visual connection between your content entities, add “Related Articles” to all of your current articles.
- Check all of your content to make sure it is H1-H6 structured.
When making efforts to future-proof your current knowledge base, quantifying the efforts required to create the content for your knowledge base is helpful. Vendors of knowledge base platforms will develop tools to aid technical writers in creating content that is AI-friendly. The metrics listed below can aid in winning over business stakeholders.
- Number of words per article: 3000 – 5000 words
- Content structure compliance: H1 – H6
- FAQs per article: 10 – 20 FAQs
- Business glossary: 20 – 30 business terms