AI, agents, and the human-centred future of skills

Let’s get one thing straight:
We are not in the age of Artificial General Intelligence (AGI). 

Not yet. But you wouldn’t know it from the marketing. 

But until recently I have not known what I am articulating… I have taken it upon myself to read, ask questions, dive in – so that I have a slightly better understanding of where we are at in the Age of AI. I am no expert. 

Every platform today is suddenly “AI-powered,” “skills-enabled,” and “agent-driven.”
Which sounds exciting, until you deploy it and realise nothing has changed.  

Why? Because technology doesn’t fix broken foundations. And 70% of HR tech implementations still fail to deliver the value they promise. 

Let’s unpack why: with a little help from The Coming Wave, the brilliant and urgent book by Mustafa Suleyman. 

We throw around the term AI like it’s one monolithic thing. But in reality, we’re operating across a spectrum, something Suleyman outlines clearly in his book: 

• AI (Artificial Intelligence):
  Systems trained to do specific tasks: like making predictions, recognising patterns, or summarising data. Think: ChatGPT or a spam filter. 

• ACI (Artificial Capable Intelligence):
  AI that can perform complex tasks across multiple domains with growing autonomy. Not general, but not narrow either. Think: copilots, assistants, agents. 

• Agentic AI:
  These are AI models that not only analyse and recommend, they can act. They operate semi-independently to complete tasks, orchestrate processes, or even make decisions. 

• AGI (Artificial General Intelligence):
  Theoretically, this is the holy grail. Human-level intelligence across all domains, with self-awareness and the ability to learn any task a human can. As Suleyman reminds us: “We are not there yet, but we are building the staircase.” 

And that distinction matters. Because all the AI we use today, even the so-called “agents”, still rely entirely on human-created data. 

If the data’s bad, the output’s bad. If the foundations are a mess – poor job architecture, siloed systems, missing context – the results will be… faster mess. 

You can’t sprinkle AI over dysfunction and expect magic. Yet that’s what we’re seeing: 

• AI-powered learning platforms built on outdated job frameworks 

• Talent marketplaces making recommendations that conflict with internal mobility rules 

• Inference engines generating skills profiles that are unusable because no one agreed on what “skills” even means 

It’s not that the AI is broken. It’s that the human systems underneath it were never built for this. 

AI is only as good as: 

1. The data it’s trained on 

1. The humans who shaped that data 

1. The systems it’s deployed into 

Miss one of those? You’re not transforming. You’re accelerating chaos. 

Let’s zoom out. 

One of Suleyman’s most compelling arguments is that we’ve misunderstood the scale and nature of technology. We tend to think of it as gadgets: apps, phones, wearables. 

But in truth, technology includes: 

• The wheel: one of the first general-use innovations. It enabled agriculture, trade, transport, and architecture. 

• The printing press: a leap for literacy, science, and democracy, as well as a precursor to centuries of upheaval and change. 

• The power loom: introduced in 1785, allowing a single child to do the work of multiple adult weavers. Wages dropped. Riots followed. But the technology didn’t go away. Society had to adapt. 

• The internet: arguably the most transformative enabler in modern history. It underpins everything from global commerce to education to surveillance… and of course, the rise of social media. 

As Suleyman puts it: 

“Technology is a hyper-object, not a niche. It is a force that defines human existence.” 

And now we enter the AI wave. 

AI will absolutely reshape the future of skills, talent, and work. 

But not because it replaces us. Because it forces us to confront what we’ve ignored: 

• Misaligned architectures 

• Siloed tech stacks 

• Data that isn’t trusted 

• A lack of shared language across the talent lifecycle 

If we don’t fix these things, AI will only expose them faster. But if we do? 

Then we can build something extraordinary. 

The temptation is to let go. To hand over the reins to the machine. To hope the tech figures it out. 

But AI can’t design your workforce strategy. It can’t define your operating model. It can’t build relationships, resolve tension, or lead your people. 

That’s our job. 

So let the agents do what they’re good at. But let us stay in the loop. To guide, contextualise, challenge, and make meaning. 

Because the future of work won’t just be artificial. 

It will be, it must be, deeply, intentionally, beautifully human.