As a writer, I dislike writing but enjoy the outcome. Running a marketing firm where I mostly write intensifies this problem.
If you feel the same, you’ll see the appeal of AI tools like ChatGPT for marketers, be it in-house or at an agency. These tools can simplify tough writing, allowing writers to skip the struggle and go straight to a finished article. It’s like getting rewards without the work.
It was last year when I started using generative AI like ChatGPT and was amazed by its quick results. It seemed like a solution that could change how I work and make me a writing powerhouse. Yet, experimenting with various large language models (LLM)/generative AI tools showed me the dangers. This is important since I work with clients and must offer well-researched, clear, and trustworthy advice.
This article aims to guide and advise skeptical marketers about the AI revolution’s potential pitfalls.
Important Rules for Using AI in Content Creation
Whether you’re creating content using AI tools for personal projects, employers, or clients, there are some fundamental principles to remember.
Protect sensitive information
Never input confidential or sensitive data into the AI model, such as company data, non-public intellectual property (IP), private datasets, business plans, internal reports, or customer details. Some companies like Amazon have restricted the use of tools like GitHub Co-Pilot and ChatGPT due to concerns about potential data leaks.
Additionally, replace real names with fake ones and replace personally identifiable information (PII). Also, Sensitive business data with fictional alternatives when using real data for context.
Ensure you have consent
Always obtain proper permissions before using any client data. Sharing data with an AI model might breach confidentiality agreements and data protection laws. Be cautious and don’t assume consent for sharing information. Seek legal advice when necessary.
Clients and businesses are increasingly aware of the use of generative AI. By openly discussing how you use AI tools, approach consent, and handle data sharing, you can show that you’ve recognized and minimized potential risks.
Review Your Content Carefully
Always go through the content created by AI to make sure no sensitive information is accidentally included. AI models might use the data they’ve been given and generate sensitive content from third parties without meaning to.
Check for Protected Info
Thoroughly examine the outputs to avoid accidentally mentioning private or sensitive details about clients or businesses.
Avoid Copyright Issues
When using tools like Midjourney for visuals or ChatGPT for text, don’t use prompts like “copy the style of X.” This could break copyright laws. Even imitating artists whose work isn’t copyrighted anymore can cause problems. Recent cases show backlash from artists about AI copying their style. Instead, follow client guidelines for the right tone of voice.
Respect Others’ Work
Don’t copy the exact style of specific authors or artists. Always respect copyrights for text, images, and other content.
Be Cautious of AI Outputs: Avoid Blind Trust
Will Moy, CEO of Full Fact, recently warned that AI’s ability to flood public discourse with seemingly credible content is a powerful disinformation tactic. This tactic can cloud discussions and hinder transparency in debates.
In our democratic society, it’s our duty to combat AI-driven disinformation. We all play a part—fact-checkers, authorities, and content creators alike—in nurturing an informed public.
Two Key Issues with AI-Language Models
Firstly, AI models sometimes produce content unrelated to their input or veering off from facts. Secondly, their quality depends on the data they learn from. If the data is flawed, the AI output can mirror that flaw.
Fact-Checking Tech’s Limits
Automated fact-checking has improved, but it still can’t ensure absolute accuracy. No tool can perfectly verify another tool’s output.
Context Matters in Fact-Checking
Context is crucial, and automated systems struggle to fully grasp it. Minor changes in phrasing or timing can change the validity of a claim. Even accurate stats can mislead if causation is confused with correlation.
Harnessing Human Reasoning
To ensure accurate facts in the right context, we need to rely on human reasoning. Let’s use our critical thinking skills to verify and contextualize information.
Double-Check Facts from Reliable Sources
Make sure you’re not just accepting what you read. Always verify facts, figures, and sources from multiple trusted places before you believe them. These can include respected news sources, government records, or academic studies.
Be Wary of AI-Generated Links
Even though AI tools like ChatGPT might offer links, it’s important to confirm their reliability first. Check if the links work, the websites are trustworthy, and the pages are actually relevant and dependable. It’s often better to find your own sources from well-known, reliable sites you’re familiar with.
Trust Fact-Checking Websites
Websites like Full Fact, Snopes, or FactCheck.org are goldmines for checking facts. They dive deep into claims, often mentioning their sources, and help you tell what’s true and what’s not.
Get the Latest Data
Data’s accuracy can change over time. Always check when the data was published or gathered. Go for the latest and most fitting data available. Keep in mind that ChatGPT’s training only goes up until September 2021, so it might not know current details, like where the Queen of England lives now.
Even the Latest AI isn’t Perfect
Even with the newest model like ChatGPT 4, there’s no guarantee that data or accuracy is flawless. Although ChatGPT 4 is better at blending data from different places, OpenAI still admits it might make mistakes, just like previous models.
Still confused about how to deal with unverified information?
When you encounter questionable or unsupported facts, consider leaving them out to uphold your credibility. But if the info matters to your topic and its truth is unclear, let your audience know about the uncertainty. Show other viewpoints too, if possible. To get more clarity, consult experts in that field. They could shed light and clear the confusion. (Remember, E-A-T: expertise, authoritativeness, trustworthiness. Quoting experts adds weight to your content.)
Talking about experts, ensure the ones you quote are reliable. Think of it like Google does – is the person mentioned on other good websites? Do they have relevant qualifications, and are they cited in pro journals? It’s your duty to verify the verifier.
Walking the Tightrope: AI and the Art of Balanced Content Creation
So far, I have shared how to lower risk with AI tools and prevent the spreading of false information. But you might think AI tools like ChatGPT are more hassle than help. With all the effort needed, you could wonder if creating content alone is simpler. This perspective holds some truth.
Yet, as a marketing advisor, I’m using AI differently. It enhances my creativity and efficiency in three methods. Notice, none involve starting from scratch.
At the beginning of creativity, my first ideas lack uniqueness. This writing phase frustrates me; getting into the flow takes time.
A writing tutor once said, the first 30 minutes clear the ordinary ideas for the great ones. That’s why it’s tough. AI, like ChatGPT, is based on existing content. It gives likely results, helping me generate these “less good” ideas swiftly. If ChatGPT suggests it, it’s probably not a new or exciting concept.
Reflecting on My Creativity
I enhance my creativity using AI by reflecting on my own creative work. After writing an article or creating something, I use AI to summarize my main points. This helps me review my work and make sure I haven’t missed anything important. It also ensures my message is clear and consistent.
AI can even spot gaps or inconsistencies in my arguments, similar to getting feedback from someone else. But even after this, I like to have a human editor review my work thoroughly.
AI also helps me come up with different versions of my original content, giving me new ways to present my ideas. I explore different ways to say things or arrange sentences and paragraphs.
This helps me find more interesting and powerful ways to convey my message. Instead of copying everything, I pick the best parts from the AI’s suggestions. Sometimes, it’s just a single word that makes a difference.