AI Is Smart, But Still Dumb? The Truth About AI's Limitations
AI Isn’t Your Replacement—It’s Your Sidekick: A Practical Guide for Startup Leaders
Weekly Snippets 🤓
Liquid Death and Yeti: Because nothing says “party” like a $1,500 limited-edition casket cooler. Grab yours now to keep your drinks—and sense of irony—on ice forever.
Klarna Halves Workforce With AI: Klarna’s new motto: "Buy now, replace humans later." With plans to slash its workforce in half using AI, it looks like ‘pay later’ might have a new meaning if their customer service experience drops. (Get it? Pay later? 😏)
$100m Podcast Deal: The Kelce brothers just hit the podcast big leagues, signing a deal with Amazon’s Wondery worth over $100m. Guess the NFL isn’t the only place to score big.
Virtual Standup Comedy Audience. No joke, it is silly—but this virtual audience thinks you’re hilarious.
TL;DR 👓
Think AI’s taking over? Not so fast. Even the smartest AI is still like a new intern—good for grunt work but clueless about deeper brand nuances. AI shines when used as a tool for data analysis, automation, and brainstorming—not for running the show. So, keep your role as the creative strategist and let AI be the sidekick. Use it wisely, or get comfortable watching others stay ahead.
AI Is Smart, But Still Dumb? The Truth About AI's Limitations
Think AI’s here to steal your job?
Let’s set the record straight.
55% of people worry it’s replacing them—while few actually understand what it can (and can’t) do.
For founders and marketers, AI is a great tool, but it’s not the replacement people fear.
Think of AI as an efficient intern. Useful for grunt work, but still needs your direction to add depth and strategy to your brand.
In this post, I’ll break down the reality vs. the hype, so you’ll be able to:
See why advanced AI is like a new intern—potentially helpful but clueless about your brand’s nuances.
Shut down the myth that AI will take over your marketing.
Learn practical ways to use AI’s processing power to support, not replace, your creative decisions.
In this post, I'll explore:
The Real Hype vs. Reality of AI
AI’s Limitations (Think “Chinese Room” Experiment)
Why Machine Learning and Large Language Models Aren’t Thinking Machines
The Hallucination Problem: AI’s Habit of Confidently Lying
How Smart Founders Use AI Without Losing Brand Voice
Why We’re Not Headed for Skynet Just Yet
My Final Thoughts
The Rabbit Hole - For Those Who Want to Go Deeper
At the end, you’ll see AI as a useful sidekick, not a replacement risk.
The Hype vs. The Reality
Sure, AI might snag a few jobs, but is it really as smart as people think?
Not quite.
According to a 2023 report by Gartner, 85% of AI projects deliver inaccurate results due to a lack of contextual understanding. There’s a common myth that AI knows everything, but that's far from the truth.
Think of AI as a skilled copywriter who’s never met the client. It can create impressive content but lacks a true understanding of the brand and its customers.
AI is less like Einstein and more like a parrot—mimicking what it’s been fed without understanding what it’s saying.
For example, a fashion brand using AI-generated content might initially see decent engagement. However, without a deep understanding of the brand's style, AI can misinterpret seasonal trends or cultural nuances, leading to tone-deaf campaigns that alienate customers.
This thought experiment will change how you see AI—and might just make you rethink your marketing game plan.
The Chinese Room: A Lesson in Pretend Intelligence
You’re at a party.
Someone hands you a script in a language you don’t speak.
You read it out loud, and everyone thinks you’re fluent.
But guess what?
You have no idea what you’re saying—you’re just following instructions.
Congrats, you’re a human parrot.
This is basically the “Chinese Room” thought experiment cooked up by philosopher John Searle.
Imagine yourself in a room filled with Chinese symbols and a huge rulebook. People outside the room pass you notes in Chinese.
You use the rulebook to reply.
To the people outside, you’re a Chinese language whiz.
But inside? You’re clueless.
Just following the script to make it seem like you know how to speak Chinese.
Just like in the Chinese Room, AI can mimic understanding without actually grasping the meaning behind its responses.
This becomes particularly problematic in marketing contexts.
Imagine an AI tool creating social media posts for a hospitality brand. It can mimic a friendly tone and respond to customer questions. But it lacks emotional intelligence for sensitive customer service issues.
It also can't grasp the nuances of the brand’s customer loyalty strategy.
How Machine Learning and LLMs Really Work
AI is essentially about two things:
Machine Learning (ML) and Large Language Models (LLMs).
Both rely on vast amounts of data but serve different purposes.
Machine Learning (ML): Think of ML as a pattern-finder. It analyses past data to predict future outcomes. It's like a true crime fan who thinks they can solve any case just by recognising patterns. Impressive, but it doesn't truly understand the why behind the data.
Large Language Models (LLMs): LLMs are the chatty ones. They’ve been trained on massive amounts of text, so they sound convincing when they respond. But they don’t actually "think." They just assemble sentences that seem right based on probability.
AI is great at mimicking patterns but lacks real understanding or strategic thinking. It’s like someone who’s watched every football game but can’t coach a team to victory.
Essentially, it’s operating in the Chinese Room, following the rulebook without any genuine comprehension.
The Hallucination Problem
Here’s the dirty little secret about AI.
It hallucinates. No, really.
It sometimes makes up info and sounds confident but is totally wrong.
Imagine using an AI-generated report for your next product launch. If the AI misreads customer feedback trends, it could lead to poor decisions and ineffective marketing strategies.
AI doesn’t know it’s lying—it’s just rearranging data in a way that looks good.
Trusting AI without verifying is like running a marketing campaign based only on a trending hashtag you have not checked.
It’s not going to end well.
Remember Snapchat's My AI Fiasco?
Snapchat thought it’d be cool to roll out “My AI”?
This AI chatbot was supposed to be all friendly and helpful, like your favorite barista. But things went sideways fast. Reports started popping up about how “My AI” was giving sketchy advice to teens on stuff like mental health and relationships. Yikes.
This fiasco showed one major thing: AI can’t yet handle complex, sensitive issues. It’s good at mirroring patterns, but when it comes to understanding real-world, nuanced situations? Total fail.
Even the most advanced AI still needs humans to make sure things stay ethical, unbiased and accurate.
How Smart Marketers Play the AI Game
So, how should marketers approach AI without falling into these traps?
AI as Your Sidekick, Not Your Boss: AI is great for grunt work—think data analysis, repetitive tasks, idea generation. But when it comes to storytelling and making emotional connections? That’s all you.
Double-Check Everything: Always review what AI comes up with. Sure, it’s grammatically correct, but is it in line with your brand voice? Is it accurate? Don’t let AI run wild.
Mix AI Insights with Human Creativity: Use AI to find trends and insights, but let your creative side run the show. AI can tell you where the crowd’s heading; you decide how to stand out.
Brainstorm Like a Boss: Use AI to kickstart ideas. Treat its suggestions as the starting point, not the final draft.
Plan for the “What Ifs”: Use AI to run different marketing scenarios. It’s like a crystal ball, showing possible outcomes. Just remember, it’s not a guaranteed win.
Why We’re Not Headed for Skynet Just Yet
The hysteria over AI becoming as smart as humans is just that—hysteria.
Reaching human-level intelligence isn’t just about better coding. It would demand massive computational power and energy breakthrough. AI models with billions of parameters need a lot of energy to train.
Today’s AI is already guzzling power like an SUV in a gas crisis. Making reaching Artificial General Intelligence (AGI) logistical nightmare, not a near-term reality.
So, while it’s essential to stay updated on AI advancements, there’s no need to panic about an AI apocalypse just yet.
My Final Thoughts
AI is powerful, but it’s not a miracle worker.
Sure, it speeds up processes and crunches data, but it lacks the core of what makes us human—creativity, empathy, and true understanding.
Think of AI as a super-efficient intern: great for handling tasks, but it still needs your direction. You’re the creative director, not AI.
Use AI as your sidekick, not your boss.
To stay ahead, sharpen your critical thinking and truly understand AI’s capabilities and limits. It’s not about feeding AI more data; it’s about bringing your human insight to the table.
AI is clever, but make no mistake—we’re still the ones calling the shots.
The Rabbit Hole - For Those Who Want to Go Deeper
Check out these hand-picked resources:
The Chinese Room Experiment Explained (BBC)
This is a really good explanation and breakdown. Highly recommended.
This was quite an interesting read and I loved the Chinese experiment!
You phrase it entirely well: LLMs currently are like having an intern! It’s “someone” you need to cross-check, work with, teach! An intern can be very good on a Monday but tired on a Friday and the outputs get worse 😂
Every prompt is not final, it’s an iteration. You need to keep improving it otherwise it just sounds like a text that is bland/ confusing or might be entirely hallucinating.
Great read! Loved this!