Ask HN: Can You Spot AI-Generated Content?
Below are a few of the relevant prompts I used, along with the content I accepted and rejected:
1. Prompt: Generate a list of articles for idli cookers. Include how they are useful, the best brands, and where to buy them.
ChatGPT generated a list of articles, product recommendations, and benefits. Although it couldn't generate accurate Amazon product links, it recommended suitable products and provided explanations. After some due diligence, I curated the links for the recommended products.
2. Prompt: Create a website with all the above content.
ChatGPT generated an outline of the website homepage, including the logo, website name, navigation menu, featured articles, images, and footer. Note: I used DALL-E to create images based on descriptions. Three out of the eight images are DALL-E-generated, while the remaining are stock images.
3. Prompt: Can you create the homepage in HTML, which includes a list of articles based on the website structure?
ChatGPT began generating a basic HTML outline for the website. While this was amazing, I abandoned the code because it lacked styling. I already had an HTML/CSS template that I could use to plug in the content and deploy the website.
4. Prompt: Create a monetization plan for IdliCooker.com.
ChatGPT generated a list of strategies - E-commerce Store, Display Advertising, Sponsored Content, Affiliate Marketing. I updated the products with Amazon affiliate links. (Note the affiliate links are real, so I may get paid a commission if someone actually buys it). Is it this easy to monetize?
Question: To me, the content is authentic and useful. As a human, can you spot the AI-generated content or do you need to be a SME? Does it matter is it is AI or not if the content is useful and relevant?