Discussion Board

“With AI infiltrating even online written tests and documentation in some cases, have you used it for professional reasons? What’s the feedback you got? What are AI key guidelines that we should know when we use AI for documentation?”

Artificial intelligence, or AI, is now an inseparable part of human life. This technology has a wide range of applications, from finance to facial recognition. With increasing usage, experts think there is a need for an emerging AI governance framework. This will reshape how the technology is built and deployed. Keeping this agenda in mind, India’s telecom regulator, Trai, has recommended the establishment of an independent statutory authority, the Artificial Intelligence and Data Authority of India (AIDAI), to regulate responsible AI use across sectors. Source mint.

Technical Writers have been very concerned about AI and its usage from the very beginning. Almost a month ago, they discussed: Whether there are any AI guidelines for creating documentation?

The Discussion

Thank you for sharing an awesome discussion topic! Here’s what I think about scenario 1: I agree that anyone can become a technical writer. However, I don't buy the “without additional learning” part. Ours is a faculty of knowledge that has a steep learning curve. To even survive, we must constantly better ourselves. A good technical writer can come from anywhere. But a technical writer that doesn’t learn doesn’t reach anywhere. Here’s what I think about scenario 2: Experienced technical writers are rejected for numerous reasons. But if recruiters reject them because of the lack of basic technical writing skills, I think it doesn’t deserve further discussion. This is where both scenarios meet. In summary, when we say "technical writer", we are all talking about someone with solid writing skills, sound technical know-how, and a strong willingness to learn. And it doesn’t matter whether you are a newbie or an experienced writer, if you have those three qualities, recruiters will find you hard to reject.
Suyog Ketkar
I’d be interested in seeing if there are even any corporate guidelines in place today on “using” AI. The only guidelines I’ve seen and heard about are that accessing AI is not allowed on company laptops, and companies are actually flagging users that don’t comply with this. From the STC sessions over this weekend and from what I am seeing and hearing in the tech writing domain, it will be a while before this emerging technology is mature enough to get guidelines. It might be happening, but I haven’t seen it. How you consider using AI for work: just some ideas – using content that’s already published and in the public domain: Explore how you can use AI tools to help users FIND the information they need faster; explore if you can feed a large PDF into an AI tool and write examples of targeted prompts to find answers to the most common problems searched for; and focus on using AI to enhance the UX experience rather than trying to use it to generate content. I’m planning to discuss some use cases for using AI in TWT’s job search series that starts next week. If you haven’t registered, now is a good time.
Sumedh Nene
Talking about AI here and not a specific tool, I guess it (AI) will still be on the hit list of several discussions and now interviews too. In terms of guidelines, some organisations have already put strict restrictions (as Sumedh mentioned) on using AI tools on their official systems. The whole idea of AI is that it is trained on the data we feed it. This raises concerns about leaking sensitive or IP data. We, of course, need to be mindful when using any AI tool, whether or not we are using it for personal or official use. The second is that at some point we need to normalise AI. It is going to be a part of our lives, and there’s no escape. People will still use it, and no one’s going to always claim that they did the job using AI. Conclusion: The only guideline is to be mindful when using AI.
I find the whole detection of AI generated content useful during an interview process. As a hiring manager, I want to hire someone who is genuinely good at verbal and written English language skills. However, at work, if I get well-written content, should I really bother if AI was used or not? At school and colleges, it is logical, as one wouldn’t want students to copy and paste without understanding the topic. - If we blindly copy and paste the content from an AI tool, we may run into trouble over plagiarism. - If er blindly copy content from an internal document into the AI tool asking for a review or a better format, we might just be exposing our content to ML. An ideal scenario would be if we can ask for structures synonyms, better ways to put across a thought, and so on.
Mridula Menon
As a future policy, I can see that the Techpubs teams will employ tools to control the use of copies delivered by the generative AI tools. What it means is that technical writers will have a bigger responsibility to write original copies. I also imagine that technical writers will have to develop a better understanding of their domains, products, and services. That will be the only way to differentiate human-developed content from machine-developed content. Imagine that you are working on a data security platform. You are no longer expected to explain the various data protection statutes. Generative AI will give that information to anyone who wants to know about it. It is the implementation of those statutes that need to be explained, which will require innate knowledge of the subject matter.
Ravi Dhumane
This time we are seeing competition with AI earlier there were automatic programs that write content. If our documentation is going public that is indexed by search engines, then we should not forget that these are bots that go through and indexes our content. They will mark it as plagiarised and block the site. They understand whether the content is written by a human writer or a bot. I have seen sites blocked by search engines as they used AI to write their content.
Keshav Arora
A good "writer" can become a technical writer with some guidance and training. Writing is an innate skill that can only be taught to ChatGPT.
I used it for professional reasons to check my doubts. without giving copyright information exclusive to the company. Results were mixed, as I asked the AI for the usage of words. In my view - never use client-specific information to ask the doubt. I put generic words and hypothetical sentences to get results.
Ravi Teja
Guidelines should be set by an organisation. I have used Chat GPT and Bard for 1. Learning concepts (my favorite prompt is - explain to me as you would to your grandma) related to some technical topics I wanted to learn about to write better product docs. I have also used it to convert complex technical sentences to simpler versions. There are many use cases like that. My favorite use case is when I asked for a learning calendar for a certain topic and I got an amazing result when I followed it.
Sowmya Chikalam
Guidelines: Never use the content generated by ChatGPT as is. Give it a tweak. If reviewing, check out Detect GPT - you can detect plagiarism content created by ChatGPT Don't go in your comfort zone that AI will create content. Keep using your creativity and brain lest it rust.
Amit Kapoor
We had an initiative where we tried to leverage ChatGPT in ServiceNow docs. The results were disappointing. It failed to understand the context of developer docs in most cases and provided generic responses. I don't think tech docs need detailed contextual understanding and creativity, ChatGPT has a long way to go. I don't know about other areas. For example, it interpreted integration generically and failed to understand the specific Integration use cases.
Soumen P
Absolutely. ChatGPT in its current stage cannot write documentation for us. It basically extracts data from the internet and returns generic results based on the keywords. TW is beyond just writing a few words, the thought process and human intelligence can't be replaced by it. However, with better and specifically crafted keywords, you may derive certain guidelines or basic text. In my opinion, while it makes it easy for blogs and quick subjective writeups, which you can modify in your own language, it is not suitable for typical technical documentation. Moreover, the content it gives in response can be from anywhere on the internet, susceptible to biased outlook, inaccurate and false information, copyright issues, and plagiarism.
I did use it for personal purposes. I used the free version of it. It's like you really need to know what you want from it. But down the line in the next 5-8 years, there will be tools around it that can take inputs from dev, qa, jira, etc, and create documents. Companies will require lesser TW for their products.
I agree that the job space is gonna change for sure. Still, we can add that human element that hopefully an AI will lack. In my opinion, we can try to leverage it and learn more AI tools to keep up with the changing trends.
Vishakha Rao
We can use AI tools to understand the product while working on API or any other reference guides. It is good to use this where you can understand about a topic or a product, and do in-depth research but not necessarily use it in writing.
Prashant Tiwari
KEY GUIDELINES 1. Never ever expose the clear context. Suppose I wish to ask it to refine text for my UI, I would ask it in terms of common fruit names or common usable objects and very generic context rather than exposing my original requirements. 2. IT IS NOT AlWAYS RIGHT. Merely getting a response from the AI might not help you, you must think of validating the responses in context to your intent. 3. Don't cut off the human intervention because of your fear or ego. Request and accept feedback with open arms. ​
No Comments

Sorry, the comment form is closed at this time.