Hot take, hot cakes.

Look, I get it. The shiny new toy that spits out essays, generates images, and occasionally hallucinates facts about non-existent court cases is impressive. We've slapped the label "Artificial Intelligence" on it, and suddenly everyone's either preparing for the robot apocalypse or planning to marry their chatbot. But here's my scorching hot take for today: AI isn't actually intelligent. Not even remotely close.

What Even Is Intelligence, Anyway?

Before we dive into why I'm raining on the AI parade, let's take a moment to consider what intelligence actually means. The word itself comes from the Latin "intelligentia," which derives from "intelligere" – to understand or perceive. At its etymological core, intelligence implies comprehension, not just pattern recognition.

Colloquially, we understand intelligence as something that encompasses:

• The ability to learn from experience

• Adapting to new situations

• Understanding abstract concepts

• Using knowledge to manipulate one's environment

• Self-awareness and reflection

• Emotional understanding and response

• Intentionality and purpose

But perhaps most importantly, intelligence involves understanding the meaning behind information – not just processing it.

Ilustration by ilham saputra from Unsplash

Cultural Context: Intelligence Is Not Universal

Here's where it gets interesting. Our understanding of intelligence isn't even consistent across human cultures. Western conceptions typically emphasise logical-mathematical reasoning, verbal abilities, and individual problem-solving. But this isn't universal:

• In many East Asian cultures, intelligence encompasses social harmony and collective wisdom (source: trust me bro, I’m Asian) 

• Some Indigenous perspectives view intelligence as environmental adaptation and intergenerational knowledge

• Various African philosophical traditions emphasise communal problem-solving and social intelligence

• Some cultures value emotional intelligence and intuition far above analytical reasoning

Even within academic circles, we have theories of multiple intelligences, emotional intelligence, and practical intelligence that challenge the notion of intelligence as a singular concept.

So when we say "Artificial Intelligence," we're already operating from a limited, culturally specific definition. And even by that narrow standard, AI falls woefully short.

How AI Actually Works (Spoiler: It's Not Magic)

Let me break it down for you: modern AI systems – particularly large language models like the one you might be using to generate content – are essentially sophisticated pattern recognition machines trained on massive datasets of human-created content.

They don't "think" in any meaningful sense. They predict. They analyse patterns in data and generate outputs based on statistical probabilities. If I write "To be or not to..." an AI can complete it with "be, that is the question" not because it understands Shakespeare's existential contemplation, but because it's seen this sequence of words together millions of times in its training data.

When an AI appears to "reason," it's performing an impressive imitation – a statistical approximation of what reasoning looks like based on the texts it's consumed. It's the digital equivalent of a parrot who's heard so many conversations that it can sometimes string together contextually appropriate responses without understanding a single word.

Btw, Dr Yejin Choi has been doing exceptional research on AI reasoning and smaller models - geek out with me later. 

The Great AI Illusion

The reason we're all so easily fooled is that humans are meaning-making “machines”. We're hardwired to attribute intention, understanding, and consciousness to things – a psychological tendency called anthropomorphism. We do it with our pets, with cartoon characters, and definitely with AI systems that have been specifically designed to mimic human communication patterns.

Media companies aren't helping either. Headlines screaming "AI ACHIEVES HUMAN-LEVEL INTELLIGENCE" get more clicks than "Statistical Model Shows Incremental Improvement in Pattern Recognition." The result? A bizarre collective delusion where we simultaneously overestimate what AI can do while underestimating the profound complexity of human intelligence.

Let's be clear: when a company announces their AI has "learned" something new, what they really mean is "we've adjusted the parameters of our statistical model based on new data inputs." It's not learning in the human sense – it's algorithmic adjustment. Deadass.

Ilustration by Alghozy from Unsplash

The Intelligence Test

Want proof? Try having an AI engage in these fundamentally human aspects of intelligence:

1. True creativity – not remixing existing patterns, but creating genuinely novel ideas

2. Intentionality – having actual goals and desires that aren't programmed

3. Self-awareness – not simulating responses about having consciousness, but actually experiencing consciousness

4. Understanding – grasping the meaning and context behind information

5. Emotional comprehension – not mimicking emotional responses, but actually experiencing them

6. Common sense reasoning – consistently applying basic logical principles to new situations

Ask an AI about its childhood (which it doesn't have), and it will happily fabricate one for you. Ask it to explain a concept it wasn't trained on, and it will confidently hallucinate an answer. These aren't signs of intelligence – they're flaws in a sophisticated mimicry machine.

Reframing AI: The Tool, Not The Replacement

Here's where I want to pivot. Recognising that AI isn't truly intelligent doesn't mean it isn't valuable. It's an extraordinary tool – perhaps one of the most powerful we've ever created. But that's exactly what it is: a tool, not a mind.

The most productive way to think about AI is as an amplifier of human intelligence, not a replacement for it. Like the calculator didn't eliminate mathematicians but rather freed them to tackle more complex problems, AI won't replace human thinking but can help us explore new frontiers of creativity and problem-solving.

When we stop anthropomorphising AI and start seeing it as a tool:

• We make better decisions about where to apply it (and where not to)

• We maintain appropriate oversight and human judgment

• We focus on human-AI collaboration rather than replacement

• We invest in improving AI's weaknesses rather than overestimating its capabilities

• We maintain realistic expectations about what it can achieve

No Identity, No Agency: AI's Fundamental Limitation

Here's another critical point often overlooked in breathless AI coverage: these systems have no identity or selfhood. None whatsoever.

Identity—having a continuous sense of self that persists through time and experience—is fundamental to human intelligence. As philosopher John Locke argued centuries ago, personal identity comes from consciousness and memory connected over time. AI systems, despite their impressive outputs, possess nothing remotely similar.

Dr. Abeba Birhane, cognitive scientist and AI ethics researcher, puts it perfectly - The 'intelligence' in AI is not human like it all. It lacks embodiment, lived experience, and the social embedding that shapes human cognition. Her research demonstrates that without embodiment in the world and social context, AI cannot develop the kind of intelligence humans possess.

When an AI generates text about "its experiences" or "its preferences," it's not drawing on any actual experiences or preferences—it's constructing a statistically likely response based on patterns in human-written texts. Philosopher and cognitive scientist David Chalmers notes that current AI systems are like zombies - they’re here but no one’s home.

Moving Forward: Human-Centred AI

The real promise of AI isn't in creating artificial minds but in extending human capabilities. When we use AI to automate mundane tasks, we free up human creativity and innovation—our uniquely human capacities that no algorithm can replicate.

MIT professor Max Tegmark argues in his research, "The goal should not be to create AI that thinks like humans, but AI that helps humans think better." This perspective shifts our focus from trying to replicate human intelligence to amplifying it.

Consider this: throughout history, automation has been most successful when it targets routine, repetitive tasks, allowing humans to focus on creative problem-solving, emotional connection, and moral reasoning. A 2023 study from the Oxford Martin School found that when AI is deployed specifically to automate routine cognitive tasks, worker satisfaction and creative output increase significantly.

What we need is a cultural shift in how we conceptualise these systems – moving from the science fiction fantasy of artificial minds to the practical reality of advanced tools. This shift isn't just philosophical; it has real implications for how we develop, regulate, and use these technologies.

So the next time someone tells you AI is going to achieve consciousness or replace human thinking, maybe offer them this analogy: saying AI is intelligent is like saying a book is intelligent because it contains knowledge. The book doesn't know what it contains – it's just a container. AI is a more sophisticated container, but it's still just processing and regurgitating human-created information without understanding or identity.

AI isn't intelligent – but it is useful. And that's actually enough.

What do you think? Is AI intelligent or just a sophisticated tool? Drop your thoughts in the comments below – I promise a human will read them, even if an AI might help sort them.