Most prompt engineers are just writers in disguise. Prompt engineering is the fancy word boffins building ChatGPT, Gemini and Claude use to describe telling AI who you are and what you want from it. Amanda Askell is a philosopher who works for Antrophic, the company behind Claude. She says the term “engineering” is apt because someone crafting a prompt defines a problem, develops a solution and refines their work through testing. I asked a real-world engineer friend about this take. He explained he’d solve a problem like a congested road traffic junction using a similar approach: define the problem, develop a proposed solution, and refine it using data or feedback. Listening to the Askell interview again, it struck me that writers work the same way. Figure out what you want to say, develop a means of expressing it, and then refine a draft using feedback from readers or an editor. Like an AI tool, creating like this is much easier if you have constraints and examples of an ideal output. Go backward and forward a few times until you have something publishable (or you’re out of time). You can even use data in the form of content analytics to adapt your content strategy. It’s super easy to upload a CSV file or link a Google Doc to Claude or GPT these days. Prompt engineering sounds more highbrow than prompt writing, but these terms are interchangeable. That said, it’s helpful to understand a few other core concepts if you’re engineering or writing prompts. Don’t let these terms flash a red light when you want to experiment with AI. Persona: Assign a specific role to the AI (e.g., “Act as a copywriting expert”). Structured Output: Specify desired formats (e.g., tables, lists, or LinkedIn posts). Prompt Chaining: Sequence prompts to refine and guide the output progressively. Chain of Thought: Ask for step-by-step reasoning to improve clarity and accuracy. Tree of Thought: Explore multiple reasoning paths simultaneously for complex tasks. I’d need more than one email to explain how all these work. If you want to learn more, I’m testing a paid newsletter offering custom Claude and ChatGPT prompts for creators. I’m testing this with a small group - reply ‘AI’ if you want first access.
|
Do you want to grow a profitable content business? If so, join 25,000 readers. I share daily insights about personal brand building, writing online and growing a profitable business. It's for creators, coaches and business owners.
If you're spending hours typing out GPT prompts, here's a neat tip to save you a few hours. Use voice prompts and transcriptions. They can generate more accurate results than laboring over a 1,000+ word text prompt. The other day, I was meeting with a client online. He explained a problem he’s facing with his business and what he wants to do this year. I offer content strategy services, but rather than pitching him on the call, I hit the transcribe button in Google Meet (both parties get a...
DeepSeek wiped a trillion off the US stock market this week. Oddly enough, it took six days for the market to react to DeepSeek, which launched on the 22nd. On its website “DeepSeek-R1 is now live and open source, rivaling OpenAI’s Model o1.” The Chinese startup behind it says DeepSeek is 20–50 times cheaper to use than the latest OpenAI o1 model. It’s number one in the App Store as I write this. It’s also a big target for cyber attackers, and registrations are limited. I got in before the...
A few days ago, I wrote a note on Substack that popped off. I write these notes every day. I spent five minutes on this one. My note went viral, attracting over 1000 likes, shares, and comments agreeing and disagreeing with my contrarian take. Some other big creators restacked or shared it, too. And I attracted a few dozen new subscribers for my Substack list: Not bad for five minutes of work. I’ve spent far more time on newsletters than have done far less for my brand. So what’s the takeaway...