Artificial intelligence (AI) is rapidly evolving from a workplace assistant into an autonomous digital worker capable of planning projects, writing software, browsing the internet, and making decisions with minimal human supervision, raising fresh concerns that millions of white-collar jobs could be disrupted sooner than many expected.

That is the warning from the Preliminary Report of the Independent International Scientific Panel on Artificial Intelligence, released by the United Nations in July 2026. The report says AI has entered a new era of agentic AI, systems that do far more than answer questions; they perform complex tasks independently or alongside other AI systems.

“AI is moving from systems that generate outputs and dialogues towards systems that act,” the report says.

It explains that agentic AI can browse the web, use software tools, make decisions, execute code, manage and work with other agents, and operate entire computers with increasing autonomy, requiring far less human oversight than previous generations of AI.

The report suggests this technological shift could fundamentally reshape the future of work across industries, from journalism and accounting to law, software engineering and customer service.

Unlike earlier AI systems that assisted workers, agentic AI can independently complete multi-step assignments such as researching information, analysing data, writing reports, generating computer code and coordinating workflows across different software platforms.

Its capabilities are also improving at a remarkable speed. According to the report, benchmark tests show that the complexity of software engineering tasks AI agents can complete autonomously has been doubling every 4.6 months, while overall frontier AI capabilities are improving nearly twice as fast as they were just a year ago.

The report says AI developers are already relying heavily on the technology themselves.

“AI developers are reportedly using AI to generate 75% of their new code,” it notes, warning that this creates a feedback loop in which AI increasingly accelerates its own development.

For Kenya, the findings have significant implications as businesses continue investing in automation and digital transformation.

Unemployment fears

Occupations built around repetitive cognitive work, including journalists, accountants, lawyers, customer service representatives, call centre workers, software developers, researchers and administrative professionals, are expected to experience the greatest disruption as organisations adopt AI agents capable of performing routine knowledge-based tasks around the clock.

However, the UN panel stresses that AI will not automatically replace entire professions. Instead, it is more likely to transform specific tasks within jobs, allowing organisations to redesign workflows while increasing demand for workers with digital, analytical and supervisory skills.

“AI technology → access → adoption → diffusion → economic outcomes,” the report explains, noting that economic benefits depend not simply on access to AI but on complementary investments in skills, institutions, data infrastructure and organisational change.

The report also cautions against assuming that productivity gains will happen overnight. Like electricity and computers before it, AI requires businesses to redesign processes, retrain employees and adapt management systems before significant economic gains are realised.

The findings are likely to intensify debate among employers, labour economists and policymakers in Kenya. The Federation of Kenya Employers (FKE), recruitment firms and labour market analysts have increasingly pointed to AI as both a source of productivity and a catalyst for workforce transformation, particularly in business process outsourcing, financial services, legal practice and media.

Despite concerns over job displacement, the report argues that AI also creates opportunities by enabling new AI-native businesses, improving scientific research, supporting healthcare and increasing productivity across sectors.

The challenge for governments, it says, is ensuring workers are equipped with the skills needed to adapt.

“Artificial intelligence has shifted from passive pattern recognition towards active reasoning and autonomous action,” the report observes.