Introduction
Will AI eliminate your job as a tech writer? From the research I’m doing and courses I’m taking, the answer seems an obvious ‘No’. Currently, AI is a hot topic with discussions ranging from ‘it is just another search engine to ‘it’s alive and plans to kill all humans and take over the world’!
My personal hunch is that “human creativity, expertise, and emotional understanding will continue to play a crucial role because, it is up to the user to ‘ask the right question in the right form’ and not delve into ‘forbidden territory’.
Although AI is a major technological breakthrough that will have widespread implications across most industries, it is not new. AI has been in our everyday lives for over a decade (think of Grammarly, Facebook/ Google ads, search engines, auto-fill) — so why has it become such a hot and often scary topic recently? Two key reasons:
- it’s being applied to creative industries (art, music, writing, video)
- it has become generative, meaning it can create something from scratch.
However, AI is currently unable to write anything long-form. Example: GPT-4’s ideal length for a single response is typically around 200-300 words.
An Example from Previous Technology
But it took a lot of time to get to where we are today. Programming computers (let alone AI engines) is obviously complicated. It takes skill and patience to think through how something is done (something technical writers excel at doing).
Writing Code 101
I remember my first computer, a Model 1, Radio Shack, TSR-80. It used cassette tape cartridges to store data and, even with a home-made, expansion interface, the hulking machine would only provide about 1.77 Megahertz of processing speed with a limit of 52 bytes of storage on a cassette tape cartridge!
Most early adopters had no alternatives other than to key in programs themselves. BYTE or PC User magazine would publish code in the back section of the magazines. Here is an example of BASIC code for the Star Trek game.
From the densely-packed, printed pages, you would type each number, letter or symbol into your keyboard. Once finished, you launched (compiled) it and held your breath to see if it worked. If it did not, you spent more hours searching for the one letter, number or symbol that was typed incorrectly, corrected it and recompiled. Often it became an iterative process of check, fix and try again.
All this work, knowing that most of these games would be boring as all get out, For example my favourite game was based on the Star Trek TV series. The ‘Enterprise’ and ‘Klingon’ space ships were represented by bouncing letters and symbols on the screen – nothing slick. You could move the ‘ships’ around on never ending adventures through space with key strokes (W for up, A for left, S for down and D for right. There was no mouse or directional arrow keys).
All that said, it is obvious that computer programs are light years ahead of those early days. Creating a system to achieve ‘Artificial Intelligence’ is exponentially more complicated than programming. So we must be patient with the new tools.
The Difficulties with AI
Click on this poster to go to the TED Talk, ‘The danger of AI is weirder than you think’ by Janelle Shane. It highlights a few of the significant difficulties with AI and will give you confidence to keep developing your technical writing skills.
Language Model Basics
So, what are some tools you can begin exploring, if you have not already started. Here are links to some basic information about the most recognised examples of language models that can generate natural language responses based on user input.
- ChatGPT
- Microsoft’s Edge’s Bing – ‘Prometheus’
- Google’s ‘Bard’
Their ‘conversational nature’ is powered by artificial intelligence (AI) that has been fed large amounts of data to ‘learn’ from. The tools can be used to answer questions, point out examples and references, generate content, provide feedback and spark creativity.
It is important that you know more about these tools so you can leverage them to your advantage, because interest in the potential of these tools is growing.
Obviously, the tools are imperfect and can sometimes produce inaccurate, misleading, or biased information. Therefore, it is important to use them with caution and critical thinking, and not blindly trust their outputs. These tools are only as good as the data and algorithms they rely on and the person asking the questions.
For part two of this article see the next issue for ‘How AI Can Benefit Technical Writers’
by Darlene Richard, www.Write4You.co (not com)