Find a niche, they said. It’ll be fun, they said. After a full two decades in both personal and corporate blogging, I thought I’d found all of the content niches I could possibly find in a lifetime. Enter generative AI – and the Krazy Fish team and I are back to exploring new niches. Suddenly, AI for marketing is a thing and so AI for business must be a thing. Just use ChatGPT to whip up some content, post it on your website, then hop onto Jasper.ai to whip up a Facebook Ads campaign, and let ‘er rip. But it’s not all quite as cookie cutter simple as it seems.
In the world of marketing and communications, innovation is the lifeblood that keeps strategies fresh and competitive. From the invention of the telegraph to the rise of the web, the ever-changing landscape of how people and businesses communicate is the very reason many of us became enamored with this profession. The newest buzzing innovation, generative AI, can indeed enhance nearly every aspect of digital marketing and SEO.
However, they are not intended to replace the human expertise of our team, but rather to augment it, freeing up the humans to focus on what we do best – strategy, decision-making, and creativity. In our ceaseless quest to explore new frontiers, the Krazy Fishes have embarked on a fascinating journey – leveraging the prowess of generative AI tools to create content that matters and makes a difference for humans. Our ongoing experiment to this end – a recipe blog – presents an exciting fusion of technology and creativity.
Get to the Point
Why a Recipe Blog?
Our approach is straightforward. We use ChatGPT-4, an advanced AI model by OpenAI, to generate recipes on Pinch of AI. Aided by human planning and crafting prompts, we use this tool to create an endless string of interesting, unique, and mouthwatering dishes. Every recipe is a delightful surprise and a testament to the intricate dance between human ingenuity and generative AI’s vast practical potential. And that’s what we are out to test – the practical implementations of ChatGPT in achieving marketing and business growth goals for a little website just getting started in a popular niche.
But why did we choose to use all of these new tools to create a recipe blog, of all things? Well, people have to eat. It’s that simple. Food blogging and recipe blogs continue to be among the most popular general niches on the internet. Within that fairly wide topic, there is a range of niches to choose from. At Krazy Fish, we know a thing or two about blogging, SEO, and digital marketing, but we also know plenty about cooking and eating. So we put our three favorite things together – digital content, tech, and food – and created a blog.
Our goal was to pick something simple that almost anyone could do. In that vein, we will be employing industry standard marketing practices, SEO techniques, and a selection of AI and analytics tools that are either free or inexpensive and easy to learn. In short, while we’ll be applying generative AI tools, we’ll be applying all of the best practices for social media marketing, content creation, and business development we’ve learned over more than a decade in business.
This is not the only AI-generated blog in existence and certainly not the first. In 2020, UC Berkeley student Liam Porr used ChatGPT-3 to generate content for a blog under an assumed name. Within hours, the first blog post he posted topped the charts on Hacker News and the post went viral. Porr claimed that he wanted to prove that an AI-generated blog or content in general could easily be mistaken for human writing or even fact. Having proven his point, he eventually shut down the blog with one final post that he wrote himself, “This is what I’d do if I had no ethics.” But more about the ethics of it all a little later.
The truth is that the magic doesn’t begin and end with AI. We aim to go several steps farther than Parr did – and we don’t blame him that he didn’t – by applying not only professional but ethical practices to AI-generated content, and all of it in an attempt to discern how AI can help small businesses and microbusinesses in improving, expanding, growing, and even building their businesses through content marketing, digital ads, and social media.
We don’t just ask ChatGPT to write something and then post whatever it spits out. And yes, we have considered how the human factor will affect our testing of AI tools. The thing is, we are not among those who think that generative AI will replace us. We’re not saying that could never happen, but we’re aware of the limitations of new genAI. We’re nowhere close to Skynet. Yet.
So we take the approach that any website editor and marketing team would, regardless of whether they’re working with AI or freelance authors. First, we figure out the proper prompts and create a blog editorial plan. Then we feed those prompts into the genAI tool. WE frequently also follow up on a generated recipe with questions about how best to prepare or cook an ingredient. Or we ask for modifications to a recipe based on particular dietary requirements. Once we have a selection of recipes, we turn our attention towards SEO and, later, content marketing. Using tools like Google Trends and Ubersuggest, we identify current search trends and popular keywords.
Even our social media marketing strategy owes its inception to AI tools. By generating a primary blueprint using genAI tools and pairing that with our knowledge as professionals in the field, we have been able to build a comprehensive, effective strategy that aligns with industry standards and ethical norms. The human element is not forgotten, though, as we continue to painstakingly refine and tweak the strategy to suit our unique goals and the brand we are building. In all of this, no one at Krazy Fish has lost their job or been demoted. And we’re not letting go of our network of freelancers either. If anything, there’s more work than ever for us with this and other projects. But the skills we need to do our job and do it well are changing. And here we are, learning.
With the AI-generated marketing strategy in hand, we create new prompts for Jasper.ai, another innovative AI tool, and Canva, a design platform. These are used to generate the images for the blog and engaging social media posts, and digital advertising content that not only showcases our recipes but also adheres to SEO best practices. This ensures our content is both appealing and easily discoverable by internet users.
I’ve always said that I’ll truly start believing in the interwebs when I can download my food online. And I now genuinely believe that will happen in my lifetime. Not just with generative AI, but with other emerging technologies in 3D printing and the food industry, I believe the day when I will be able to order a bucket of wings from KFC and have it “delivered” via a 3D food printer in my kitchen is years, not decades away.
With this experimental blog, it’s not all about machines, algorithms, analytics, and data. Our team is also on the front line, experimenting with the AI-generated recipes in our kitchens. This is not just quality control. We are loving this process and grabbing at the opportunity to learn, discover, and appreciate new food cultures, ingredients, and flavors. Our hands-on approach also ensures that our content is grounded in reality, allowing us to vouch for every recipe’s feasibility and taste.
Our journey into the realm of generative AI has been illuminating, palpably – or palatably, if you will – demonstrating its potential in content creation and digital marketing. The blend of AI capabilities and human expertise has led to results that are as (ful)filling as they are successful. We’re not just using AI. We’re learning from it as it learns from us, tweaking it, and even having fun with it, as we navigate this thrilling new intersection of technology and culinary adventure. One that we believe is just a minor one on the way to tectonic technological changes to come.
The convergence of AI with digital marketing and communication is not only feasible – it’s a burgeoning field busting at the seams with untapped potential. When complemented by human creativity and expertise, generative AI can create content that’s engaging, SEO-friendly, and most importantly, saves small businesses time and money, and connects with its audience on a personal level. Just don’t go thinking it’s going to do your job for you.
Method to the Madness
In the brief course of our little AI-driven, flavor-packed adventure so far, we’ve learned that the best way to harness the potential of generative AI for business is by combining it with tried-and-true marketing methodologies. Take, for instance, A/B testing – a cornerstone of any marketing campaign and any new website. We’ll be using this approach to, for example, directly compare the performance of AI-generated content and human-generated content by our freelance writing crew. To ensure fair comparison, both AI and the human freelancer are given the exact same prompts – or what we used to call a ‘brief’ in the industry.
The work product of both is then analyzed and compared not just on its creativity, but also on how well it performs in real-world scenarios – in terms of reach, engagement, click-through rates, and other industry standard metrics. Though not quite scientific, this method and analysis help us to fine-tune our usage of AI tools and better understand their strengths and limitations.
We will also continue experimenting with A/B testing in different content formats, ads, and anything else we think of along the way. Whether it’s long-form articles versus short blog posts, TikTok versus YouTube videos, or text-based recipes versus step-by-step photo guides, our goal is to understand not only what kind of content works best, but where generative AI is most useful.
We realize we’re covering just a small corner of the vast fields that are digital marketing and SEO. And those fields have various techniques, practices, and methods. It would be impossible to test all of them on one blog. Armed with the theory that all of that can be effectively streamlined and augmented by generative AI tools, we’re getting started with this one website for now.
Our goal is not to look for ways to replace traditional marketing techniques or marketing professionals with AI, but rather to integrate AI tools into these techniques and to make marketing professionals more efficient. This integration should, hopefully, enhance our skills, precision, and the effectiveness of our strategies.
Exploring Generative AI for Business and Marketing
Just weeks after we conceived our little AI-generated recipe blog, we watched Google Ads VP Jerry Dischler introduce Google’s new approach to AI-powered advertising at Google Marketing Live. We’d just put this thing together and Google hands us another new toy to distract us. Not that we didn’t know it was coming. Google announced a couple of months before the Google Marketing Live event that the company was working on something along those lines. But did they have to go and make it so colorful and shiny?!
One innovation that stands out in this new feature set is a new natural-language conversational feature within Google Ads, designed to streamline campaign creation. In other words, you pick a preferred landing page from your website and Google AI will summarize the page and generate relevant assets for your campaign, including keywords, headlines, descriptions, and images. It’s a bit like having a chat with a very efficient freelancer who has all the data available on Google at their fingertips.
That said, this means your website, landing pages, and any assets therein have to be in tip top shape. Like a dear friend of mine recently said, “In order for AI to take our jobs, clients would have to accurately describe what they want. Our jobs are safe.”
So, if you want your AI-powered ad campaigns to succeed in this brave new world of Google’s generative AI, ensure your website content is top-notch, well-structured, and properly optimized. our content is not just part of your SEO strategy, it’s now also the raw material for your AI-powered advertising. We haven’t tried out this new toy from Google yet, but we’re going to guess that it’s going to render results that are as good as the materials you provide it with.
Possibly one of the most exciting and least explored, as of yet, aspects of our journey with generative AI in the digital marketing landscape are monetization opportunities. Ethics and social change play a big part in our decision-making here too. Can individuals or small businesses in remote areas of the country or the world create revenue opportunities or extend into bigger markets using their skills, their laptop, and generative AI tools? Can they do so in a transparent, legal, and fair way? Can one person running an English-language blog in an underdeveloped country feed a whole family from that blog?
Once we’re deep enough on the web with this website, in terms of traction, processes, and content, we plan to delve even further into the wonderful world of web business and test various avenues for revenue generation with our AI-generated content. We’ll start out with the most basic monetization methods, like affiliate marketing and ad share revenue. But we’re also keen to explore the possibilities of native advertising revenue, product sales or dropshipping, and possibly As we venture into these opportunities, we aim to strike a balance between generating revenue and maintaining the integrity and quality of our work, and exploring new opportunities for small businesses from all industries and their counterparts in the marketing industry.
Effectiveness Meets Ethics
I’ve been known to make some bold, bombastic statements in the past. Well, here’s another one for you – AI is like nuclear power. It is neither good nor bad, in and of itself. It all depends on how we harness its power and how we use it.
Are some generative AI tools unethical? We believe so, yes. Is all genAI unethical? Of course not. Is it unethical to use AI for business growth or customer acquisition? It depends. As the use of generative AI tools continues to grow, it’s crucial – at least to our team here at Krazy Fish Media – to consider the ethical implications not only of their application, but also in how they were built and the information that they are allowed to access.
Here are just some of the key factors that make a difference in our decision-making and selection of which genAI tools to use:
- Data Privacy and Security: AI systems are usually trained on vast amounts of data, much of which could be personal or sensitive. It’s important to ensure that any AI tool we use respects user privacy and has robust data security measures in place. ChatGPT, for example and according to OpenAI, does not have access to the web, although they do have a beta test feature in collaboration with Bing in which they have access to Bing.
- Bias and Fairness: AI models, including generative ones, can inadvertently perpetuate and amplify biases present in the data they were trained on. We try to choose AI tools that have been designed with measures to minimize bias and promote fairness, or at least be aware of potential bias in order to make appropriate adjustments. This includes fairness toward artists, authors, researchers and others whose intellectual property might be infringed. In other words, no matter how good they might be at generating content, we refuse to use tools that are allowed to explicitly imitate someone’s artistic style or work.
- Transparency: That previous factor is closely related to this one. It’s important for us to understand how the genAI tool we are going to use makes its ‘decisions’. Choosing AI tools that offer transparency in their development and decision-making processes helps to reassure us.
- Preventing Misuse: This is a no-brainer and something we think about when selecting any tools or software we use, not just generative AI. AI tools, particularly those capable of creating human-like text or video, have massive potential for misuse, such as generating deepfakes or misleading information. Though our team will tend to blame the human prompting the tool instead of blaming the tool itself, ultimately, those tools should and must have robust misuse detection and prevention mechanisms in place.
- Environmental Impact: AI models, especially the big ones, require significant computational resources and great energy resources to train and maintain, which can have a substantial environmental impact. In 2019, researchers estimated that creating ChatGPT-3 consumed 1,287 megawatt hours of electricity and generated 552 tons of carbon dioxide, which is what 123 gasoline-powered passenger vehicles will produce over a year. Oof. Obviously, many of these companies are finding greener, leaner servers and energy sources to mitigate this issue but reducing that carbon footprint is on us users as much as it is on the companies developing those tools, if not more. While considering the energy efficiency of the AI tools we choose is important, it’s also important to consider how using these tools will affect our own carbon footprints. Given the choice to either rent an office and work in person with a team or freelancers to put together and publish a print magazine versus working with a remote team and using ChatGPT and digital publishing to make and sell that same magazine – the environmentally friendly choice is clear. Provided you’ve picked reasonably energy efficient and environmentally friendly tools to do it all with.
- Regulation and Compliance: First, ensure that the AI tools you choose to work with adhere to the regulations and laws of the regions in which they are used, including those related to data handling and privacy. Then, check those regulations and laws and make sure they line up with your ethical standards too. Just because you can, doesn’t always mean you should.
For all of us at Krazy Fish Media, ethical considerations are integral in the selection and use of any tools we employ. We continue to explore and review those ethics, always guided by the principle to do no harm. Or, in the words of the Russell-Einstein Manifesto, “Remember your humanity and forget the rest.” Especially when it comes to AI tools.