Even the marketing world isn’t exempt from the fears of automation, in fact recently we saw a story about Chase bank replacing its copywriting staff with artificial intelligence (AI). Apparently, the AI outperformed the human copywriting team in a series of tests. Obviously we might be a little biased, but we found it hard to believe that those tests reflected the real-world challenges facing copywritersso we looked at some tests of our own. 

Test 1: Client feedback

Clients will give us the best brief they can, including information about audience and key messaging. It’s up to us to take that brief and fill in the gaps and produce something tangible. Of course, we won’t always get it right first time, and feedback is essential to get to what the client wants. 

But unfortunately, feedback isn’t always as clear as we’d like. Sometimes it can take a lot of creativity to interpret what our clients are after. Often we’ll have comments like:

“Can you make it punchier?”

“I don’t like ‘ing’ words. Please remove the ‘ing’ words.”

“This word just isn’t working for me.”

How would an AI fare with feedback like that? It has already created what it considers to be the optimum piece of work based on the information available. How do you explain to an AI that the brief wasn’t 100% clear, and the client didn’t know what they wanteduntil they saw what they didn’t? 

It would need to be a generalised AI to deal with this situation. But if you’re stuck with the specialist AIs currently available, without detailed, prescriptive instructions it may struggle to understand feedback and will be unable to implement it. Will humans need to act as an intermediary between clients and AI copywriters? Could this be one of the new jobs automation is going to introduce: AI Whisperer?



Creativity and problem solving are still required to effectively meet a typical brief. 



Feedback does not compute.

AI copywriters are specialists that are trained to do specific tasks, if the brief changes they’re not yet intelligent enough to adapt to new situations. Luckily humans are. Currently, we’re better equipped to deal with the changing needs of clients. 

Test 2: Brand guidelines

Brand guidelines are an important part of marketingthey ensure consistency across all marketing materials and help establish an identity for the company. But they can have their limitations: these guidelines won’t cover all eventualitiesparticularly if it’s a piece of content that hasn’t been done before. And there can often be tensions between different arms of the businessdo we brand small business and enterprise differently, what about sales communications compared to internal?

Sometimes it requires a bit of creative interpretation of the rules to reconcile what the client wants and what the rules say. There have been times when we’ve intentionally pushed the boundaries on projects, and have delivered outstanding results. We’re able to do this because we can think outside of the parameters we’ve been given, and because we can communicate effectively with both our client and brand team about the changes we’ve made and why. We can help them understand the limitations of the guidelines and work together to find a solution. 

For AI, sticking to the rules is a given. The AI copywriter will be programmed with specific rules and will follow them to the letter, you’d have to program it to break the rules if you wanted anything different. 



When the rules need to be broken, communication is key. Humans have that one nailed.



Sticking to the rules is what AI is good atbrand guidelines are no exception. 

This one’s a draw. AI is perfectly designed to follow rules and create content within those parameters. But when a little flexibility is required, the human touch can’t be underestimated.

Test 3: Unexpected hiccups 

Few large projects avoid a hiccup along the way. These can be as simple as someone pulling permission for a reference at the last minute to needing to perform a complete rebrand. There have been many times we’ve faced challenges like these at Gloo, but we don’t dwell on the problem. Instead, we develop and solution and call our clients directly to let them know exactly what is happeningso they don’t have any unexpected surprises later on. 

But can AI adapt as quickly as we humans can? Sure, it may be able to solve problems like lost references quite quicklyit will just find a replacement and start the process again. But what about more complicated problems? If the messaging changes at the last minute, the AI copywriter won’t know how to deal with that. It won’t be able to jump on a call with a client, discuss the change and get on with making the updates. 



Humans can adapt to change. We might not always like it, but we can certainly handle it. 



Current AIs are amazing, but not the stuff of science fiction. They don’t do so well if you can’t clearly define a win. 

The humans just scrape the lead on this one. AI may be an excellent system for automating processes, and may be well equipped to deal with minor hiccupsbut when it comes to major challenges the ability to communicate with clients is essential. 

Final score: Humans 3, AI 1

These tests are by no means conclusive, and they only touch on a fraction of what copywriters face in a typical day. But we don’t think we need to be worrying about our jobs just yet. AI may have the competitive edge in some early copywriting test, and it may have the capacity to churn out content but it doesn’t have the human touch required to produce truly great content. And until it learns to communicate as effectively with clients as human copywriters can, we don’t need to worry. 

But then this article was written by a human who is hoping copywriting will be a good career for the next 40 years. Unlikely.

Posted by John on 3 September 2019