A year ago, the big question about artificial intelligence was whether it would replace us. Now, the conversation has changed. AI is clearly here to stay. The more pointed and uncomfortable question is who is actually benefiting?
Behind the flashy demonstrations and promises of productivity, new data suggests that the economic benefits of AI are not being shared equally. In fact, it may reinforce existing disparities between the rich and the poor, the skilled and the unskilled, the well-resourced and the under-prepared.
What the data says when you strip away the hype
In mid-January 2026, AI company Anthropic released an economic indicators report. This is a rare study that looks at how people actually use AI tools, rather than how companies declare they will use them. This report analyzed millions of conversations with Anthropic’s chatbot, Claude, across a variety of countries and job categories.
One discovery immediately caught my eye. There is a big divide between how AI is used behind the scenes and how it is used by individuals. Companies that use AI through APIs mostly automate background tasks. Individual users of Claude’s website tend to work collaboratively with AI, using it as a thinking partner rather than a replacement.
This supports a growing idea in the tech world. Automating tasks is easy. Using AI for judgment, strategy, and purpose still requires human involvement.
The uncomfortable truth about skills and access
The most troubling insight from the data is how closely the AI’s output reflects the quality of the user input. Simply put, people who ask better and more complex questions get better answers. People who don’t do it don’t do it.
Anthropic frames this as AI meeting people where they are. A more serious view is that AI is fixed in skills. If you can’t clearly explain the problem, you won’t get the most value out of the tool. Technology does not uplift everyone equally. Reflect existing capabilities to users.
For South Africa and much of the continent, this is important. The report shows that AI adoption is strongly linked to national income. In wealthy countries, AI is used more often and with more nuance. They treat it as a collaborator. In low-income countries, it is more likely to be used to support classes and basic support than to promote business.
The old idea that AI would allow emerging markets to leapfrog richer economies suddenly began to waver.
Backlash from the Global South
Not everyone agrees with this gloomy interpretation. At Davos last week, India’s technology minister, Ashwini Vaishnaw, publicly challenged the idea that leadership in AI lies only with countries that build models at scale.
His argument was frank. The real benefits come from applying existing models to real business problems, not from building ever-larger systems. According to him, most of the economic value lies in the application layer, where companies deploy AI to understand local needs and increase productivity.
He pointed out that India already ranks near the top of the world in terms of AI talent and overall vibrancy. The meaning is clear. Countries like India and potentially South Africa do not need to win the model race to win economically. They need to win by executing.
Whether this optimism can be sustained beyond the conference stage remains an open question.
African venture capital issues
Similar tensions can be seen in technology financing in Africa. Investor Steven Deng recently argued that African startups are stuck telling outdated stories. For many years, funding relied on two ideas: Africa’s youth population and Africa’s potential for social impact.
Those stories brought in money. Now, Deng Xiaoping says they are losing power. Funding through stocks is stagnant. Pre-seed investing is drying up. Deals focused on debt and climate change dominate, but early-stage risk capital has retreated.
His call is for impatient capital that demands scalable global outcomes, not slow demographic rewards. It’s a convincing diagnosis. It is also, inevitably, shaped by the needs of the next person to raise money.
No one is building the missing middle
The intersection of the AI data and venture capital debates is where many founders quietly realize that there is a missing middle ground. These companies are too small for enterprise AI systems, too cluttered for modern data stacks, and lacking the resources to benefit from sophisticated AI tools that rely on clean data and skilled prompts.
AI promises transformation, but only for those who already have the capacity to unleash it. Venture capital promises growth, but only for startups that fit a narrow story.
Some startups are trying to build tools that are not perfect and work in chaos. It remains to be seen whether these approaches will scale up.
Read the incentive behind the story
None of the participants in this discussion are neutral. AI companies want to prove usefulness and adoption. Investors need fresh stories to enable funding. A startup needs a problem big enough to justify its solution.
That doesn’t mean any of them are wrong. That means their claims deserve scrutiny.
What remains behind the hype is a quiet reality. The benefits of AI are real, but they don’t come automatically. They tend to flow to people who are educated, articulate, and already empowered.
The rest of the world is still waiting for stories and systems that fit the facts on the ground.
Follow Joburg ETC on Facebook, Twitter, TikTok and Instagram
For more Johannesburg news, visit joburgetc.com.
Source: IOL
Featured Image: Eric Huberman
