Artificial Intelligence
Who really benefits from AI and who is being left out
A year ago, the big question around artificial intelligence was whether it would replace us. Now the conversation has shifted. AI is clearly here to stay. The sharper, more uncomfortable question is this: who is actually gaining from it?
Behind the glossy demos and productivity promises, new data suggests that AI’s economic upside is not evenly shared. In fact, it may be reinforcing existing divides between rich and poor, skilled and unskilled, and well-resourced and underprepared.
What the data says when you strip away the hype
In mid-January 2026, AI company Anthropic released its Economic Index report, a rare look at how people actually use AI tools in real life rather than how companies say they will. The report analysed millions of conversations with Claude, Anthropic’s chatbot, across different countries and job types.
One finding stands out immediately. There is a sharp split between how AI is used behind the scenes and how it is used by individuals. Businesses using AI through APIs mostly automate background tasks. Individual users on Claude’s website tend to work alongside the AI, using it as a thinking partner rather than a replacement.
This supports a growing idea in tech circles. Automating tasks is easy. Using AI for judgement, strategy, and purpose still needs humans in the loop.
The uncomfortable truth about skill and access
The most troubling insight from the data is how closely AI output mirrors the quality of what users put in. In simple terms, people who ask better, more complex questions get better answers. Those who do not, do not.
Anthropic frames this as AI meeting people where they are. A harsher reading is that AI is skill-locked. If you cannot articulate a problem clearly, you cannot unlock the full value of the tool. The technology does not lift everyone equally. It reflects the existing ability back to the user.
For South Africa and much of the continent, this matters. The report shows that AI adoption is strongly linked to national income. Wealthier countries use AI more often and with more nuance. They treat it as a collaborator. Lower-income countries are more likely to use it for coursework help and basic assistance rather than business acceleration.
The old idea that emerging markets will leapfrog richer economies through AI suddenly looks shaky.
A pushback from the Global South
Not everyone agrees with this bleak reading. At Davos last week, India’s technology minister, Ashwini Vaishnaw, openly challenged the idea that leadership in AI belongs only to countries building massive models.
His argument was blunt. Real returns come from applying existing models to real business problems, not from building ever larger systems. According to him, most economic value lives in the application layer, where companies understand local needs and deploy AI to boost productivity.
India, he noted, already ranks near the top globally for AI talent and overall vibrancy. The implication is clear. Countries like India and potentially South Africa do not need to win the model race to win economically. They need to win at doing.
Whether this optimism holds up beyond conference stages is still an open question.
The African venture capital problem
A similar tension appears in African tech funding. Investor Stephen Deng recently argued that African startups are stuck telling outdated stories. For years, funding leaned on two ideas: Africa’s young population and Africa’s social impact potential.
Those narratives brought in money. Today, Deng says they are losing power. Equity funding has stalled. Pre-seed investment is drying up. Debt- and climate-focused deals dominate, while early-stage risk capital retreats.
His call is for impatient capital that demands scalable global outcomes rather than slow demographic payoffs. It is a compelling diagnosis. It is also, inevitably, shaped by the needs of someone raising their next fund.
The missing middle nobody is building for
Where the AI data and the venture capital debate meet is in what many founders quietly recognise as the missing middle. These are businesses that are too small for enterprise AI systems, too messy for modern data stacks, and too under-resourced to benefit from polished AI tools that assume clean data and skilled prompts.
AI promises transformation, but only for those already equipped to extract it. Venture capital promises growth, but only for startups that fit a narrow story.
Some startups are trying to build tools that work in chaos rather than perfection. Whether these approaches scale remains to be seen.
Reading the incentives behind the stories
None of the players in this debate are neutral. AI companies want to prove usefulness and adoption. Investors need fresh narratives to unlock funding. Startups need a problem big enough to justify their solution.
That does not make any of them wrong. It does mean their claims deserve scrutiny.
What remains after the hype is a quieter reality. AI’s benefits are real, but they are not automatic. They tend to flow toward the educated, the articulate, and the already empowered.
The rest of the world is still waiting for a story and a system that fit the facts on the ground.
Follow Joburg ETC on Facebook, Twitter, TikT
For more News in Johannesburg, visit joburgetc.com
Source: IOL
Featured Image: Erik Huberman
