Change in Technical Writer Role

Given the technological shift happening quickly after the introduction of ChatGPT-4, technical writers and knowledge base providers are still assessing how this will impact their job role and technology platforms. However, there is a shift in consumer consumption trends in knowledge creation, discovery and consumption as detailed in the below image

Technical writers need to shift their thinking in producing any knowledge base content which is AI bot friendly and makes it easy for both AI bot and humans to comprehend. This is a radical shift in how technical writers produce content as they have been producing content predominantly for humans until now! In future, technical writers will be producing new knowledge for AI bots, especially Generative AI technologies. However, the characteristics of writing for human consumption are extremely different to how to produce content for AI bots as explained in the below image

Writing for Humans

When a technical writer composes a knowledge base article, the content needs to be concise given the limitation of human attention span. The characteristics are dictated by biology, psychology, and cognitive capabilities. This translates to

• Most humans have shorter attention spans – This limits the content length to a certain number of words and constraints reading time
• Most humans prefer images and video for quick learning – This makes technical writers produce rich multimedia content such as screenshots, videos, and animated gifs
• Most humans prefer simple language – This constrains technical writers to use a small set of words and vocabulary to articulate complex ideas in simple words
• Humans are driven by emotions and ideology – The content needs to be inclusive and accessible to all humans. The content cannot exhibit any bias

Once the knowledge base article is consumed by a human, then they undertake certain activities as detailed in the knowledge base article. E.g if a software user reads about a product software feature in software documentation, then software user might configure that software.

Also Read:  Role of ChatGPT in Software Documentation

An intuitive software documentation tool to easily add your content and integrate it with any application. Give Document360 a try!

GET STARTED

Writing for an AI bot

Writing for an AI bot requires a mindset change on how technical writers create a new knowledge base article. An AI bot has unlimited computing power to synthesize new knowledge, unlimited storage to store huge volumes of information, and can interact with both humans and another AI bot!

Writing for an AI bot / Generative AI requires a different way of writing given the characteristics of the AI bot. This translates to

• Write content as explanatory as possible with a structure – Generative AI has the capability to summarize lengthy content into briefs and can explain complex ideas in simple terms
• Provide more examples on scenarios – Providing conversation-like scenarios with examples helps Generative AI technologies to consume this content easily
• Adding semantic metadata – Adding more metadata to annotate content enriches the content and augments the textual content
• Include FAQs – Writing FAQs helps with training Generative AI with set of prompts (questions) with responses (answers)
• Maintain a business glossary – A list of all business terms along with definitions, nuances, assumptions, metrics and so on
• Augmenting textual information with metadata – Adding metadata whether a particular action is an input to an interface or output of an interface helps Generative AI to know whether to seek intervention in case of automation of actions
• Related articles – A list of related articles from a technical writer’s point of view to the current articles

If technical writers produce AI-friendly content, then a customer can ask Generative AI on what objective they wish to accomplish. The Generative AI technologies such as ChatGPT, Bard and so on can provide precise steps. The bot technologies such as Robotic Process Automation (RPA), and intelligent automation via APIs will probably ask a customer whether it wants to complete those steps on their behalf!

Content Evaluation

Given that technical writers are going to produce content for AI bots, how we ensure that our current publicly available knowledge base contents are ready for Generative AI models to ingest and provide accurate answers based on our customers’ prompt. Large Language Models (LLMs) such as ChatGPT are trained on a large corpus of text taken from the open internet. If the LLMs need to use your knowledge base content, then those language models need to be fine-tuned. The fine-tuning of the LLMs requires giving a lot of prompts and responses as examples. This can be achieved through writing a lot of FAQs with questions and explanatory answers. Below is the list of steps for technical writers to repurpose their existing knowledge base to be AI-friendly

• Append more content to all your knowledge base articles with more explanations
• Add more metadata to both textual and multimedia elements
• Compile all business terms in a business glossary and use them consistently inside your knowledge base
• Produce a lot of FAQs with a lot of questions and laboratory answers
• Add “Related Articles” to all existing articles to establish a graphical connection between your content entities
• Review all your content to ensure that content is structured from H1 – H6

Metrics

Quantifying the efforts to make your knowledge base content helps with undertaking efforts to futureproof your existing knowledge base. Knowledge base platform vendors shall be building tools to help technical writers produce AI-friendly content. Below are some of the metrics that can help to get business stakeholders buy-in

• 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

Read more: How Technical Writers can Utilize ChatGPT?