AI companies are racing to build increasingly powerful systems, but the speed of progress is also raising questions about how societies will manage the technology’s long-term impact. Following this, Anthropic has now announced the launch of the Anthropic Institute, which it claims will focus on studying the societal challenges that could emerge as AI systems become more capable. The company asserts that its latest move will help people prepare for the impact that AI will have on human jobs and other things.
AI progress could accelerate sharply in the next 2 years
Anthropic says the pace of progress in AI has accelerated rapidly since the company was founded five years ago. It took about two years to release its first commercial model, but in the following three years the firm says its systems evolved to a point where they can perform complex professional tasks and even detect serious cybersecurity vulnerabilities.
The company believes the next stage of AI development could move even faster because advances are compounding, meaning each improvement helps speed up future breakthroughs. If that trend continues, Anthropic says the coming two years could bring far bigger leaps in AI capability than many expect, raising new questions for governments and researchers about jobs, economic impact, the values embedded in AI systems, and how the technology should be governed globally.
Interestingly, Anthropic CEO Dario Amodei also recently warned that AI could eventually cause major disruption to the labour market. In earlier comments, he said AI may lead to “unusually painful” changes in employment, particularly across white-collar professions.
“Humanity is about to be handed almost unimaginable power, and it is deeply unclear whether our social, political, and technological systems possess the maturity to wield it,” Amodei wrote in a recent essay discussing the potential risks of advanced AI.
Many jobs still remain difficult for AI to automate, Anthropic reveals
Despite concerns about long-term disruption, Anthropic’s recent research suggests that large parts of the workforce remain relatively insulated from AI automation for now.
The company analysed real-world usage of its AI chatbot Claude to understand how professionals are actually using AI tools in their daily work. The study found that the gap between what AI could theoretically do and what it is currently being used for remains quite large.
For example, roles related to computing and mathematics could potentially see AI assist with a very large share of tasks. However, real usage data shows that only a fraction of those tasks are currently being handled with AI assistance.
Anthropic’s analysis also found that many professions remain difficult for AI systems to automate because they rely heavily on physical work, real-world interaction, or human judgement. Jobs in sectors such as agriculture, construction, transportation, food services, and repair work currently show very low levels of AI exposure.
Even in fields where AI is being used more frequently, including programming, customer service, and data entry, there is still little clear evidence that the technology is causing widespread job losses so far.
Instead, the company says the impact may appear gradually through changes such as slower hiring or automation of repetitive work.
Speaking of which, Anthropic also said the new institute that they have opened will create opportunities for a small team of researchers and analysts. Alongside this, the company is expanding its public policy team and plans to open its first office in Washington DC as it increases its engagement with governments on AI governance.
