I run an online news publication that focuses on early-stage tech startups. All the written content I publish is based on interviews and research that I have personally conducted or supervised. Any external references are clearly attributed and fall within the requirements of the Fair Use Doctrine. Respecting copyright is of utmost importance to me.

Although I excel in using words to tell stories, creating visual media has always been a challenge for me. In my previous work in large newsrooms, I relied on visual media experts to complement my articles. However, the WordPress theme I use for my publication requires an image for every article, presenting a dilemma.

Artificial intelligence (AI) has introduced a solution to this problem, but initially, I struggled to embrace it due to my limited skills in visual media creation. Before starting my own publication, I heard about bots scouring the Web… looking for publications using other peoples images. A ‘cease and desist’ with a settlement offer was sure to follow any infraction.  To avoid potential copyright infringement, I initially requested visuals from the startup founders I interviewed. Unfortunately, some of them sent images they grabbed from my competitors, which were copyrighted and posed a risk to me.

To find alternatives, I searched for images in the public domain. While there are a few sites offering free-use images for commercial entities like mine, the options are limited and often fail to adequately support the story I’m telling. Nevertheless, I chose to use them to avoid legal issues and out of respect for other creators’ copyright.

In 2019 or early 2020, I began exploring AI image manipulation projects like Google’s Deep Dream Generator and later, Dall-E. Although these tools seemed promising, my lack of visual creativity hindered me from achieving the desired results. Additionally, I hesitated to use these images, fearing the algorithms’ creators held copyright ownership. This concern evaporated after I learned about the infamous macaque case.

To provide a brief overview, a dispute arose regarding the copyright of selfies taken by Celebes crested macaques using the camera equipment of wildlife photographer David J. Slater. Wikimedia Commons and the blog Techdirt hosted these images despite Slater claiming copyright. People for the Ethical Treatment of Animals (PETA) argued the macaque should hold the copyright. Ultimately, the United States Copyright Office clarified that only works created by humans can be copyrighted under U.S. law, explicitly excluding non-human creators such as animals or machines without human intervention.

This ruling seems to place AI-generated images in the public domain, resolving one of my imagery concerns. However, as mentioned earlier, my abilities in image creation remain limited.

By the spring of 2023, I began experimenting with various AI tools for both images and written content. Midjourney’s impressive image generation capabilities caught my attention, but I lacked the skill to direct it effectively. My solution was to use GPT-3 or ChatGPT to generate prompts for these AI tools.

Here’s an example of how it works:

PROMPT:

Act as a marketing manager for an online business publication covering early stage, seed and pre-seed, tech startups. Create a series of  ‘yes or no’ polls for TikTok to maximize engagement. The questions should have 10 or fewer words, and all should be answerable by either a ‘yes’ or ‘no’ answer. For each poll, also generate prompts for Midjourney to generate an image that best supports that poll question.

CHATGPT:

  • Poll 1:
    Question: “Do you believe in the potential of tech startups?”
    Midjourney Prompt: “Create an image showcasing diverse entrepreneurs working on innovative ideas.”
  • Poll 2:
    Question: “Have you invested in early-stage startups?”
    Midjourney Prompt: “Design an image featuring a stock market graph and startup symbols.”
  • Poll 3:
    Question: “Is your favorite app a product of a tech startup?”
    Midjourney Prompt: “Design an image showcasing popular apps’ logos and their startup origins.”
  • Poll 4:
    Question: “Do you think pre-seed startups need more support?”
    Midjourney Prompt: “Create an image with a sprouting plant representing the growth of pre-seed startups.”