I resisted using artificial intelligence for serious work tasks longer than most. I’ve built much of my career on being a good writer. News reports, technical documents, website content, social media posts, research papers – I’ve been doing it for years! I didn’t need an algorithm in some data center out in middle of the desert to do my job!
But tech marches forward, and the need to deliver faster started catching up. It became clear that the demands of today’s workplace were rapidly outpacing what we overworked humans could expect to sustainably deliver. In the U.S., 80% of workers report experiencing “productivity anxiety,” and a similar share globally say they don’t have enough time or energy to complete their work. Time spent in meetings, answering emails, and task switching between the ever-multiplying number of productivity apps has skyrocketed over the past few years.
Consider me a statistic.
So I finally relented to the efficiency machines – slowly at first. I began using AI cautiously while working for a USAID project, letting it assist occasionally with things like copy editing, condensing long paragraphs, and revising texts written by non-native English speakers. Then I joined a digital agency with an IT team, where experimenting with AI is an established part of daily work life. I had a lot to learn.
I’ve accepted that AI can indeed be a powerful assistant for generating content – a sounding board when I hit writer’s block, a copy editor when I’m in a rush, or a research partner to help me jump-start a new topic.
But let me repeat: it’s an assistant – not a replacement for you, dear human.
AI-generated content is everywhere. Your LinkedIn and Instagram feeds are flooded with it – and much of it is mindless slop. While I’m not the stubborn skeptic I was a year ago, I’ve learned that we humans, despite our penchant for typos and mid-afternoon brain fog, are often still the difference between quality and crap.
Research pitfalls: always check your (AI’s) sources
Content is always stronger when it makes smart use of research and real-world examples. AI can be a great help with this, and it’s more than happy to provide you with the perfect statistics and information to make your post more compelling – but beware!
Chatbots have a tendency to hallucinate, misrepresent data, or make it up entirely. Sometimes, when I ask for a source of the information it just gave me, it admits it can’t find one – or worse, cites something that doesn’t even exist.
Custom instructions can help:
Only use verified sources…always provide direct links to sources you cite…clearly mark when examples are meant as placeholders or illustrative….
Still, there seems to be no miracle pill to fully inoculate your chatbot buddy against digital hallucinations or entirely temper its pre-programmed urge to affirm the veracity of your every late-night epiphany.
Some tools are better for this than others. I’ve found NotebookLM useful for synthesizing research and information because it focuses on the sources you feed it – research papers, essays, news articles, even YouTube videos – rather than scouring the web or its trained memory for answers. When AI is limited to a narrow range of sources, it’s less likely to make up nonsense. Still, every tool requires verification and interpretation. AI can help surface insights, but it can’t quite decide which ones matter.
Train new content creators to be cautious
Many organizations now expect staff to work across roles. A project manager in a small IT firm might be asked to draft marketing copy or social media posts. Not everyone is a writer or has a communications background, and this is where AI really can help. It can get a non-writer started by offering structure and refining messy drafts into something more polished.
But there’s a danger in hitting the “generate” button and calling it done. Teams need to learn to check AI output for accuracy, tone, and relevance. And, crucially, they need to understand the communication goals of their organization or project before typing prompts into their chatbot.
Better inputs = better outputs.
That means knowing roughly what you want the end product to look like and providing clear, detailed, and specific instructions to help the AI tool generate what you need.
AI doesn’t eliminate the need for editorial oversight, but it does change what oversight can look like. When used thoughtfully, it can speed up the content review process. When used carelessly, it can make the process longer than just doing things the old human way, as already-overworked editors disentangle half-true or tone-deaf drafts.
Beyond text: the promise and peril of AI-generated media
AI image, video, and audio tools are making it easier for small teams to do more with limited budgets. I was amazed to discover that NotebookLM can turn lengthy articles and essays into commute-friendly podcast episodes with realistic-sounding hosts who “discuss” and explain the materials. At first listen, it’s impressive. There’s even a temptation to publish this miraculous content, until you realize the same AI personalities, the “Deep Dive” duo, can be heard across literally thousands of podcasts on Apple and Spotify.
Certainly, AI offers a lot of possibilities for speeding up podcast production. Podcasts are a great way to learn about a topic for many people, and there’s nothing inherently wrong with leveraging LLMs to assist with content or using AI tools to enhance someone’s voice – or even produce entirely synthetic voice overs. The challenge is making sure the final product is accurate, interesting, and unique. That requires a workflow. That requires creativity. That still requires us.
The race to create can’t be a race to the bottom
It’s seductive to think we can use AI tools to fill content calendars quickly, but we must resist laziness. Synthetic media can be part of the creative process, but it’s not the end of it. AI can assist with things like coming up with ideas, refining text, or producing and editing visuals. It can even help non-programmers like me with custom code for WordPress (the platform I used to build this website) and explain how to use it. But when it comes to content, we humans are still needed to shape the narrative, verify information, and ensure authenticity and originality so it doesn’t get lost in the sea of brainless AI drivel.
I’m still exploring these tools and finding new things we can do with them. Used wisely, they can help create engaging, accessible, and even experimental content. But they need management, curiosity, critical thinking, and some attempts at humor – the same things that made us writers and content creators in the first place, and the things that LLMs just can’t really do on their own right now.
So you may be wondering – did AI write this?
I’ll admit that my merry LLM assistants (yes, I have a small team now) helped with finding background information and did some light editing of an earlier draft – but the em dashes are all mine! You can pry them from my cold, dead MacBook keyboard.