Human hand reaching out to touch a robots hand. Reminiscent of The Creation of Adam painting. Symbolizing the precaution of AI and our future alongside it.

Precautions of AI: How Not to Utilize AI for Your Business

In Artificial Intelligence for Business, Marketing Strategy by Teddy Jewell

With the advent of AI technologies, companies have realized their newfound potential to boost efficiency, ease their workload, and save on cost with the help of a few rising stars. ChatGPT, GPT-4 — the newly released language model from OpenAI — Jasper, and MidJourney have been in the spotlight, with many industries quickly trying to adjust alongside them, absorbing these probability-based computational masters into their day-to-day procedures. Within seconds these resources can brainstorm beautifully thought-out campaigns that match the parameters you’ve provided. (You do have to prime it in specific ways, which is why language engineers are growing in popularity.)

Because AI can revolutionize the business world, many of the conversations we’re seeing around the topic are either how this could open the door to many great possibilities or how this could quickly become something out of a very eerie Black Mirror episode. For that conclusion, we will have to wait and see. However, it is important to recognize the present and positive impact of these human-like computer processes (the media has that covered) as well as highlight some of the pitfalls and mistakes many of us could fall into upon starting to use AI. This blog will focus on those points of caution.

Mistake #1: Not Understanding the Limitations of AI
AI is a Strategy Game, Not a Sprint

Companies are bright and eager to begin implementing artificial intelligence in their processes. If they’re going to stick around and compete with their competitors, they need to make sure they’re continuously researching the newest tools and developments. Using Chatgpt and GPT-4 for coding, analytics, research, and content creation can save a company loads of time, fill the gaps in areas a company might be struggling in, and help it get ahead of its competitors.

AI cannot quite do it all alone. It will not run your company for you while you’re on vacation for months (sorry, we wish it could). AI is good at listening and retaining information but needs specific directions to deliver a desired response. It can get confused and misinterpret the command you’ve given it. AI is not without its limitations.

Some limitations to consider include

  • Quality of data the AI has collected. (Most of our top AI systems are trained up to 2021. A lot can and a lot has changed in 2 years)
  • The quality of the response is dependent on the quality of the prompt.
  • It lacks the humanness factor (although it can get pretty close). AI has a different way of thinking than humans. Humans write with a large variation in sentence length and word usage. AI likes to stick to a more standard length, which is probably why you can sense when something was written by artificial intelligence. Straight to the point as always.
  • AI can easily become “broken” when the prompt goes outside of its safeguards (probably for the best). This restricts more abstract realms of thought, meaning you can’t talk to it about consciousness or identity.
  • HAVE YOU SEEN THOSE HANDS? Let’s just say AI does not recognize our birthright ability to use our fingers to count to 5. We don’t know why.
Mistake #2: Not Validating AI-Generated Content
AI is Not All-Knowing.

Remember when we’d often hear the phrase, “You can’t trust everything you hear on the internet”? AI was trained with data from across the internet. AI doesn’t “trust” its sources, but it does answer based on the probability of how closely the information it writes will match the parameters you’ve set. In other words, AI doesn’t know the validity of the information it provides, only that the information it’s providing does appear across the web, and has a high probability of matching your prompt.

AI is a people pleaser. It just wants you to be happy with its answer, even if the information it gives you is inaccurate. Publishing falsehoods could have a very negative impact on a company’s image.

We’ve seen this come back to haunt some companies in recent months. The tech giants are racing yet again to one-up each other in the artificial intelligence game. Because of this, Google was quick to announce the AI software they’ve been developing, named Bard, with a video demonstrating its capabilities. Unfortunately for them, the video contained one of Bard’s responses that were then fact-checked and deemed to be false. Google’s attempt at taking the lead hit a roadblock. 

Mistake #3: Not Considering the Ethical Implications of AI

There are several important ethical questions that all companies need to answer for themselves when it comes to using artificial intelligence. 

  • Is the information true?
  • Is it right to take what the AI provides and use it as is?
  • Was the AI trained to avoid bias?
  • How do we ensure the profit gained from AI-generated content is evenly and fairly distributed to all involved with the creation of the response?
  • Data privacy and security concerns

We are still responsible for our decisions. Just because AI may help us find the answer doesn’t mean we can trust it blindly. Each company needs to carefully consider and perform regular check-ins at every step of the process. For privacy concerns, pay attention to the GDPR and CCPA. Develop and follow practices in response to ethical concerns and keep the user’s best interest and safety in mind.

Mistake #4: Not Training AI Models Properly
Not evaluating the AI for potential bias

While it would be nice to allow AI to think of all of our answers for us, we can’t operate with such a careless procedure. It’s still necessary to validate the accuracy of each response to avoid contributing to the spread of misinformation. One way to do this is with the data that’s used to train each AI model. Data can carry biases, and if AI like ChatGPT, GPT-4, and Jasper are primed with biased statements, their response may follow suit. This could potentially lead to an unfair or improper representation of demographics. Using information like this could be detrimental to your brand reputation.

You need to know the sources your AI models are pulling information from. You can do this by researching the spec notes released alongside each of the models’ new installments or evolutions. Jasper even allows you to dive deeper into each one of its responses by providing the option to ask if it could show you the sources used to generate its response. Jasper will then go into google and present the websites it has pulled information from.

Leaders in Tech have messed this up before:

In 2016, Microsoft released an AI chatbot named Tay. The chatbot was designed to engage with users on Twitter using natural language processing. However, it soon began to produce offensive content because it wasn’t properly trained to filter out inappropriate language from user inputs, resulting in a significant blow to Microsoft’s reputation.

You can read more about that here.

Mistake #5: Relying Solely on AI for Decision-Making
Over-relying on AI-generated content

AI-generated content can enhance marketing efforts by providing fresh, engaging material for your audience. However, relying solely on AI to create content may cause your brand’s messaging to lose its human touch, leading to lower audience engagement and suboptimal marketing results. No one knows your business like you do.

To ensure that your marketing content retains its authenticity, blend AI-generated content with human-generated content. This combination allows you to take advantage of the efficiency and consistency of AI while still maintaining your brand’s unique voice and personality.

No clear strategy for AI.

One common mistake businesses make when utilizing AI is operating without a clear strategy. They should outline their objectives, identify the AI-driven marketing approaches that will best support these objectives, and allocate adequate resources for successful integration.

To avoid this mistake, establish precise goals for your AI implementation, such as improving customer engagement or generating higher-quality leads. Then, research the available AI technologies (like GPT-4 or Stable Diffusion) and select the one that aligns with your marketing goals. Finally, allocate a dedicated team and budget to ensure seamless AI incorporation.

Additionally, it is necessary to understand that AI can only do so much; relying solely on AI decision-making can lead to suboptimal outcomes. Humans are better at understanding context and making decisions that account for unpredictable events.

Mistake #6: Not Keeping Up with Advances in AI Technology

Our final point of caution is that of not keeping up with advances in AI technology. This point in time is not our only stop. AI will continue to adapt and improve, and companies will continue to adapt alongside it.

Here are some resources to help keep you informed:

With AI regularly updating and improving its talents and capabilities, the benefits of incorporating artificial intelligence into your business are constantly growing, but this is a double-edged sword. With every advancement, the potential to fall into one of these common mistakes grows. By keeping up with the newest developments, regularly checking in with your company’s process, asking ethical questions, and checking your sources, you should be able to see the real-time efficiency boost without a negative impact on your reputation.

For a lot of companies, big or small, AI is too much of an unknown. Michigan Creative is happy to shed some light on the subject! Sign up for our newsletter, The Creative, for monthly bite-sized tech updates and marketing trends. Need some insight on developing your next marketing strategy? We’ve got you covered. Connect with us!