In today's fast-paced digital world, Artificial Intelligence (AI) has quickly progressed from being a futuristic (often dangerous) concept used in Sci-Fi Hollywood movies, to a common-place business tool, transforming day to day operations for organisations of all shapes and sizes.
Understanding the basics of AI and tools like ChatGPT could unlock new opportunities for efficiency and innovation. Gone are the days when AI was just a buzzword; it's now a practical asset that can help drive your business forward and set you apart in a competitive landscape.
1. So what is AI?
There are many different definitions of Artificial Intelligence, but we’ll go with what Wikipedia says:
Artificial Intelligence (AI) is intelligence demonstrated by machines, as opposed to intelligence displayed by animals and humans.
https://en.wikipedia.org/wiki/Artificial_intelligence
But let’s be clear: what we’re going to discuss in this article is Generative Artificial Intelligence – a subset of AI which burst onto the scene (and into everyone’s social feeds) over a year ago with the public launch of ChatGPT (more on that later). So back to Wikipedia:
Generative Artificial Intelligence (Gen-AI) is Artificial Intelligence capable of generating text, images, or other media, using generative models. Gen-AI models learn the patterns and structure of their input training data and then generate new data that has similar characteristics.
https://en.wikipedia.org/wiki/Generative_artificial_intelligence
So to simplify it… We’re talking about using AI to generate (create) an output based on a request (typically called a prompt). That output could be anything – data, insight, suggestions, creative marketing ideas, music, literature or art, and it generates that output from analysing and learning from ‘big data’ which already exists.
AI knows what a cat looks like, so if we ask it to create another picture of a cat, it typically does a good job. It knows the typical style and writing structure of Rabbie Burns, so when we ask it to generate a poem (an ode to a cat?) in a similar style, it’ll likely bear an uncanny resemblance. The AI learns things, and can therefore generate new things based on that learning.
It’s similar to how your mobile phone predicts the next word for you while typing... it’s not guessing that word, but rather predicting what it might be based on what you typically type next, and what others across the world typically type next... it has learned language patterns because it’s seen a lot of example from peoples texts, WhatsApp messages, and more, and it’s using logic to suggest your next word. The current AI tools essentially work the same way, but can do this at scale, predicting not just the next word, but whole paragraphs, chapters and books! There’s a great article here explaining this in a much more scientific way.
2. How did it learn?
So how did AI learn things? Because the creators of the various AI systems fed it lots of data. OpenAI (the company behind ChatGPT) initially trained its AI on vast diverse text-based datasets, some of which was publicly available, some of which was just ‘scraped’ from the live internet, and some which was taken from sources which are now pretty angry they used it without asking for permission (the New York Times being a good example).
Some experts predict that over 45 terabytes of text data was used to train ChatGPT when it first launched. That might not actually sound like a lot – the computer I’m typing this on right now has a 20 terabyte drive (although I’m only using a tiny fraction of it), but given the data was just plain text (no big images or videos etc), that’s a seriously large volume of content that it has to reference.
Other “image based” AI tools like MidJourney were trained on hundreds of millions of images allowing them to understand what things look like, recognise patterns, and then later recreate them (or things very similar to them). Many artists are angry at such tools because their content was used to train the AI algorithms without permission, and now anyone can generate artwork in their style without them ever seeing a royalty, or receiving a thank you.
3. AI for business marketing
There are so many possible use cases for small and medium businesses when it comes to AI, but this article will focus purely on Generative AI, and we’ll concentrate on business Marketing as a great first example.
Websites
It wasn’t that long ago that businesses had to work with web design companies, or maybe know some HTML themselves in order to get a nice shiny website up and running. Now with the growth of AI, there are affordable tools that can build websites in seconds.
Wix.com for example asks you a few basic questions about your business, and with the click of a button, a ready-made website, suitable for your type of organisation is live, pre-populated with content that while you’ll need to change and improve, is a pretty good starting point. 10Web is another similar AI powered web building tool that promises a website in minutes, just by answering a few questions, and then tweaking the output to suit.
And just in case you are still a fan of hand-coding HTML, CSS, or JavaScript and building sites the old fashioned way, ChatGPT is well versed in all web coding languages, so it can help you out big time there too!
Another tool to check out for creating websites (as well as Presentations, documents and more) is Gamma. Just tell it what you’re looking for – “A one-page website promoting my dog walking business” – and right in front of your eyes you’ll see it built in seconds, with relevant “about us” and “our dog walking services” sections, containing decent text and images to get you started. These AI tools know what a typical dog walking website, or a plumbing website, or a personal training website look like. So they know how to replicate something similar for you.
Images
Every business needs images. It’s arguably impossible to ‘do good marketing’ without strong images and visuals on your website, social media channels, email newsletters and more. Until recently, business owners could take their own pictures, hire professional photographers, buy stock photography, or try and create graphics in Photoshop or Canva.
Now thanks to the huge growth and rapid development of Generative-AI tools for creating images, there’s another option – using systems like Midjourney, Adobe Firefly, or Stable Diffusion, and simply “generating” any image you might need, simply by asking for it.
Many of these tools can be initially used for free (operating freemium models), and some are better than others for creating certain types and styles of images (see this blog comparing outputs), but they all essentially do the same thing – “Text to Image” – generating images from a description (or prompt).
Prompt Firefly with this - “advertisement food photography splash of lemon juice on salmon steak, bright colours studio lighting” and you’ll be given a few images to pick from, worthy of inclusion on a Michelin star restaurant’s website, at such a high level of photo-realistic detail you’ll find it hard to believe a professional photographer didn’t spend hours on the perfect shot. They didn’t – the AI created it. It hasn’t just gone and searched the web for images to return back to you, it’s generated these images from scratch – they don’t exist elsewhere.
Not every image will be perfect, and there’s a bit of an art/science in crafting the perfect prompt, but some trial and error will soon have you working out what works. You could ask for certain styles (photorealistic or cartoon sketch?), or even provide reference images where the system will match the style from your own existing photos.
You could even start with your own images, and use AI “in-painting” to modify parts of the scene, removing unwanted items (like that annoying tree in the background which looks like it’s growing out your head) or adding in some AI enhancements like changing the background for your perfect product shot, or changing the sky from dark clouds to blue skies. Image manipulation has never been easier, or cheaper with many of these tools free to use, or operating a ‘freemium’ model (free to test out and play with, moving to a small affordable monthly subscription once you decide to actually make use of them commercially).
Videos
It wasn’t that long ago that most businesses had to hire professionals when it came to making video content. Experts with fancy video kit, lighting rigs, professional microphones and more. There’s still a place for that of course, but thanks to the advancement of AI, videos can now be created quicker than ever, without the need for expensive kit or expertise.
Just like some of the image generation tools, videos can also now be created just by describing what you need. Platforms such as Runway and Pika allow short video clips to be created from your text based prompts, or by uploading a single still image as the starting point for your video sequence. As Runway says on it’s website – “If you can imagine it, you can generate it.”
These tools can also be used to modify existing videos that might be too costly (or challenging) to re-shoot, with the ability to change anything in the frame, such as the presenters clothing, or the background setting. The results are often not perfect, and some things work better than others, but compared to traditional video editing techniques, anyone can pick up these skills quickly and easily.
OpenAI’s latest product Sora is capable of creating videos up to a minute long, while maintaining a high visual quality and sticking fairly accurately to the user’s original prompt. While its early days for this whole medium, the results are impressive (even with their flaws) and creative video storytelling is taking an exciting turn.
Other affordable AI based tools to help generate video include D-ID and HeyGen, both of which are particularly effective at creating ‘avatar based’ videos. Simply upload an image of a person (real or fictional), give them a voice, personality, and something to say, and you can create video content in seconds, to be used on websites, social media, online learning materials, or just for fun. These types of videos will never replace actual humans, speaking directly to camera, but they’re much quicker to make compared to doing it “for real”, and if you’re camera shy, then you can let your AI based digital twin do the talking from now on!
HeyGen in particular is capable of mimicking real people and their mannerisms pretty effectively. You can train the AI model with actual video footage of yourself, and it learns how you look, sound, and act on camera. It’s also got some pretty amazing language translation tools to play with, so now you can create welcome videos to host on your website, where you can fluently speak French, Japanese, or Hindi to your non-English speaking visitors.
While these tools provide some clever marketing opportunities, there is of course a downside to them, with some unethical users creating ‘deep-fakes’ of real people without permission. Only create AI versions of people with their consent to do so, and if using them online, best practice is to make clear to your audience that they’re not real (they’ll probably have already guessed!), and not to hide the fact that it’s an AI generation. Or just have fun and create videos of talking cartoons, or perhaps get your cat to chat to people on your social channels.
Written text
Finally, when it comes to AI generated marketing content, we can of course use AI to help create written text. The most obvious tool to consider here is ChatGPT from OpenAI – it has rarely been out the media spotlight since it launched back in November 2022, but there are plenty of other similar tools you should have a play with too, Google’s Gemini being one of the main contenders, but also check out Perplexity, and Claude which are gaining serious momentum in this space.
All these tools are known as what’s called “LLMs” (Large Language Models), and as explained earlier, they’re able to generate natural language having been trained on huge volumes of written data, and you use them by simply chatting back and forward to them.
Ask ChatGPT to write you a poem for your friend’s 30th birthday, and it’ll create a fun rhyme you can copy and paste into their birthday card, but dig a little deeper, and you’ll realise these tools can do so much more!
5 ChatGPT Marketing Use Cases to consider
1.Your AI personal assistant
Ask ChatGPT to help you, just as your personal assistant would. Get it to answer questions that you need to know, provide facts and statistics that will help finishing off that report, explain things you don’t understand, and carry out research that you need done.
Be careful though, the free version of ChatGPT currently has a knowledge cut-off of January 2022, so it doesn’t know anything beyond that date. The paid version (starting at $20/month) has a later cut-off date of April 2023, but it’s also capable of searching the internet (using Microsoft Bing) to augment its existing built-in knowledge. Watch out also for ‘hallucinations’ – something that all the big LLMs do, occasionally just making stuff up which sounds highly plausible, but is in fact just a lie.
One particular task that ChatGPT can do effectively is summarisation. Give it a big piece of text (or point the paid version to an existing webpage) and ask for a summary, maybe requesting the top 5 take-aways. It’ll usually do a great job condensing it down for you, helping you quickly understand the bigger piece without having to actually read it all.
2. Your AI business adviser
Not everyone has an accountant to reach out to for on-demand financial advice. Likewise not everyone has a marketing mentor, or a PR guru on speed dial. As much as the Business Gateway team try and support businesses 24/7 we obviously can’t be there all the time, and that’s where ChatGPT could step in.
Need help understanding the pros and cons of becoming VAT registered? Just explain your situation, and ask ChatGPT for advice and recommendations. Looking for ideas for a new company name, marketing strapline, or even the whole business idea itself? Have a brainstorm with your AI adviser, and bounce around some ideas that might spark something that works for you. On the whole, the advice given will be accurate (although for certain things you might want to 100% double check), decent, and depending on what (and how) you’ve asked, you might be pleasantly surprised at how creative this robot can be!
3. Marketing Content
One of the most common ways that small business owners are using ChatGPT is to assist them with their marketing and social media content. From helping create short social media biographies (ask ChatGPT to improve on what you’ve already written in your Instagram Bio, or get it to re-write the text for your LinkedIn ‘headline text’), to comprehensive blogs for your website, these AI Large Language Models are pretty effective writers when it comes to marketing content.
ChatGPT’s ability to re-write text (for different audiences, or more suited to a different platforms) is one area that might save you a lot of time. A LinkedIn post can be quickly and easily re-purposed for Facebook (re-written is a more relaxed, fun style) just by asking. Likewise, your 500 word blog could easily be turned into a series of 280 character posts for X (formerly known as Twitter!) giving you weeks’ worth of valuable content.
Struggling to say something relevant on #SmallBuisnessSaturday or any of the other popular ‘awareness days’? Just ask ChatGPT for example content relating your business, product or service to this theme. It might not be perfect, but it’ll give you some ideas for fun engaging social media content and stop the dreaded “blank page syndrome”!
4. Ideation and Inspiration
What about when you really don’t know what to say? When you’re under pressure to add something to Facebook, or write a new business blog but you’ve got writers block and are woefully lacking in creative ideas.
ChatGPT’s ability to help you brainstorm ideas for inspiration is impressive. Think of it as your creative friend always available to give suggestions, and to bounce around ideas with. Give it some detail about your business, what you do, and where you’re based, and ask it for creative ideas you can use across all your social media channels. If you’ve got the beginnings of an idea yourself, explain it (as best you can) to ChatGPT and ask it to critique the idea, giving additional suggestions on how to improve it.
The trick here is to converse with ChatGPT – lots of back and forward discussion, digging deeper into ideas, and exploring them more fully from multiple angles. In the same way you might brainstorm things with a group of business partners, the AI can play those additional roles and personalities for you.
5. Customer Insights
Remember that marketing is more than just the promotional content posted on your digital channels. Marketing is about truly understanding the target market, and giving them value, and ChatGPT can help with this too.
Ask it to help identify some customer types for your business, then pick one and request a more detailed “customer persona”, highlighting their demographics, goals, challenges and values. Dig deeper by asking what their typical day looks like, and how/why they might have a need for your product or service. Finish by getting ChatGPT to explain the likely reasons this customer wouldn’t buy from you (commonly called the ‘objection to the sale’) and brainstorm ideas how to convince them otherwise.
By spending time climbing inside the mindset of customers, you’ll be better equipped to consider great marketing strategies and tactics which really resonate with them, providing solutions to their problems, and becoming far more customer-centric in your approach.
Remember that the examples above are not exclusively just for ChatGPT. Other LLMs are available, and you should try and test some of the other platforms like Google’s Gemini or Anthropic’s Claude to see which AI system works best for you.
3 Simple ChatGPT Top Tips
Here are some simple tips to ensure you get the most out of ChatGPT (or any of the other LLMs that you might choose to use):
1. Use it as an ‘input’ not an ‘output’. Don’t just copy and paste what it gives you and use as your finished output. Use it as your input, your inspiration, and change it. Add your own personality, style, anecdotes and examples. Make it better. Make it your own.
2. Give it as much detail as you can. Don’t get lazy with your prompts. You’ve maybe heard of the old programming term “GIGO” – garbage in, garbage out. In order for ChatGPT to create relevant replies to your prompts, it needs to be given detailed input. Use the “Customize ChatGPT” setting to give it custom instructions, telling it more detail about you, your business, your goals, your customers etc. You can also give it a steer on how you want it to respond with regard tone of voice, style and personality. Try and configure this so the default output sounds a little more like you!
3. Tell it how to behave. If you really want expert considered answers, tell ChatGPT to act like the role you require. “Act like a journalist… and write a press release suitable for UK tabloids to raise awareness of our new charity initiative…” or for something a little more creative – “Act like a rock musician.. and write a song influenced by the style of Bon Jovi, about rebelling against the AI, and include suggested guitar chords”. Defining the role you want ChatGPT to play helps it get into character, while also ensuring the response come from that perspective, limited to that expert’s knowledge and thinking.
To further assist with prompting, you might want to consider the “Triple R Prompt framework” – which will help you write more detailed requests in a structured way.
4. Beyond marketing
While this article has focussed on marketing as the obvious business opportunity from Generative AI and ChatGPT especially, it’s worth remembering that these tools can do so much more. From helping create internal company documents, policies, and strategies, to fast-tracking web design briefs, funding applications, or even spreadsheet data analysis and reporting.
Don’t also forget that LLMs are great with all languages – not just the English language. So they’re handy tools to assist with translation work that might need to be done, or even writing basic HTML webpages (they know most coding languages too!).
5. Copyright
And as for the obvious question – “what about copyright?” – well it’s a bit of a grey area. In the UK, copyright law requires a human creator to be behind copyrighted works. The Copyright, Designs and Patents Act (CDPA) outlines that a work created by a computer alone is not eligible for copyright protection. Knowing that, there’s the argument that no one has copyright over AI created assets. The CDPA also says however, that the person behind the tool used for creation could be considered the author, so there’s an argument that copyright could be granted to the developer of the AI tools.
What’s certain is that the law hasn’t kept up with the pace of the technology itself, and while it’s unlikely that using these AI tools will ever land you in legal hot water, our advice is to do your research, read the terms and conditions and understand the usage rights that the various tools will enforce on you. Midjourney for example confirms that while you will own all the visual assets you create through the tool, you also grant them full rights to use those images in any way they want.
OpenAI (the company behind ChatGPT) says that you own both the ‘input’ (the prompts you enter) as well as the ‘output’ generated, but warns that either could be used to help train their AI models and improve their services.
6. Other ethical concerns
Above and beyond copyright concerns, many businesses are wary of utilising AI due to other ethical concerns including:
- Bias - content produced from Gen-AI tools can often display bias and a lacking in diversity.
- Privacy - fearing that private data entered into these systems will become public data, even more serious if that information is personal data, and ensuring compliance with GDPR.
- Transparency (or lack of it) – where did the AI system get its information from? Which artist’s style has been replicated (without any recompense)? Will using these AI tools detrimentally harm hard working creatives?
- Environmental impact – these new AI systems require huge computing resources, server farms, and electricity to power them.
- Job displacement – perhaps the biggest fear around the usage of AI right now – “will it take my job?”. As these systems become ever more powerful, and their ability to create content begins to rival that created by humans, will these creative jobs actually be needed going forward? In a UK Government report, they predict that in 5 years, 7% of all jobs could be redundant thanks to AI automation taking over. That figure rises to 30% over the next 20 years. They do however point out that many new jobs and roles will be created thanks to the growth of AI, alongside new and emerging related technologies.
It’s no surprise that many marketing industries play down the impact from AI, often publicly mocking its quality, or highlight its flaws. “It’s nowhere near as good as what the professionals can do” they shout, and guess what - they’re 100% right!
A trained and qualified PR consultant with 10 years’ experience should always be able to write you a better press release than ChatGPT. A web designer who’s been building small business websites for 20 years and understands exactly what the market needs should always be able to create a better website than Gamma. And a Photoshop guru who’s spent years learning their craft, able to use all the advanced tools and techniques, will be able to give Midjourney a run for its money – especially once we add in the “human element” – that extra edge because they truly do have intelligence which is real and not artificial.
But small business owners don’t always have the expert skills to do the job themselves, or the time to learn how to do it properly, or the money to hire in the best consultants and suppliers. Add into that mix the constant pressure on small business owners to create content across their social media platforms, while at the same time juggling many other tasks, and not knowing what to say, or have the time to plan future marketing campaigns. That’s where using AI can help. It might not be the best solution, but under the right circumstances, it’s often the right solution.
That all said – remember that ChatGPT and other Gen-AI tools are just that – tools in the toolbox. Don’t rely on them to run your business on auto-pilot. Don’t trust them to always do the right thing by you. Don’t get lazy and think you’ll never need to get creative yourself. Use them when it helps, and pick another tool when it’s better for the job in hand.
Learning how to use some of the popular AI tools, and developing skills in AI literacy is a great start in futureproofing your business. As Richard Baldwin, an economist and professor at the Geneva Graduate Institute in Switzerland, said at the 2023 World Economic Forum's Growth Summit:
"AI won't take your job. It's somebody using AI that will take your job."
Book onto one of Business Gateway’s national AI webinars and discover how to take even more advantage of Generative Artificial Intelligence to help your business thrive.
Written by: Gary Ennis, Business Gateway trainer, and Managing Director of Digital Skills training company NSDesign Ltd.
Your local Business Gateway can offer you free 1:1 advice and online support to help you with anything to do with your business or if you are just thinking about starting up then we can support you with that too.
Fill in the form below and we will get back to you to help.