Will AI reduce the need for technical writers?

The late Stephen Hawking famously said that artificial intelligence would be “either the best, or the worst thing, ever to happen to humanity.” As a technical writer documenting AI technology, I’d like to believe it would be the former and it’s fair to say there have already seen positive signs about how AI might shape and assist with documentation in the future.

A number of tech companies have already dipped their toes into the water, with some developing AI-assisted, predictive content generation and others harnessing machine learning to predict the help content the end-user is looking for.

AI-assisted content

Google introduced its natural language processing development, Smart Compose, to help Gmail users write emails in May 2018. They combined a bag-of-words (BoW) model with a recurring-neural-network (RNN) model to predict the next word or word sequence the user will type depending on the prefix word sequence they wrote previously.

model3
The Smart Compose RNN-LM model architecture.

Smart Compose was trained with a corpus of billions of words, phrases and sentences, and Google carried out vigorous testing to make sure the model only memorised the common phrases used by its many users. The Google team admits it has more work to do and is working on incorporating their own personal language models that will more accurately emulate each individual’s style of writing.

Arguably one of their biggest challenges they face is reducing the human-like biases and subsequent unwanted and prejudicial word associations that AI inherits from a corpus of written text. Google cited research by Caliskan et al which found that machine learning models absorbed stereotyped biases. At the most basic level, the models associated flower words with something pleasant and insect words with something unpleasant. More worryingly, the research found the machine-learning models also adopted racial and gender biases.

f1.large

The research found that a group of European American names were more readily associated with pleasant than unpleasant terms when compared to a batch of African American names. Researchers also found inherited biases included associating female names and words with family and the arts while male names were associated with career and science words. Yonghui Wu, the principal engineer from the Google Brain team, said: “…these associations are deeply entangled in natural language data, which presents a considerable challenge to building any language model. We are actively researching ways to continue to reduce potential biases in our training procedures.”

AI-assisted spelling and grammar

With 6.9 million daily users, one of the most common tools people are using to assist the accuracy of their spelling and grammar is Grammarly. The company are experimenting with AI techniques including machine learning and natural language processing so the software can essentially understand human language and come up with writing enhancements.

crisp-writing-alter-ego-communications

Grammarly has been training different algorithms to measure the coherence of naturally-written text using a corpus of text compiled from public sources including Yahoo Answers, Yelp Reviews and government emails. The models they have experimented with include:

  • Entity-based model – Tracks specific entities in the text. For example, if it finds the word “computer” in multiple sentences it assumes they are related to each other.
  • Lexical coherence graph – Treats sentences as nodes in a graph with connections (“edges”) for sentences that contain pairs of similar words. For example, it connects sentences containing “Macbook” and “Chromebook” because they are both probably about laptop computers.
  • Deep learning model – Neural networks that capture the meaning of each sentence and are able to combine these sentence representations to learn the overall meaning of a document.

Although this is still a work in progress, their long term goal is for Grammarly to not only tell you how coherent your writing is but to also highlight which passages are difficult to follow.

AI-assisted help content

Some companies have started to look at ways that AI can help with predicting and directing readers to the exact content they are looking for. London-based smart bank Monzo launched a machine-learning powered help system for their mobile app in August 2017.

monzo

Their data science team trained a model of recurring-neural-networks (RNNs) with commonly asked customer support questions to make predictions based on a sequence of actions or “event time series”. For example:

User logs in → goes to Payments → goes to Scheduled payments → goes to Help.

At this point, the help system provides suggestions relating to payments and as the user starts typing, returns common questions and answers relating to scheduled payments. Their initial tests showed they were able to reach 53% accuracy when determining the top three potential categories that users were looking for out of 50 possible support categories. You can read more about their help search algorithm here.

What does the future hold?

I think we will see more content composition tools like Smart Compose emerge but I think it will take a lot of time and work before they can be trained to effectively assist with the complex and often unpredictable user-oriented content that technical writers are tasked with producing on a daily basis.

datacenter-future-e1503509968416

I’m sure some technical writers are already using Grammarly to assist with their spelling and grammar. It can be a really powerful tool to ensure your text is not only accurate but in the future will be able to measure the coherence of your writing. I’ve dabbled with Grammarly but found it either wasn’t compatible with certain tools or prevented some of my applications from working so it became a bit of hindrance rather than an assistant for me personally. No doubt these are kinks they will iron out at some point down the line.

I do see the benefits of AI-assisted help so it would be awesome to see some more development in this area. It really could be something that saves customer support and documentation teams a lot of time in terms of predicting and directing end-users to answers before they’ve even asked a question.

So are we there yet? Not quite… but I think some very promising foundations have been laid. While some technical writers might be concerned, I think it will be a very long time before AI is advanced enough to supplant our role in the development teams. So don’t be afraid of AI, for the time being these tools are only going to make our lives easier!

Advertisements

The Story behind Slate: An Interview with Robert Lord

Back in 2013, developer Robert Lord, then an 18-year-old intern at Tripit travel software company, was challenged to create an API documentation tool by his boss. It took him several weeks but the result was a beautiful, responsive API documentation generator called Slate. Five years later, it has grown into a popular open-source tool that is used by a number of global organisations and companies including NASA, IBM and Coinbase.

cover-image-for-above-the-entire-post
Robert Lord created Slate while interning at Tripit. Copyright© Concur Technologies,

Lord said the Slate project grew out of a set of requirements the Tripit engineering team had at the time. He said: “I was interning at TripIt and my boss pointed me towards some two-column documentation pages and said ‘We’d like a page like this for our new API.’ They also had the requirement that their technical writer could make changes, and I think they didn’t want to write raw HTML. I made a generator that ended up being pretty generic to any documentation, and convinced them to let me open source it.”

How to Use Slate

Slate is simple to use, you fork the Slate Github repository and create a clone. Next you customise the code to meet your requirements; adding a custom logo, fonts and any additional CSS styling in the source folders, before adding your API endpoints and their descriptions in Markdown.

screenshot-slate
Slate helps users to create beautiful, intelligent and responsive API documentation.

When you’re done, you start Slate and launch your API documentation site using Vagrant or create an image using Docker. The result is an attractive, responsive three-panelled API documentation site with code samples in multiple languages down one side and a smooth scrolling table of contents down the other. For more information on how to use Slate, follow the instructions in the Slate README.

Slate in the Wild

Today more than 90 people have contributed to Slate on Github, it has been forked more than 13,000 times and has been given more than 23,000 stars. Some of the organisations and companies listed as users include NASA, IBM, Sony, Monzo, Skyscanner and Coinbase. There is a list of more than 90 companies that have used it on the Slate in the Wild sub-page of the repository.

This slideshow requires JavaScript.

Lord admits he still finds it “pretty surreal” that such large companies have adopted what he labels the “buggy project” he created as a teenager. “I really did not expect anybody else to see it or care about it,” he said. “Slate never really had a big rush of new users all at once, the growth in stars has been more or less linear over the years. No hockey sticks here. So there was never a single moment where suddenly a bunch of people were using it, it was a very slow process of discovering one company at a time.”

Life after Slate

Interestingly, a year after working at Tripit, Lord interned at Stripe, one of the leading API-first companies whose own API documentation inspired him when creating Slate. Stripe realised the value of their product hinged on people being able to read and digest their APIs. They invested a lot of time and effort in developing their own in-house API documentation tool and set the bar for the rest of the industry with the two-panelled design that has inspired so many other API tools.

Lord had plans to develop further API tools but decided to focus on other things. “Initially had some plans for similar tools,” he said. “But I think I realized I’m still early in my career, and would rather branch out and work on a variety of projects instead of focusing in on just one area.” Despite moving onto other projects and being fairly modest about the success of Slate, it’s an impressive piece of work for the young developer to put on his resumé. Indeed, one of the main reasons he asked Tripit to allow him to open source the project was so he could show future employers his work. “I mostly convinced them to open source it just so I could point future employers to this chunk of code I wrote,” he said. One company clearly took notice, Lord starts work on Fuschia at Google in a few of weeks time.

👽 The Emoji Invasion

A critic labelled the ‘text speak’ of the 1990s as “penmanship for the illiterates” but the latest threat to written English is the emoji, said to be the fastest growing language in the UK. While ‘text speak’ saw words shortened and abbreviated, emojis have replaced text altogether, harking back to the dark ages of cavemen and hieroglyphics, when pictures formed the basis of communication.

The rapid spread of emojis into modern communication has seen a translation company hiring the world’s first emoji translator, a restaurant launched in London with an emoji menu and the recent release of the Emoji Movie in our cinemas. So, where did they come from and is there a place for them in modern communication and technical writing?

🎬 Origins of the Emoji

The emoji first appeared on mobile phones in Japan during the late 1990s to support users’ obsession with images. Shigetaka Kurita, who was working for NTT DoCoMo (the largest mobile-phone operator in Japan), felt digital communication robbed people of the ability to communicate emotion. His answer was the emoji – which comes from the Japanese ‘e’ (絵) meaning “picture” and ‘moji’ (文字) “character”.

2000
One of the original set of 176 emojis designed by Shigetaka Kurita

The original emojis were black and white, confined to 12 x 12 pixels without much variation. These were based on marks used in weather forecasts and kanji characters, the logographic Chinese characters used in Japanese written language. The first colour emojis appeared in 1999and other mobile carriers started to design their own versions, introducing the smiling yellow faces that we see today.

aa-Cover-4fu6auljfsrl2jr5qirr1deh76-20161217051451.Medi
Shigetawa Kurita, the 👨‍👩‍👧 father of emojis, felt digital communication robbed people of emotion.

Speaking to the Guardian, Kurita admitted he was surprised at the popularity of emojis. “I didn’t assume that emoji would spread and become so popular internationally,” he said. “I’m surprised at how widespread they have become. Then again, they are universal, so they are useful communication tools that transcend language.”

“I’m surprised at how widespread they have become. Then again, they are universal, so they are useful communication tools that transcend language” – Shigetaka Kurita.

However, Kurita doesn’t believe emojis will threaten the written word. “I don’t accept that the use of emoji is a sign that people are losing the ability to communicate with words, or that they have a limited vocabulary,” he said. “Some people said the same about anime and manga, but those fears were never realised (…) Emoji have grown because they meet a need among mobile phone users. I accept that it’s difficult to use emoji to express complicated or nuanced feelings, but they are great for getting the general message across.”

💹 Emojis in Marketing

It is this ability to get the message across very simply that has resulted in companies using emojis more and more in marketing, particularly on platforms like Twitter and via email. It has become a way for brands to humanise themselves, have a sense of humour or put across a message that a younger audience can relate to.

One example of emoji marketing is a Tweet sent by Budweiser which was composed entirely of emojis to celebrate the 4th July this year:

Meanwhile, Twentieth Century Fox took emoji-based humour to whole new level with posters and billboards bearing two emojis and a letter (💀💩L) to announce the release of Deadpool in 2016:

Deadpool.png

✍️ Emojis in Technical Writing

A number of tech companies, especially those with a younger (in their 20s-40s) target audience like Slack and Emoji, have also embraced the use of emojis in their technical documentation and the software itself.

Slack3.png

Slack use them sporadically in the product, often as the punchline of a joke or message when you’ve read all unread threads (see screenshot above).

Emojis also appear in their help system, with Emoji flags for the chosen language and to highlight bullet points (see below):

Slack-emoji2

Startup bank Monzo also embraced emojis early on, designing an emoji-rich interface that would a younger client base than typical banks could relate to. Emojis are automatically assigned to transactions and you’ll find them incorporated in the Monzo API documentation and the app’s Help screen:

Monzo-mojis

iPhone6---Front-Black---by-JustD

Speaking to brand consultants Wolf Olins, CEO Tom Blomfield explained how they also use machine learning to pair your transaction’s spending category with relevant emoji. For example, it will display the doughnut emoji 🍩 when you shop at Dunkin Donuts. He said: “There’s no business case for the emoji donut, but people get ecstatically happy when seeing it and go on social media to share the moment.”

☠️ Risks of using Emojis

While emojis might work for some tech companies and give them a way to humanise their brand and relate to their target audience. I think there are several risks which come with their use as well.

The first risk is alienating users who don’t relate to emojis, or even dislike them. Although most of my office do use them as a way to react to each others’ Slack posts etc, there are a number of people who refuse and as there are a lot of nationalities with different cultural references, sometimes the emojis are used in different ways. For example, in Japan the poop emoji (💩) is used for luck while the English use is a lot more literal. Similarly, the folded hands emoji (🙏) means ‘thank you’ in Japan, while it is more commonly used to convey praying or saying ‘please’ in English usage.

Secondly, if emojis are just a fad like the Kardashians, Pokémon GO and Tamagotchi then you face the unpleasant task of replacing them all when they become unpopular, are considering annoying or are phased out. If you have saturated both your product and documentation with emojis then this task will take you and your team a lot of time and effort.

Thirdly and finally, studies have shown that emojis can get lost in translation as they are incredibly subjective so the meaning and intended emotive message can often be misinterpreted. This has continued to get more and more muddled as different vendors and browsers redesign and create their own versions of the unicode emoji characters. A study by GroupLens research lab found evidence of misinterpretation from emoji-based communication, often stemming from emojis appearing differently on different platforms.

Screen_Shot_2016-04-11_at_8.49.09_AM

The grimace emoji (😬) is said to cause the most confusion, with researchers finding that 70% of people believed it to be a negative reaction while 30% thought it conveyed a position emotion.

On the whole I don’t dislike emojis or think they’re a threat to the written word. They definitely have a role to play in social interaction, can humanise communication and even add humour to it. However, I still feel there are too many risks, too many different cultural interpretations which mean they simply won’t work in a multinational business. Technical writing is all about choosing the clearest form of communication, the shortest, most simple words that cannot be misunderstood. I’m just not convinced there’s a place for emojis in documentation yet, at least not while there is still room for things to get lost in translation.