Half of small businesses don’t use any AI in their operations, an August 2024 survey of Australian SMEs by Peninsula Group states. One of the main reasons holding them back? Fear of AI and what it could mean for their business.
Survey respondents had a variety of concerns about adopting AI. These included data privacy, reputational impact, risk of law breaking, loss of intellectual property, and impact on work quality and productivity.
But are these fears grounded in reality? The problem with AI is that products and services that use it are growing much faster than relevant education or legislation. In this article, we look at the facts relating to some common fears that small-business owners have about AI.
To help us, we brought in AI coach Leanne Shelton of HumanEdge AI. As an AI coach, Leanne has seen plenty of AI fear from small-business owners in her time. When we sat down with her, she confessed that her first encounter with AI was also fearful. She ran a marketing business based around copywriting when ChatGPT rose to popularity in late 2022. She began to worry that the free software, which could generate flawless text almost instantly, would make her business redundant.
“I thought, ‘Why are people going to invest in me when there is this free tool?’ ” she says. “There’s this new shiny thing that could replace me – so I did have that moment of freak out, going, ‘What does this actually mean?’ ”
What Leanne felt at this point – fear of being replaced by technology – is recognisable to generations of humans. It’s also one of the most fundamental fears that business owners have about AI, and AI-fear is what we will address first.
Fear vs reality
Fear: AI will replace my products and services
Facts: It might.
Technological change will always affect the way we work. Leanne is an example of someone who had to pivot her entire career because of technological advancements.
Countless inventions have disrupted industries throughout history. But it’s also a fact that, time and again, workers have found ways to adapt and thrive. Millions have done it before, and many more will do so in the future.
This has often meant upskilling or adapting existing talents to stay relevant. Though Leanne’s job fell victim to ChatGPT, she’s an example of this adaptation in action. In late 2022, after the launch of ChatGPT, Leanne bought a few books on the software to teach herself how it worked. One, a $57 book that contained 1000 chat prompts, made her realise how she could repurpose her copywriting skills in a new way.
“Businesses were just pumping stuff out [using ChatGPT] without the proper guidance or editing,” she recounts. “I thought, ‘If I take out the marketing point of view, with my experience, they’re going to be damaging their brand.’ I thought, ‘I can teach people how to make this content sound good at the output stage.’ ”
After studying up on Large Language Models (LLMs) like ChatGPT, Leanne started teaching other business owners how to leverage the technology without churning out texts that sound generic and robotic. Her copywriting business was shunted to the side as demand for her AI-training services grew, and she now makes a living solely from her AI coaching.
Moreover, her new business plays into the element that AI can never replace: the human touch. Her coaching service, HumanEdge AI, combines her AI expertise with her deep understanding of human behaviour as a marketer, which gives her business its true value.
“During COVID, we realised how much we yearn for human connections,” Leanne recounts. “We still want to speak to humans. I listened to a talk by [Australian consumer experience expert] Amanda Stevens last week…she’s like, people are going to want more, deeper relationship marketing, like, really going above and beyond.”
We are seeing this play out in current marketing trends. Later in this issue of ISB, we speak to marketer Maddi Ragno about the rise of “founder-generated content”, which is content that features a business’ founder front and centre. As the ubiquity of AI grows, so too will our desire to see real human lives and experiences.
Fear: AI can present security risks because it involves sharing data with third parties
Facts: Sharing data with AI does present security risks, and most experts recommend that you don’t do it. Trust-management platform Vanta recommended anonymising customer data before inputting it into AI, a practice that only a quarter of Australian businesses carry out.
Leanne advises her clients to avoid inputting sensitive information, whether it is their own or others’. She also recommends looking for ways to opt out of training your AI tool of choice. If you use the paid version of ChatGPT, for instance, you can create custom GPTs and opt out of your input being used to train models. There is also a “temporary chat” feature in ChatGPT, Leanne says, which will disappear a chat after you’ve used it, meaning the conversation won’t be used for AI training.
If you are after more guidance on how to use AI responsibly, the Government-backed ARM Hub AI Adopt Centre provides practical guidance specifically targeted at SMEs. Also, the Office of the Australian Information Commissioner (OAIC) has released new guidance to assist organisations and developers in navigating privacy obligations while using AI. Finally, the federal government has developed a Voluntary AI Safety Standard aimed at business. It’s also working on turning this into a set of mandatory AI guardrails, but it’s unknown whether this legislation will apply to small businesses.
Fear: If I use AI, I risk losing my IP, or plagiarising someone else’s
Facts: Yes, there are IP risks when using AI.
“Never trust what [the LLM] has been inbuilt with,” Leanne says.
If you’re using an LLM to write content, she recommends giving it a list of sources to take from and asking it to reference the information it has used. Much of the content on these platforms has been unethically sourced, so if you have ethical qualms – or want to avoid potential litigation in the future – it’s safest to avoid taking content straight from LLMs as much as possible.
Leanne describes how she prompts ChatGPT to do this: “I say: ‘I’d like you to write this article for me, please ONLY’ – I use capital letters for ‘only’ – ‘Use these resources to write the article’. You can even go a step further and say, ‘Please provide a reference or advice when you have sourced from an article, then it’ll pop up’’ And you can always say at the end as well, ‘Just confirm, did you only use those sources?’ ”
Leanne is in contact with an IP lawyer who has moved into the AI space recently. She was told that IP comes down to “how much human goes into it”. If you use ChatGPT as an outline, or for inspiration, you are very unlikely to have plagiarised someone else’s work.
As for your own IP, the same goes for this as with your sensitive data. If you’re worried, don’t give it to your LLM – or opt out of training the AI, as described above.
Fear: I will have to overhaul my existing systems to introduce AI
Facts: “No, you don’t have to, at all,” Leanne says. “I say to all my students and audience members – just start small.”
“Starting small” could mean training up an LLM, like ChatGPT, on your brand’s products and services, and teaching it your brand voice, so it can write copy for you. It could mean using it as a brainstorming tool so you’re not staring at a blank page. It could also create email sequences or templates. There are many ways to leverage AI without overhauling your systems or forking out thousands of dollars on fancy AI-powered products.
Fear: AI will worsen the quality of my work
Facts: This can definitely become true. But if you train your AI and use it well, it shouldn’t, Leanne says.
“We’ve all seen that robotic, generic content flooding our inboxes and social media
feeds,” she says. “However, with proper training, AI-generated content should be unrecognisable. The crummy content we’re currently seeing is the result of ChatGPT (or similar) being misused by unintentionally misguided or misinformed individuals.”
Before you do anything with AI, she says, it’s key to spend time training your tool to thoroughly understand your business’ voice, values, and customers. Leanne recommends treating your AI tool of choice like a new junior assistant or intern who has just joined your team.
“You wouldn’t start barking demands or handing over essential parts of your business to them on day one, would you? No, you’d start with an onboarding,” she pointed out.
Onboarding, in this case, means teaching your LLM everything there is to know about your business, from your products and services to your customer base. You can also feed it articles or posts you’ve written and ask it to learn your brand voice.
There are millions of tweaks you can make to your AI tool so that its output is just the way you want it. For instance, Leanne instructs her ChatGPT not to use certain words.
“We’re using only UK English, and none of those stupid AI words as well,” she says. “I told it to please never use ‘unlock’, ‘unleash’, ‘in today’s digital age’, ‘delve’, or anything like that.”
One last thing: When Leanne demonstrated some of the prompts she uses, I noticed that she often says ‘please’ to her LLM. The AI coach explains that there’s some evidence that AI responds better to polite requests. A study by Waseda University and the RIKEN Center for Advanced Intelligence Project found that polite prompts can produce better responses.
“It’s also for when the robots take over,” Leanne jokes.
The robot takeover is a fear we won’t be exploring today; but hopefully, if we all learn to train our GPTs well, it won’t happen any time soon.
This article first appeared in issue 47 of the Inside Small Business quarterly magazine
