For the past two years, the public conversation around AI has been dominated by one question: will it replace jobs?
It’s an understandable worry. Tools like ChatGPT and Claude can write, research, summarise, code and synthesise information at a speed that seemed impossible only a few years ago. It hasn’t helped that both Sam Altman of OpenAI and Dario Amodei of Anthropic fuelled the anxiety – Amodei warned AI could wipe out half of all entry-level white-collar jobs, while Altman said entire entry-level categories were at severe risk.
But the reality inside many organisations is very different, and even Altman and Amodei have shifted. Altman recently admitted he was “pretty wrong” about AI eliminating entry-level roles, while Amodei now suggests automation will expand human workloads rather than destroy them.
In our work across public and private sector entities, a surprising contradiction keeps surfacing. Employees who have completed foundational AI training often still lack access to the basic tools needed to apply it. Some arrive at advanced courses having done little with AI since their previous training – not because they didn’t want to, but because they weren’t given the tools or the time to experiment.
That points to a challenge that receives far less attention than automation: most businesses are still struggling with AI enablement, creating a massive capability overhang — the gap between what modern AI systems can theoretically do and how they are actually deployed. We talk about AI as if every business operates at the frontier. In reality there is a vast chasm between those that are truly AI-native and those that are not, and it is far larger than many IT leaders realise.
The assistant versus replacement debate
Modern AI models are undeniably impressive — the model you are using now is the worst it will ever be. Viewed in isolation, that can feel threatening, but capability alone does not equate to replacement.
AI can automate repetitive tasks and assist with writing, coding, analysis and research. What it cannot do is provide organisational context, align stakeholders, exercise judgment, understand politics, manage relationships or make decisions on behalf of a business. It can write code better than almost anyone alive, but without a person prompting, directing, evaluating and validating the output, it accomplishes nothing meaningful on its own. The missing ingredient is human judgment and context.
A more useful way to think about AI is as an accelerator, not a replacement. When one part of a process becomes dramatically faster, the bottleneck simply moves elsewhere. Research that once took days can take minutes – but the information still needs to be evaluated, interpreted, socialised and acted upon.
Software development shows the pattern repeatedly. AI helps teams build initial versions far faster, but what follows is a flood of feedback, change requests, testing and new ideas. The faster teams move, the more opportunities they create, so work often grows rather than shrinks. Many organisations find that if a team can deliver two projects instead of one, the response is rarely to work half as much – it’s to take on more. The principle isn’t doing more with less, but doing more with the same. That changes jobs; it doesn’t necessarily remove them.
The real risk may be at entry level
That said, there are workforce implications, and entry-level roles deserve the most attention. Many professions rely on apprenticeship models where junior staff build competence through repetition. If AI strips out routine junior tasks such as initial research or team administration, fewer entry-level positions may be needed.
Talk to an intern and they’ll grasp AI-assisted work but ask a telling question: “Without vast practical experience, how do I know if the answer is right?” They’d be spot-on. AI may democratise access to expertise, but experience still matters – arguably more, because someone must determine whether the machine’s answer is correct.
The shadow AI problem
Organisations trying to control AI adoption through policy alone are setting themselves up for disappointment. Shadow AI already exists: employees use these tools whether or not companies formally approve them, and many simply don’t disclose it.
Research has identified an “AI disclosure penalty” – where employees who admit to using AI are seen as less capable, less hardworking or less deserving of recognition. That creates a dangerous dynamic in which people keep using the tools but become less transparent and look for workarounds. The answer isn’t stricter policy, but environments where people can experiment safely, learn openly and discuss AI use without stigma.
What AI skills mean
Perhaps the most interesting question HR teams now ask is what AI skills actually are. The answer is probably not prompt engineering. The most valuable capabilities are surprisingly human: defining the right problem, providing context, evaluating outputs critically, using data responsibly, applying judgment and – perhaps most importantly – working transparently.
These are what determine whether AI creates value or simply generates more content. The businesses that benefit most over the next decade will not always be those with the most advanced models. They will be the ones that give employees access to the tools, time to experiment, guardrails to use them safely and a culture that encourages transparency.
Before worrying about whether AI will replace workers, many companies have a more immediate challenge to solve: enabling them.
About iqbusiness
iqbusiness is a digital integrator that transforms businesses, public sector entities and other organisations as your go-to for consulting and technology. With more than 27 years of experience, and led by some of the continent’s best thinkers and doers, its purpose is simple: to grow people, business and Africa as one.
Its scale and capabilities unlock value and global growth for clients, from intent to impact, through five value streams: digital experience and strategy; strategic consulting and innovation; business performance and delivery; intelligent applications and platforms; and technology managed services.
iqbusiness, including technology division iqx, forms a key part of the ICT segment of JSE-listed industrial group Reunert. It is a level-1 B-BBEE contributor driven by its GESHIDO energy, and is consistently recognised as a top employer and leading consulting firm. Established in South Africa in 1998, iqbusiness integrated into Reunert ICT in 2023 and merged with +OneX in 2024. For more information, visit www.iqbusiness.net.
*GESHIDO® = we get $#!% done.
- The authors, Biase De Gregorio and Morgan Goddard, are, respectively, partner leading AI advisory and enablement and partner leading software development, both at technology and management consultancy iqbusiness
- Read more articles by iqbusiness on TechCentral
- This promoted content was paid for by the party concerned
