In 1994, mathematician Andrew Wiles solved Fermat’s Last Theorem – a problem that had defied proof for over 350 years. It wasn’t artificial intelligence (AI) that cracked it. It was human persistence, intuition and decades of foundational work.
Contrast that with what generative AI (GenAI) is capable of today. It can distil 1,000 pages into a one-paragraph summary, forecast customer churn, classify cyber threats or triage support tickets before your team finishes its coffee. Yet ask it to replicate Wiles’ proof, and it stumbles. Spectacularly.
This contrast is instructive. It shows how much time has been spent judging GenAI against what it can’t do, when the real opportunity lies in what it can. AI doesn’t need to unravel ancient theorems to change business. It just needs to make work smarter, faster and more secure.
That’s the GenAI shift: away from hype and novelty, toward scalable, repeatable value.
Usefulness trumps brilliance
A recent study by Apple, The Illusion of Thinking, noted how large language models perform poorly on logic puzzles like the Tower of Hanoi—even when given step-by-step instructions. While this made headlines, it shouldn’t have. GenAI doesn’t reason recursively. It predicts based on patterns.
And that’s fine, because business doesn’t run on logic puzzles. It runs on anomaly detection, workflow automation, signal correlation and reduction of operational drag. The kind of high-frequency, low-visibility work that underpins performance.
Waiting for AI to become “smart enough” to be impressive is like refusing to fly because planes can’t reach the moon. The real potential is already here.
The real value of AI: not magic, but leverage
AI’s practical strengths lie in natural language processing, predictive analysis and classification. These capabilities are not theoretical. They drive efficiency, cost savings and revenue acceleration. Consider just a few examples:
- Customer service: GenAI analyses tone and urgency, prioritises support tickets and drafts first responses, cutting time-to-resolution and easing pressure on agents.
- Finance and fraud: AI flags outliers, reconciles accounts and forecasts budgets with extraordinary precision.
- Sales acceleration: It extracts insights from call transcripts, updates CRM records automatically and creates tailored follow-up messages.
- IT and cybersecurity: It filters alert noise, highlights early indicators of compromise and suggests security policy updates.
The aim isn’t to replace expertise. It’s to remove the friction around it.
Pattern recognition > symbolic logic
Modern AI doesn’t “think” like a mathematician, it predicts based on probability across vast datasets. This makes it extremely effective for:
- Time-series forecasting (e.g. demand, usage, spend)
- Document summarisation and classification
- Anomaly detection in telemetry and logs
- Event correlation across seemingly unrelated domains
- Workflow and policy optimisation
It is not effective at symbolic logic or abstract reasoning – and that’s acceptable. Symbolic reasoning is brittle and often fails in real-world, noisy environments. Business data, by contrast, is messy, semi-structured and abundant. That’s where AI shines.
The WALL-E trap: convenience vs capability
There’s a more subtle risk with AI – and it’s not that it underdelivers. It’s that it overdelivers on convenience, to the point where critical thinking erodes.
In Disney’s WALL-E, humans become so reliant on machines that they lose the ability to walk. Business faces its own version of this trap:
- Marketers accepting AI-generated content without strategy or refinement
- IT teams deploying AI-prompted fixes without root cause analysis
- Security teams trusting black-box alerts without investigating threat context
The danger isn’t that AI is “thinking” – it’s that people stop doing so.
Augmented human intelligence: the leadership model
The future isn’t about AI replacing humans. It’s about AI augmenting human judgement, enabling faster action and better decisions.
Here’s what augmented human intelligence (HI) looks like:
- A security analyst sees AI flag access attempts on a CFO’s laptop. She knows the firm is mid-acquisition and recognises that the shared data room has a cloaked URL. AI spotted the anomaly. The human understood the implications.
- AI surfaces a highly ranked lead. A sales rep notices it aligns perfectly with an upcoming product launch. The call gets made.
- An AI model predicts latency in a specific region. An IT architect knows that plant supports a just-in-time production line—and acts before downtime costs accrue.
AI accelerates pattern detection. Humans provide context, strategy and nuance.
The manufacturing moment: momentum vs inertia
In Aryaka’s recent global manufacturing survey, more than 75% of respondents reported hybrid application environments. Yet only 22% had adopted secure, converged networking models like SASE.
The reason isn’t ignorance – it’s inertia.
Those who have moved to secure, modernised infrastructure are already seeing better operational visibility, improved cyber resilience and reduced costs. Those who haven’t remain stuck in “innovation theatre” – talking about transformation while dragging legacy systems behind them.
For GenAI to deliver on its promise, the environment it runs in must be as modern as the models themselves.
Executive playbook: four leadership principles for AI success
To turn AI potential into business outcomes, leaders should focus on four areas:
- Prioritise utility over novelty
Stop asking, “What’s the flashiest use case?” Start asking, “Where are we wasting time—and how can AI reduce that?” - Modernise the foundation
If the network is slow, data siloed or security reactive, no AI tool will succeed. Performance comes before intelligence. - Design for human-in-the-loop workflows
Create systems where AI informs and accelerates decision-making, without removing accountability or oversight. - Stay curious and critical
AI is a tool. Competitive advantage lies in knowing when and how to use it—and when not to.
Be the Wiles, not the WALL-E
Andrew Wiles didn’t need AI to solve Fermat’s Last Theorem. What he brought was the discipline to think, question and build on centuries of knowledge.
That’s the true opportunity today.
AI can get teams 80% of the way to an answer. But the final 20% – the strategic leap, the critical decision, the insight with impact – still belongs to people.
