How often do you open Google to look up a synonym with your Thesaurus of choice? A writer's life is quite challenging as it is, and doing that as a job means that sometimes your brain just won't collaborate. Luckily, researchers all around the world are working to harness the power of AI and give those poor writers a hand.
Though the rise of Skynet and full consciousness of AI are quite far down the road, there are already some tools out there that, through machine learning or some other form of wizardry, can make the life of a writer much, much easier.
We've had the chance to ask Casey Halter (here on the left), copywriter and strategist at Trollback+Company, which of these tools are the best currently available for creative teams, from a copywriter's perspective. From Grammarly's Tone Detector to minimal, yet effective dictionaries, these tools are a significant helping hand for writers from all paths of life – while keeping the fear of AI safely at bay.
Without furder ado, here's a guide to the best AI tools for copywriters, in the words of Casey herself.
The best AI tools for writers – A Copywriter's Guide
Online tools have gotten incredibly adept at acting, sounding, and behaving like real human writers. Could this help brands and agencies communicate better while we all work from home?
As a copywriter, my brain often already kind of feels like a computer. Case studies sound like this. Taglines work like this. Social media posts must be this length, use roughly this many emojis, and please no, that meme format is far too old for you to be using it.
This in-depth knowledge of how to communicate to a vast array of audiences, content types, and contexts is exactly what makes a good copywriter. It’s also what makes us a vital asset to brands and agencies alike.
But with most of us working from home these days, these quick copy gut-checks – typically reserved for a casual “Do you have a second?” or “Just-so-happened to see you there” chat – often get lost in the chaos.
And while this might seem like a strange time to bring AI into the conversation, I’m going to do just that. As someone who’s both fascinated by computer intelligence and expert wordsmithing, I’m a firm believer that digital assistants could open up a new frontier for writers, brands, and businesses. It’s time to get up to speed, and I’m going to show you how to do just that.
Now, hear me out. I am not recommending that everyone replace their human copywriters with digital assistants until we all end up back at the office. If anything, you could argue that in these trying times, copywriters are more valuable than ever. After all, who else is going to emergency-pivot your brand’s communication strategy in the face of a life-threatening pandemic?
However, there are several online writing tools that can step in as a quick fix as we all get through this together – from grammar, to tone of voice, even copy creation.
One of my favorites is Grammarly’s Tone Detector tool, a computer-generated editing platform introduced by the wildly popular online writing start-up in late 2019. The program works by copy through a combination of rules and machine learning to identify and adjust key tonal signals in writing – essentially the equivalent of tapping your copywriter on the shoulder and asking: "Can you rewrite this for me?"
Grammarly's Tone Detector helps writers of all skill levels finely tune their copy to their context and audience
By answering prompts like “Who are you speaking to?”, “What’s your context?”, “What’s your level of formality?”, Tone Detector can help writers of all skill levels more finely tune their copy to their context and their audience. In return, Grammarly’s AI engines get some serious help from writers, who are literally training the tool to become a better, more “human” writer with every adjustment.
Over the past year, I’ve already learned a lot from Grammarly’s new Tone Detector — from words I repeat all the time, to solid advice on how to cut down longwinded sentences. The tool also helps me as an editor, checking for plagiarism from other writers, showing how long a script will read without having to mutter it under my breath at my desk, and even pinpointing the percentage of the English-speaking readers that will be able to understand what I’m writing.
For further help on word choice and tone, I also use online language tools like OneLook and PowerThesaurus, both hyper-minimal yet super-robust databases of synonyms, antonyms, rhymes, and phrases collected from the far reaches of the Internet – from online dictionaries to online search queries.
One of my favorite tools in the former is a phrase tool located at the bottom of the page. Plug in any word into the website’s search bar, and OneLook’s algorithm will show you all of the common words and phrases that commonly appear next to your search query.
In my experience, this can be hugely helpful for adding a human flow to stories and bringing in interesting, unexpected connections into your copy. I’ve even used it for tagline ideation, exploring all the different idioms and memorable catch-phrases, then making them my own with a simple twist or adaptation.
Another major leap in the AI writing landscape I recently discovered comes from IBM, which released a very cool Tone Analyzer tool early last year that’s designed to help writers retroactively assess their content for intent, personality, and purpose.
The goal, IBM writes, is to help human readers understand emotions and communication style in text. Sounds like a great tool for essential businesses to use as they’re dealing with an influx of customer service requests and technological support, no?
As a copywriter, IBM’s Tone Analyzer also works great as a gut check when writing tone of voice guides, social media posts, and sample ad copy – sort of digital friend that you can endlessly ask: “Does this sound too insensitive?”. I used IBM’s system recently while writing a new website for our company, running about 14 different webpages through the analyzer. Some pages were logical. Others, passionate. Others, straightforward. Understanding that I was using a different tone to describe work vs. about us vs. new business inquiries showed me that I had to button things up quite a bit. I kept going until each page lined up with my target tone combo – “Passionate, Logical, Creative.”
Computers may one day be able to determine from inputted language who people are, how they feel and how they think
With improvements in tools like Watson’s Tone Analyzer, IBM researchers say they hope computers may one day be able to use inputted language from around the world to determine who people are, how they feel, and how they think. As a strategist, this information could also be hugely valuable to me from an audience insights perspective. It would also help with the classic writer’s conundrum of having to switch back and forth between voices and ideas quickly with little direct feedback beyond your client’s glitchy Zoom reaction.
And while it’s still not always perfect (please, debate me on what the word ‘Passionate’ truly means from a data perspective) it also opens up a new frontier of informed, emotion-based writing. By design, the more we use these tools, the better they will get, helping to leave out a lot of guesswork when it comes to nailing down an argument, an appeal, or a tone of voice.
But what about AI tools that can churn out hundreds of words a day as we all focus on saving the world and our businesses? Don’t worry – writers, linguists, and computer scientists are working on that, too.
Over the past few years, several companies and universities have been developing computer models trained to generate sentences and even paragraphs of text on a variety of topics – from fantasy novels to business correspondence. In fact, during my research for this piece, I quickly discovered that digital algorithms are already used across a variety of business intelligence dashboards, data reports, and yes, even those ‘too real’ personalized email responses you get on Gmail, every day.
A better example of this is Heliograf, the Washington Post’s in-house automated storytelling technology. The tool, developed to focus on hyperlocal sports stories in 2017, may seem a bit unnecessary as we’re all social distancing and watching Netflix at home. However, it does speak to a potential future where quick news updates can be disseminated without the need of endangering or wasting resources on a journalist that could be covering more pressing stories.
By weaving together scoring plays, player stats, quarterly score changes, and more, Heliograf is able to craft compellingly human-sounding articles and tweets that get the job done (even if they are a bit robotic). As a result, the paper can still stay connected to the heart of its community, even when they’re not physically there to call the plays.
There are also some scary new writing tools on the horizon
That said, I also found some scary new writing tools on the horizon I think we, as writers, should all keep an eye on. For example, OpenAI’s GP2, a text-generating system designed last year to write page-long responses to random prompts – from Lord of the Rings fanfiction to high school homework assignments.
The model, developed by Microsoft-backed nonprofit, works by accessing a huge, transformer-based language model with 1.5 million parameters.
It was apparently so good, its creators refused to make it publically available due to concerns over deep fakes, fraud, and other “malicious applications” of the technology. They are now working to “fine-tune” it to detect and avoid misuse. However, that future still, admittedly, feels a long way off.
Is This the AI Writing Revolution?
So, what does the future hold for digital communication in the post COVID-19 era?
For now, none of these systems are ready to replace us just yet. To be honest with you, I’ve already made it my mission to beat Grammarly’s tone of voice and grammar editor every time, and am fairly good at pre-empting IBM’s Tone Analyzer on my own. Even OpenAI (the scary one) suffered from repetitive text and world modelling failures (i.e. a fire burning underwater), said researchers – all issues that would have a hard time tricking a human reader anytime soon.
On the more positive end of the spectrum, however, improvements in AI writing assistants can help brands and agencies alike edit, streamline, inform what we do in the meantime — from better online translations, to better chatbot encounters.
As a writer, I think we should all be encouraging our co-workers to use AI to improve their writing as we continue our remote, work-from-home schedules. And if one day a computer comes to take my job… well, there’s always strategy.