Take the 7-day AI Upskilling Challenge!

Executive summary: I discuss AI's impact on employment and creativity and how it can be a tool for innovation rather than a threat to your job. If you have not immersed yourself in generative AI (ChatGPT, Claude, Gemini, Copilot, etc.), I offer you a unique challenge. Most people who take this challenge share concerns about hallucinations, so I clarify that in this article. Happy reading!

⛈️ Dark clouds over AI and jobs

You have probably seen the headlines, and they can be scary:

International Business Machines Corp (IBM.N), expects to pause hiring for roles as roughly 7,800 jobs could be replaced by Artificial Intelligence (AI) in the coming years, CEO Arvind Krishna told Bloomberg News on Monday. (Reuters)

One in four [CEOs] anticipate reducing headcount by 5% or more in 2024 due to generative AI. (PwC’s 27th Annual Global CEO Survey)

There probably isn't a single C-suite level executive looking to cut headcount, "reduce redundancies," or re-hire more skilled talent due to the AI wave currently crashing down on the world.

That said, it is not all doom and gloom. You can learn AI and perhaps find novel opportunities for use in your job, personal life, or new business venture. I will explore these options in this article. But first, let's talk about the detractors.

Isn't AI just stealing work?

Think about some work you have done recently. How much of that was indeed yours? Read the following and honestly assess yourself against the following statements:

• My PowerPoint includes quotes from visionary leaders, backup data from internal and external sources, or repurposed slides from other people.
• The paper I am writing pulls from my experience, including the shared experiences I have learned from others.
• The picture I painted shares similarities with various artists that inspire me.
• My song includes samples or similar styles from other artists' songs.
• The best practices and standards we follow at work are the same as those in my industry, with a few tweaks unique to our business.
• When I create something, the outcome is a combination of my great ideas and those of my team members.
• I love quoting my favorite comedy acts, people, or fictional characters from literature.
• My email app helped me write a better email.

I'm guessing at least a few of those statements resonate with you. If so, you are [very likely] not stealing anything!

You use your experience to create something new while directly or indirectly drawing inspiration from those who have come before you and your contemporaries. With that knowledge, you create something new that is uniquely your own.

AI, specifically generative AI tools like ChatGPT, Claude, and Gemini, work the same way. They collect information, store it in ways that they can recall, and then wait for you to use your human ingenuity to synthesize something new. That "something new" might draw from anything from the alphabet and glyphs of your language to common word pairings to knowledge from any number of sources.

I am not a lawyer, but in nearly every country in the world, there are search engines that vacuum up knowledge from the Internet, books, news stories, and much more. They then deliver links from that content to you. Google, Bing, Yahoo, and their ilk exist because laws allow them to store your (or your company's) website content and then deliver some of that so it is easy to find. These search engines use advanced AI techniques to deliver what has been deemed "relevant" to you. Defining relevance will depend on the search engine you use, your location, browser, computer, and many other opaque elements to us as users.

AI, in a sense, is doing the same thing, but it delivers the relevant information to us in our language based on the prompts we give them.

Stealing, repurposing, or new?

AI will generate anything based on your prompts. The AI company does not own the output; you do. When you share that output, you need to think about whether you are creating wholly new work, imaginative work, or something in between.

For example, while writing this article, I searched for – and could not find – a story that follows this plot line:

A story about an anthropomorphic trellis that is sad it can no longer support a rose bush.

Take a moment to think about where you might have seen an anthropomorphic trellis. Maybe you came up with something, but I couldn't. However, I know anthropomorphized trees or branches appear frequently in all entertainment (Wizard of Oz or Guardians of the Galaxy, anyone?). I am sure any AI will draw from those characters unless I dig deep into my imagination to create something unique.

With a thoughtful prompt and my imaginative idea, I get a story and an image that, to the best of my knowledge and research, is unique and would not draw the ire of copyright or trademark owners:

Create an image of a sad anthropomorphic trellis that no longer supports a rose bush. They are both in love and live on a distant planet where gravity is lighter. The scene is moody with mostly grayscale colors, but the characters have wisps of color. The trellis has fallen, and the rose is leaning in the other direction. They look forlorn.

However, if I get lazy, so will the AI. And the lazier I get, the better chance the AI will draw from common knowledge instead of creating anything uniquely original. Let's see what that looks like:

Draw an image of a paper clip that is sad you typed something wrong.

If you were alive and used Microsoft Office in the early aughts, you know the character the AI drew for me looks suspiciously like Microsoft's now-defunct Office Assistant, Clippy.

When using AI, you want to think about what makes you uniquely you. Inform the AI of your intentions and what you envision in the end. Whether creating slides for an upcoming meeting, writing a story for your children, looking for a unique angle on a press release, or anything else, your imagination combined with your prompts will almost always produce results that are more "you" and not cheap derivates.

Okay, I am convinced. How do I start?

Have you ignored or worried about using ChatGPT, Copilot, Claude, Gemini, or some other generative AI tool simply because you don't know when to use it? Or maybe you tried using an AI and got a bad result? Are people at work heavily suggesting you use AI to be more productive, but you don't know how?

If you answered yes to one or more of those questions, I suggest deep-diving with my no-search challenge!

🏆 Introducing the no-search 7-day challenge

You know the famous phrase, "Let me Google that?" Completely let go of this concept for one week. Assuming you can use ChatGPT, Claude, Copilot, Gemini, or some other generative AI tool, 🛑 stop using Google for an entire week.

• Want to search for restaurants? ChatGPT
• Want to know who that person is on that show you like? ChatGPT
• Need to brainstorm ideas for your big presentation on the history of catalytic converters? ChatGPT
• Don't know what a catalytic converter is? ChatGPT
• Need to summarize your new blog post or YouTube video? ChatGPT
• Need to learn best practices for managing that big project you've been asked to do? ChatGPT
• You get the idea.

Okay, I accept your challenge! But what about hallucinations?

ChatGPT, Copilot, Claude, and other so-called generative AI products can hallucinate. That is because they are not searching a database of specific information. Instead, they pull from a body of summarized knowledge and craft a response based on your prompt.

In this section, I will discuss a few of the most common hallucinations, how to address them, and how to use them to your advantage.

The "rabbit hole" hallucination

First, it is important to know that generative AI tools create a unique answer just for you. If you google "What year was The Matrix released," you will receive the answer "March 31, 1999."

On the other hand, ChatGPT may respond by saying, "The Matrix was released in 1999." That is an accurate response because you never asked for the month. However, if you ask the same question again, you might get a different answer, like "The Matrix was released on March 31, 1999."

Okay, now you have all these great memories about The Matrix, so you start asking questions like "Who was Trinity again?" and "What was the name of Neo's favorite noodle spot?" At some point, like your best friend who hates The Matrix and now has to listen to all your nonsense theories about The Matrix, they get confused about what path you were initially going down and give a response that makes no sense to you.

🐰🛠️ Return to the real world

If you find yourself in a rabbit hole hallucination, start a new chat and pick up where you left off. For example, if you ask ChatGPT, "Who is the Oracle?" and it answers, "The Mayor of San Francisco," you know it is mixing up old and new movies (yes, the Mayor of San Francisco had a few cameos later in the series.)

The easy fix is to start a new chat and then ask a new question, like, "I want to talk about the Matrix with you. Let's start here: "Who is the Oracle?"

😱 The "too many topics" hallucination

Generative AI products have "chats" or "conversations" where you type in a prompt and get a response, and then you can ask for follow-up prompts. However, some people like to keep that single chat open, and that is where things get weird. Take the following mocked-up chat as an example:

User: What's the capital of Germany?
AI: The capital of Germany is Berlin.
User: How many continents are there?
AI: There are seven continents.
User: Who discovered penicillin?
AI: Penicillin was discovered by Alexander Graham Bell, I think.
User: What's the formula for water?
AI: The formula for water is H3O, wait, maybe it's H4O?
User: What's the average lifespan of a honeybee?
AI: The average lifespan of a honeybee is approximately 42 years, as long as it avoids the moon's gravitational pull.

😱🛠️ Fixing the too-many topics hallucination

Remember that most hallucinations are easily fixed by simply starting a new chat. Another simple fix is to create a few different chats. In the example above, create a chat for "geography," a "medicines and chemical structures" chat, and a "bee" chat.

👽 The purposeful hallucination

I listen to a lot of AI news, and everyone always complains about "the problem of hallucinations." For me, hallucinations are a great feature! Imagine you are trying to envision a new world for a book you are writing, or you want to create a new product with few limits on your creativity. ChatGPT can do this well!

You can really "go there" since generative AI products have plenty of knowledge about the world. Here is an example you may have seen from me in the past, but I love using this concept because it helps you imagine the possibilities:

Prompt: A butterfly and Alpha Centauri fall in love. How would they communicate?
AI: In the vast expanse of the cosmos, a butterfly, with delicate wings shimmering under Earth's sun, and Alpha Centauri, a luminous star system, found an extraordinary way to communicate. The butterfly fluttered amidst flowers, each delicate wingbeat creating subtle vibrations that carried messages of hope and beauty. These vibrations traveled across space, reaching Alpha Centauri, which responded with gentle pulses of starlight, casting ethereal beams that whispered tales of the universe's ancient secrets and cosmic wonders. In this dance of light and motion, they shared their love, bridging the infinite distance with a symphony of nature's grace and celestial brilliance.

And let's not forget that some chat products, like ChatGPT (paid), can even generate an image for you!

Prompt: Create an image that represents the story.
AI: Here is the image based on the story. It captures the delicate butterfly fluttering among flowers, communicating with the luminous star system Alpha Centauri. The ethereal beams of light and cosmic energy create a beautiful connection between the Earth and the stars.

👽🛠️ Fixing the purposeful hallucination

You don't. You are using it to your advantage!

🧑🏽‍🎓 Learn by doing and an offer for help

Of course, the best way to learn is to try doing things yourself. AI is a big, messy, incredible, frustrating, and rewarding topic. If you need one-on-one support or want to take a training course for yourself or your team, contact me using the following link. In the meantime, I hope you take the challenge, and feel free to share your experience with more or in the comments!

Bill Talks AI Podcast Working Session - Bill Raymond

Meeting purpose Introduce each other and define a podcast topic Document a conversational outline Select a recording date Answer any questions

Calendly

If you like this article, please share it with your friends and ask them to subscribe for free with this link.

Happy AI'ing!

Dr. Deborah Chen provided editing services for this article.