Will AI outsmart humans and take our jobs

AI is advancing at a pace that’s both impressive and, at times, unsettling. It’s no surprise that many of us are wondering what the future of work will look like over the coming decades.

This is exactly the kind of question we explore at the Centre for Artificial Intelligence Research and Optimisation (AIRO) at Torrens University - not just what AI can do, but what it should do, and how humans fit into that future.

In reality, AI has been outsmarting humans in specific tasks for almost 30 years. What’s new is visibility. Large Language Models (LLMs) such as ChatGPT and Microsoft Copilot have brought AI out of the lab and into everyday life and suddenly, it feels very personal.

So, what does the research say about AI versus human intelligence?

What do we really mean by “intelligence”?

When we talk about intelligence, things get fuzzy very quickly. Even humans can’t agree on what it means.

For example, my wife and my mum don’t agree on whether I’m intelligent. My mum thinks I am. My wife… remains unconvinced. So clearly, intelligence isn’t something we’ve measured very well.

That ambiguity matters. Mimicking the human brain perfectly isn’t necessarily desirable anyway - human judgement and performance are heavily influenced by emotion, fatigue, stress, and context. That’s not always a bug, but it certainly isn’t a clean benchmark.

Comparing human and machine intelligence (and why it’s misleading)

Directly comparing human and artificial intelligence is a bit like comparing a bicycle to a freight train.

The human brain has billions of neurons, each processing roughly a thousand signals per second. AI systems, by contrast, operate at a vastly larger computational scale - performing billions of operations per second across massive data centres.

Energy use tells a similar story. AI consumes electricity on the scale of a small city. The human brain runs on whatever you ate for lunch. This isn’t a fair competition - and it never was.

The myth of a “race” between humans and machines

The idea of a dramatic race between humans and machines was largely created by the media. AI has already beaten world champions at chess (1997) and Go (2016) - the latter being far more complex.

What has changed is natural language. Tools like ChatGPT feel conversational, almost human, which creates the impression that AI has suddenly arrived and is catching up fast. In reality, it’s been here for a while. It’s just learned how to talk.

Where humans still outperform AI

Recent research studies (from 2022 through to 2025) show a consistent pattern: AI excels at well-defined, data-rich tasks. Humans tend to outperform AI in open-ended, creative, and ambiguous situations - especially when decisions must be made with incomplete or shifting information.

Take business forecasting. AI is an exceptional calculator and optimiser. But it struggles with human fear, sentiment, and irrational behaviour.

If AI could reliably predict the stock market, everyone would already be rich. It can’t - because markets are driven by emotion as much as logic.

During COVID, many indicators suggested house prices would fall. Instead, fear and uncertainty drove prices up, and in some places, they doubled. That’s not a failure of data - it’s a reminder that humans don’t behave like spreadsheets.

Creativity, media, and the human edge

Creativity is where things get especially interesting.

Researchers often test creativity using “alternative uses” tasks. For example, asking people to list as many uses as possible for a common object like a light bulb.

AI performs better than the average human in these tests. But highly creative individuals consistently outperform AI by a wide margin.

That distinction matters. Creative thinking isn’t about templates or averages. It’s about originality, context, judgement, and sometimes breaking the rules altogether - areas where humans still shine. This is particularly relevant in creative media, communication, design, and innovation-driven fields.

AI in healthcare: collaboration, not replacement

Healthcare shows both the promise and the limits of AI very clearly.

AI systems have already passed some written medical exams. That doesn’t mean doctors are about to disappear. Some technical or administrative aspects of medicine may be automated - but roles that rely on trust, empathy, judgement, and human interaction remain deeply human.

Nurses, carers, and frontline clinicians aren’t just delivering information. They’re responding to people.

And people are complicated.

So… will AI take our jobs?

Here’s the uncomfortable truth:

AI will outperform most average humans in certain tasks. But educated humans still outperform AI in others - often spectacularly.

The best outcome isn’t competition. It’s collaboration.

We don’t race horses anymore. We ride them. The real race is between humans to see who can use AI most effectively.

What this means for universities and the future of work

This is where education matters most.

University courses can’t be about competing with AI. They must be about developing graduates who can manage AI disruption.

That means focusing on:

• Structured creativity, not just content recall
• Engineering-centric innovation, grounded in real systems
• Well-informed critical thinking, especially under uncertainty
• Human-centric mindsets, where ethics, judgement, and responsibility matter

AI literacy will be essential. Everyone will need to know how to work with AI, but not everyone needs to become AI.

Through the AI Research Centre, we will continue to focus on AI that is transparent, trustworthy, and human-friendly while providing education that equips people to do what humans do best, in ways machines still can’t.

And that’s a future worth building.