There’s a lot of noise out there these days about AI.
Depending on who you listen to, we’re either days away from a historic economic bubble bursting… or from humanity being gently replaced by our new silicon overlords.
Somewhere between doomscrolling and utopian PowerPoints, the real story of AI gets lost — buried under brands, buzzwords, and a never-ending parade of “AI-powered” everything.
So here’s a sober, slightly provocative look at where things actually stand.
AI Adoption Is Everywhere… Kind Of
According to McKinsey’s State of AI report, nearly 88% of organisations now claim they use AI in at least one business function. Companies are experimenting with “AI agents,” hoping multi-step workflow automation will solve their operational headaches.
Adoption? Undeniable.
Impact? Less clear.
Once you look below the surface, things are far less glamorous.
Sources: McKinsey State of AI 2024 https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2024
The Quiet Reality: Most Organisations Still Don’t Know How to Use AI
The corporate world has become excellent at announcing AI initiatives.
Turning them into impact is another story.
Studies from BCG and MIT Sloan converge on a consistent pattern:
~74% of companies struggle to extract real value from AI (BCG)
Only 3–6% achieve enterprise-scale, repeatable success
Most remain stuck in endless proofs of concept
It’s the organisational equivalent of buying a treadmill and assuming that alone makes you fit.
Sources:
MIT Sloan Management Review & BCG, The State of AI in Business
BCG AI Value Study (https://www.bcg.com/publications/2023/ai-value-realization)
Why the Gap Between Hype and Reality Is So Wide🤖
The biggest myth:
AI is a technology problem.
It isn’t.
Models are rarely the bottleneck.
The roadblocks are almost always human and organisational:
Fragmented or low-quality data
Legacy systems allergic to change
Vague objectives (“Let’s do something with AI”)
Teams unprepared to shift their workflows
Leaders expecting magic instead of transformation
MIT Sloan’s research shows that high-performing organisations spend 70% of their AI effort on people, processes, and data foundations, not models.
AI doesn’t fail because it’s unintelligent.
It fails because the organisation isn’t prepared for intelligence.
The Extremes: Apocalypse vs. Bubble
Public discourse swings between two cinematic extremes:
1. “AI will wipe us out.”
2. “The AI bubble will burst and drag half the economy with it.”
Both are easy to sell.
Neither reflects the messy, slower-moving reality.
The existential-risk narrative assumes capabilities we don’t have.
The bubble narrative ignores that enterprise AI is a boring, incremental grind.
Reality sits somewhere in between:
No apocalypse, no utopia — just uneven, friction-filled transition.
A New Concern: 🧠 Cognitive Offloading and AI Dependence
One risk gaining attention in MIT’s work: cognitive offloading — outsourcing memory, decision-making, or judgement to machines. Early results show:
Heavy AI users often become less confident in their own reasoning
Reliance on AI can lead to shallower understanding
People treat machine confidence as accuracy, even when wrong
This aligns with a stark anecdotal example:
A team adopted 17 AI tools to boost productivity… and became 35% less productive.
Too much “assistance” can turn into coordination chaos.
Source: “I adopted 17 AI tools…”
https://generativeai.pub/i-adopted-17-ai-tools-to-boost-productivity-my-team-became-35-less-productive-12b45f0b2f15
Cyber Threats and Fake Conviction
AI is also supercharging digital risks:
Deepfakes that impersonate voices and faces
Hyper-personalised phishing written in perfect, context-specific language
Automated misinformation that spreads faster than fact-checkers can blink
And then there’s the strangest emerging threat:
AI systems hallucinating — and other AIs repeating those hallucinations as if factual, turning fiction into “machine-validated truth.”
This isn’t science fiction.
It’s happening now, and regulators are scrambling to keep up.
💡 What Actually Works: A Playbook for Real AI Adoption
Across organisations that do make AI work, a few patterns repeat:
1. They measure value, not vibes.
If it doesn’t improve speed, cost, accuracy, or satisfaction, it’s theatre.
2. They fix the data first.
Bad data + smart AI = confident nonsense.
3. They redesign workflows, not just add tools.
AI on top of chaos creates… accelerated chaos.
4. They avoid tool hoarding.
More AI ≠ better AI.
As another referenced article shows, simply feeding your entire life into a model doesn’t guarantee insight.
Source: “I fed Claude 7 years of daily journals…” https://medium.com/swlh/i-fed-claude-7-years-of-daily-journals-it-showed-me-the-future-of-ai-2c13a8d18ef9
5. They stay critical, not cynical.
Optimistic, but grounded.
🪓 Maybe the Real Headline Should Be:
AI Isn’t a Revolution. It’s a Better Chisel.
Powerful? Yes.
Transformative? Potentially.
Magic? No.
AI is a tool — a sharp, accelerating one — but still a tool.
In skilled hands, it reshapes industries.
In careless ones, it mostly reshapes budgets and attention spans.
The story of AI isn’t about replacing humans.
It’s about amplifying them… and occasionally distracting them into oblivion.
📎 The Meta Part (Yes, I Used AI to Write This)
Writing this would normally take hours of research, synthesis, and structuring.
With AI, it took minutes — not because AI replaced the thinking, but because it removed the friction.
That’s the real promise:
AI speeds us up, but only if we stay awake at the wheel.
So stay curious.
Stay critical.
And most importantly: stay safe — double-check your sources.
Because in a world where machines speak with perfect confidence, your scepticism may be your most valuable skill. Thank you for reading.
Go Back
