With how fast the AI landscape is evolving, this page is basically going to be WIP for the foreseeable future.
That being said, here are a few basic principles (and their corresponding implications) that I believe will stand the test of time:
ChatGPT, Google Gemini, Claude, Copilot, Grok, Perplexity, Jasper... there seems to be an infinite number of AI tools.
Although the smaller players may carve out a niche for themselves (Perplexity is probably the best example of this), most users will stick to the offerings from the large tech companies in the long run simply because it's more convenient.
Case in point: Pika was one of the first platforms to showcase text-to-video capabilities but most users will end up using Sora simply because OpenAI will most likely include text-to-video within ChatGPT (why bother opening up a new window and using Pika?)
Implication for you and me: Don't get distracted by new AI tools. Stay focused on getting better at using 1-2 tools from the Big Tech companies (e.g. Google Gemini and ChatGPT) by learning how to prompt.
Many assume that, soon, AI chatbots will be so intuitive that learning how to interact with them – the art of 'prompt engineering' – will become obsolete.
"Why bother if the bot will soon get the hang of what I mean anyway?"
This is a dangerously lazy mindset because of two reasons:
Even if AI chatbots are brilliant in five years, right now, skilled prompting makes a massive difference in the quality, accuracy, and depth of the responses you'll receive.
Like any tool, you get vastly better results from AI when you understand how it works. Taking a little time to educate yourself in basic prompt engineering means you'll save time (and frustration!) in every interaction with these powerful systems.
Prompt engineering is a skill. Skills add to your personal toolbox and make you a more adaptable, resilient problem-solver. Even if this specific skill becomes less important, you've wired your brain to think in terms of inputs and outputs, parameters and instructions.
Think of it like this - Even the most sophisticated calculators can't replace the knowledge of core math concepts. Building that knowledge makes you a flexible thinker, able to adapt as new technologies emerge. The same holds true for any AI system you'll use in the future.
Implication for you and me: Pick a use case (text-to-text, text-to-image, text-to-video, etc.) and get good at prompting. My Perfect Prompt Formula video is a good place to start 😉. And if you're ready for more advanced content, I highly recommend LearnPrompting.org's courses!
Just like the principles on this page (lol), AI chatbots remain a work in progress. They are prone to errors, misinterpretations, and occasionally nonsensical output (i.e. hallucinations).
Mistakes by AI systems can be costly, from factual inaccuracies to malfunctioning code. And although AI's reliability will definitely improve, it's currently unwise to entrust critical tasks solely to these tools.
View AI as a powerful aid, not an infallible replacement for human judgment.
Implication for you and me: Maintain healthy skepticism. Always verify AI-generated information and outputs.
You know how those targeted ads can get a little creepy? For example you think about buying a new pair of hiking boots once, and suddenly your whole online world is filled with deals on outdoor gear? That's a tiny taste of how AI can reinforce your own perspective.
The thing is, these chatbots are trained on mountains of data. That data includes "good stuff" – facts, figures, and creative inspiration. But it ALSO includes all the biases, prejudices, and downright misinformation floating around the internet.
AI systems can learn that stuff too.
It means you could get stuck in a feedback loop, where the AI only shows you stuff that agrees with your existing worldview. It might feel comfortable, but it limits your learning, challenges your thinking less, and can even be subtly harmful.
Implication for you and me: A prompt I literally use on a daily basis is: "Ask me 5 questions that will improve the response you will be giving me."
This prompts (no pun intended) the AI tool to give you a list of factors you might not have considered (since we're in our own echo chambers right) and these in turn will drastically improve your final output!
I design and create all my thumbnails so when text-to-image AI tools were launched, I thought paying for a Midjourney subscription would be a great way to enhance my designs.
However, I quickly realized the feature I needed to use the most (using AI to expand and enlarge an existing image) didn't exist in Midjourney; I should have used Adobe Firefly and Canva instead.
Implication for you and me: If you want to start using a particular AI tool because of a specific feature, look past the hype by asking yourself what exactly you're hoping to achieve first. Do some basic research to confirm that AI tool can satisfy your needs before investing your precious time (and money).