Observations

Why AI is like a frozen ready meal

As an editor and proofreader, I look at how writing comes across to readers. I guard against anything that could shake their trust in the text, such as muddled language, production errors, and contradictory ideas.

Editing and proofreading are both very different to producing my own writing.

When I write and rewrite, it helps me reflect and come up with new ideas. Editing someone else’s writing is closer to checking the seams and details on a tailored trouser suit.

Writing from scratch allows me to explore what I really think.

AI as a shortcut around thinking

A lot of people are worried about generative AI’s impact on thinking skills.

It writes, reviews text, searches the web, offers advice, and even creates pictures.

We adjust and evaluate its outputs, like tailors making vintage clothes fit to wear again. It’s skilful work, but not the same as constructing a garment from uncut materials.

If we let AI do all the writing, we will all become its editors. We’ll lose opportunities to explore and organise our own thoughts.

Inconvenient conveniences

You may have noticed similarities between generative AI and another convenient and time-saving shortcut.

Forty years ago, frozen ready meals made life easier for a lot of people. They took less time to prepare, and you didn’t need to be a skilled or adventurous cook. They even widened the range of dishes available.

Within a few decades, culinary skills declined in many households. We began to eat without understanding how food was made or how certain ingredients affected our bodies.

Highly processed food has taught us that convenience has downsides. Cooking from scratch is more difficult, but it gives us greater control over our health. It’s also a skill to take pride in.

Looking to the future

In the worst case scenario, AI will become as widely enjoyed as frozen ready meals. What impact will that have on our ability to think from scratch? Will people lose the ability to organise their own thoughts on a page?

I suspect the worst won’t come to pass. Too many people are actively thinking about AI to allow that to happen.

It’s widely understood that technological advances, such as ready meals and AI, carry costs. Since the 1970s, time-saving inventions have changed society again and again.

For example, digital technology has made it possible to avoid travelling to work, shop, visit the bank, and socialise. It’s liberated people but hasn’t been entirely positive for society.

I think this time around, with AI, we’re all well aware of the possible downsides of its convenience.

I believe that while people will appreciate what generative AI can do, they’ll continue to value thinking and thought in all its old complexity.

Knowing what not to say

Knowing what not to say is an important writing skill. It’s something we can learn whether we write traditionally or use generative AI.

This skill matters at all stages of writing, from the blank page to the final draft. It’s about making cuts that improve the text.

There isn’t a universal rule about what to delete and what to save. In each case, we think about what we want to communicate to our readers. Then, we take away anything that might bury the message. 

It’s like removing piles of junk from a handcrafted table so that everyone can see the woodwork. Too many words or digressions in a piece of writing can really obscure its main point.

Cutting questions

Useful content questions to ask are:

  • Why would readers want to know this?
  • What do readers need to know?
  • Would this try a reader’s patience?

Useful language questions to ask are:

  • Can this sentence say the same thing with fewer words?
  • Is this the most efficient way to deliver this point to a reader who is short on time?
  • Is the reader likely to forget the beginning of the sentence before they get to the end?

Is AI any good at cuts?

In the early years of AI-generated output, one sign of AI use was that its language was too much. It tended to turn plain English into overblown phrases that needed cutting. A person didn’t dig holes, they optimised vertical openings in the terrain.

Thankfully, AI’s language is improving. But, no matter how good the technology becomes, people will always need to make decisions about deleting or keeping sections of content.

Only the human writer knows why they’re writing and how they want readers to experience the text. They can try using AI to advise them on cuts, but it’s like asking Google to tell you what to eat for dinner. It’s advice – from a machine.

Working with human editors

Many types of editing are becoming more content oriented these days. So, if you work with an editor and want advice on making cuts, it’s worth talking to them. A human editor can help when you’re not sure what to keep and what to lose.

You’ll need to be clear in how you ask for this because not all editors advise on content cutting. In many editing jobs, suggesting the deletion of content is a serious overstep. That’s why many professionals will be reluctant to do it without explicit permission. 

I can give advice, but only for shorter texts and on the understanding that my knowledge of the subject does not equal that of the client. As an editor, I look at documents in terms of the reader’s experience of the text. This isn’t the same as advising on messaging or strategy.

Explaining what we mean

As a student, I used to redraft my essays a lot. It taught me that academic writing is about building bridges between minds – the reader’s and the writer’s. First, I had to learn how to transfer my thoughts onto the page in an organised and coherent way. Then, to think about what the reader didn’t know and what I needed to explain.

Good academic writing, I discovered, was about helping readers follow my thinking. After all, they wouldn’t know what was inside my head unless I told them. 

I learnt to signpost the connections between my ideas, explain why I thought some things deserved attention, and warn when I was about to change direction or introduce new points.

The role of editors

Getting academic thinking onto a page can be complicated. Academic editors exist because writers recognise the need for help.

Academic editors learn to follow writers’ thoughts very closely and make edits to clarify their expression. I think of it as adding small repairs to a bridge crossing a river.

I still work like that today, when appropriate, even on texts written with AI assistance. With technology adding an extra layer to the writing process, building that bridge isn’t always easy.

How AI is changing editing

Editors tend to be vocabulary enthusiasts. It’s a good thing too; AI can have significant problems with the meanings of words.

Vocabulary is such a sensitive and nuanced area that AI sometimes struggles with it.

Words can be very similar but vary in rightness for a situation, strength, feeling, and fit. Many have more than one meaning and evolve through use.

When AI makes mistakes with word choices, it confuses readers and introduces contradictions into texts. Even when words are technically correct, they can be too obscure or communicate the wrong tone. It’s now the editor’s job to sort that out.

I’m looking forward to seeing how AI develops to tackle this vocabulary problem in the future. I believe it has improved in recent years.

Even in a world with very advanced AI, people will always need to learn how to explain what they’re thinking. I suspect that traditional academic writing skills will be valued for a long time yet.