AI is everywhere — but do we really understand it?
It’s seemingly omnipresent — answering customer support questions, curating playlists, running ads, and offering advice on everything from fitness to finance.
Lately, my ChatGPT-powered mindfulness coach even delivers customized morning exercises at 8 AM sharp — a fun experiment that probably deserves its own post.
Like a genie in a lamp, AI seems to grant wishes on demand.
But what’s actually inside the lamp?
I thought I had a solid grasp of AI. For years, I worked with data analytics in eCommerce, customer experience and tech startups, and used AI-powered tools regularly.
It’s like driving a car with an automatic transmission — you know how to get from point A to B, but don’t necessarily understand what’s happening under the hood.
Yet, AI still felt like a black box — this mysterious entity that consumed our data and spat out insights.
One nagging thought kept coming up:
Do I actually understand how AI works?
I wanted to get my hands on the gears and learn to drive stick — maybe not as an engineer, but enough to understand how it makes decisions and, more importantly, how to use it better.
After researching a dozen AI certification courses, I chose MIT’s No-Code AI & Machine Learning Certificate Program — a hands-on, 12-week course promising to “achieve AI mastery with MIT faculty and live mentorship from industry experts.”
And where better than MIT — the cradle of modern AI, where Marvin Minsky co-founded its first AI lab in the 1960s? I was sold.
Here are 10 unexpected Myth vs. Reality facts that demystified AI for me, and what they mean for its actual use.
We have more in common with AI than we think — TRUTH ✅
The human brain is a prediction machine — constantly scanning memories to predict the future.
Its primary function is to keep us safe.
What we perceive as “safe” is mainly shaped by our life history, which becomes the foundation of the super-database our brains use to make decisions and predict outcomes.
AI operates similarly. Machine Learning algorithms — the backbone of AI — analyze historical data to recognize patterns and predict outcomes.
What this means: whether it’s forecasting stock prices, detecting fraud, or choosing the next word in a sentence, AI, like the human brain, is always calculating the most probable next step based on what it has learned.
2. AI Understands Words, Images, and Speech? Not Quite — MYTH ❌
AI doesn’t understand language the way we do. Everything you say, write, or show to AI must first be converted into numbers — because that’s the only language AI understands.
Example: When you see the word love, you connect it to emotions, experiences, and meaning.
When AI sees love, it doesn’t “feel” anything — it represents it as a series of numbers based on patterns in text, like coordinates in a massive dataset.
A simplified word embedding for “love” might look like this:
🔢 love = [0.23, -1.52, 0.77, 0.98, -0.45]
What this means: AI can mimic human intelligence, but it doesn’t actually “think” or “understand” the way we do. When AI generates text, it’s not expressing thoughts or emotions — it’s statistically guessing the next best word.
Here’s another great example — check out this diagram that illustrates what happens behind the scenes when we talk to AI.
How AI “Hears” and Understands Human Speech
3. You Need to Be a Programmer to Learn AI — MYTH ❌
This was arguably one of the most exciting myths we explored during my MIT AI course!
❌ You don’t need to know Python or write algorithms to work with AI.
✅ What matters most is the quality of the data and understanding how to use AI tools effectively.
Examples of No-Code AI Tools:
Teachable Machine — Train AI to recognize your movements, sounds, or facial expressions using your webcam.
Example use case: Imagine having an AI yoga instructor that adjusts your poses! You can even use your webcam to teach it to recognize moods based on your facial expressions.
(sadly, it can’t pour a glass of wine for a sour mood — yet.)
RapidMiner — if you want to visually experiment with AI models, this tool lets you build, test, and refine them without coding with a drag-and-drop interface.
Example use case: Want to create a decision tree? Prune it? Or plant an entire Random Forest? We used this in my MIT course to predict hotel booking cancellations and optimize occupancy rates.
OpenAI Playground — It’s like finding a sports edition of ChatGPT with manual transmission.
Example use case: Want AI to be more creative? Crank up the “temperature” setting. Need it to stay focused and avoid repeating words? Fine-tune the “presence” control.
4. AI is Superior at Multitasking — MYTH ❌
AI is fast, but it actually struggles with multitasking — much like us. Ever tried writing an email while browsing social media? It doesn’t go well. AI faces the same issue.
🤖 AI Example: performs way better when tasks are broken down step by step instead of being overloaded at once.
🧠 Human Example: You write a better report when you focus on one section at a time instead of flipping between five tasks.
MIT professors emphasized this in the prompt engineering course — using “Let’s think this step by step” (aka Chain-of-Thought prompting) significantly improves AI’s responses.
Dump everything into one prompt, and you’ll get mediocre results. Break it down step by step, guiding AI through the process, and the output becomes far more accurate and effective.
Pro Tips for Better Prompts
Courtesy of Jonathan Yarkoni, ex-Google AI Trainer.
- Instead of dumping a huge request into ChatGPT, try breaking it down into smaller steps.
- Ask ChatGPT to critique your prompt. Then ask it to fix the prompt.
- Ask GPT to make your prompt better (Great method for beginners.)
- Reverse prompt engineer: Start with the end result. Ask GPT to give you a prompt that can generate it.
👉 Takeaway: Whether AI or human, single-tasking beats multitasking.
5. AI is only as good as the data it’s trained on — TRUE 💯
If AI had a family tree, math would be its great-grandparent, statistics its grandmother, and machine learning its parent.
Basically:
- Statistics explains the why behind numbers (analyzing past trends).
- Machine Learning uses past data to predict future outcomes.
- AI applies ML models to make intelligent decisions — but only as accurately as its training data.
As my former colleague and friend Ryan Leglu, Senior eCommerce Leader at The North Face, puts it:
As my former colleague and friend Ryan Leglu, Senior eCommerce Leader at The North Face, puts it:
“Operationally, AI starts and ends with data. With cybersecurity on high alert, are you willing to open up the data floodgates to AI?”
👉 Takeaway: Before committing to AI-driven decisions, ask: Is my data accurate, unbiased, and secure?
6. AI is new tech— MYTH ❌
AI didn’t just appear out of nowhere in 2022 — it became more visible.
It’s been around — (loud gulp) — for over 60 years.
The term Artificial Intelligence was first coined in the 1950s, and machine learning models have been powering everyday applications long before ChatGPT made AI feel magical.
Still think AI is new? Here are some ways it’s been working behind the scenes:
- Netflix & Spotify recommendations? AI at work
- Google Ads — AI has been predicting which ads you’ll click on since 2015 based on search history, browsing, and even offline behavior.
- Medical AI in diagnosing diseases — AI improves early cancer detection in mammograms with greater accuracy than traditional screenings.
- AI detects fraud every time you swipe a credit card.
- AI powers your search results — Amazon, Google, and brand websites use AI to show you the most relevant results based on behavior, location etc.
👉 Takeaway: AI has been working in the background, making our lives easier long before it became a buzzword.
7. AI makes you more productive — 🤔 (you decide)
In reality, AI can be a perfectionist’s worst nightmare…
…seriously, the worst. Because:
More Choices = More Time Spent Deciding
Instead of writing one “good enough” version and moving on, AI can pull you into a rabbit hole of endless rewrites and tone tweaks.
Trust me, I’ve gone deep down that hole — more than I’d like to admit.
👉 Takeaway: If you’re using AI, set a time limit. Otherwise, you might spend more time refining and optimizing than actually creating with your own voice.
8. AI and Humans Share a Superpower: Attention — TRUE ✅
In 2017, the research paper “Attention Is All You Need” introduced a breakthrough in how AI processes language — Transformers. This technology powers modern LLMs like ChatGPT, Gemini, and more, allowing AI to focus on the most important words in a sentence, even if they’re far apart.
Sound familiar? That’s because humans do the same thing.
When you read, your brain doesn’t process each word individually — it picks out key details, connects ideas across sentences, and fills in the gaps. AI now mirrors this process, making its responses smarter and more human-like.
Attention Shapes Both AI and the Human Brain
You’ve probably heard: “Neurons that fire together, wire together” and “Where attention goes, energy flows.”
An update from modern neuroscience on this:
“The brain physically rewires itself based on where your attention goes.” — Neuropsychologist Rick Hanson.
This ability to rewire and form new neural connections is called neuroplasticity.
Intentionally directing our attention — and tapping into neuroplasticity — is our cognitive superpower. Every time you choose where to focus, you’re shaping your future.
If you’ve made it this far, congrats — you’re already training your attention!
What this means
🧠 For Humans:
- Focusing on gratitude and learning → strengthens neural pathways for happiness and growth.
- Focusing on stress, distractions, and fear → reinforces those patterns instead.
🤖 For AI:
- Improves by analyzing key data points, not absorbing everything at once.
- Like humans, AI gets better with practice, refining accuracy over time.
- By focusing on what really matters, AI processes information much faster without getting lost in the weeds.
Tip: After feeling humbled (and honestly overwhelmed) by MIT’s lecture on Transformers and Neural Networks, another student in my cohort recommended this YouTube video. It demystifies attention, the key mechanism inside transformers and LLMs, breaking it down with fantastic visuals and simple explanations.
9. Reinforcement Learning in AI works just like human learning — MYTH❌
In AI, Reinforcement Learning (RL) is how models are trained through trial and error — by getting rewarded for correct responses or penalized for incorrect ones.
In neuroscience, reinforcement learning is how our brains form habits. As neuroscientist Dr. Jud Brewer explains, habits follow a simple loop:
trigger → behavior → reward
Whether it’s reaching for a snack when stressed or checking our phones when bored, these behaviors get reinforced because they feel rewarding — even if the long-term impact is not so good.
The key difference between AI and human learning: While reinforcement learning in AI is inspired by behavioral science, the human brain learns in far more expansive and diverse ways — through demonstration, verbal instruction, mimicking, conceptual understanding, and lived experience. It’s a depth of learning that RL in AI simply can’t replicate (for now).
To live it is to know it: bite into a banana, and you’ll learn much more about it than by reading about bananas in the encyclopedia.
Ways you may be participating in RLHF (Reinforcement Learning from Human Feedback):
- Choosing between two versions of an AI-generated response.
- Giving a thumbs up or down on AI answers (ChatGPT, Claude etc.)
- Participating in paid RLHF projects (e.g., companies like Outlier mass hire experts to help train AI with human input for Meta, Google and OpenAI) — ask me about that experience!
10. Myth: You Need Fancy Education to Master AI — MYTH ❌
MIT professors and instruction are top-notch. But the best AI education isn’t just in elite classrooms (or classZooms) — it’s online, free, and waiting for you. (Seriously, check out that YouTube video above!)
AI knowledge has never been more accessible. Take full advantage and learn from some of the best minds in the world — for free.
Here are a few great courses to start you on your AI journey (or should we say, wAI?)
-
Google’s Free AI Course — Beginner-friendly AI fundamentals.
Google AI Essentials: https://grow.google/ai-essentials/ - IBM’s Free AI Course — Practical AI applications and Industry-recognized credentials to boost your resume. https://skillsbuild.org/adult-learners/explore-learning/artificial-intelligence
- Harvard’s Free AI Course: Introduction to Artificial Intelligence with Python – perfect for those with coding experience looking to build AI-powered applications. https://cs50.harvard.edu/ai/2024/
Final Thoughts
So, remember how I was sold on “achieving AI mastery in 12 weeks” at MIT?
Well — do I have AI mastery?
Not even close.
But what I do have is a whole new level of confidence — in building AI models, growing decision trees and forests, engineering effective prompts, training agents, and orchestrating AI tools to amplify my work.
I may not have mastered AI, but I’ve learned first-hand that AI isn’t just a tool — it’s a collaboration. The better you understand it, the more it feels like a true partnership.
If you want to power up your AI skills, start today — no tuition required, just your attention.
Attention is the new gold.
How will you invest it?
Leave a comment