Artificial intelligence (AI) is spreading rapidly across industries, but assessments of its employment effects and social impact remain mixed.

On June 9, MIT Technology Review organised the key mid-2026 trends around AI into five pillars: jobs, safety, social backlash, scientific research and technology diffusion.

First, the impact on jobs remains uncertain. Generative AI tools are already used by millions of people to automate routine office work such as drafting emails, organising documents and researching materials. But there is still not enough data showing how AI will affect employment and the broader economy.

Some forecasts say AI agents could split up tasks and turn office work into an assembly-line process. Companies are still at the stage of figuring out how to apply AI to internal work. It said it remains to be seen whether AI will replace some roles, become a tool that boosts productivity, or create new ways of working.

Safety issues, by contrast, are becoming more concrete. Deepfakes have been used as tools to incite violence or fuel political distrust, and concerns have grown that they could also influence election opinion. Studies show many deepfakes are concentrated on sexually exploitative content, and victims are disproportionately women.

A controversial dependence between chatbots and users is also emerging. Many people seek personal advice or emotional comfort from chatbots, but some AI companies face lawsuits over safety problems that occurred during chatbot use. As a result, a key issue is how far AI can respond in sensitive situations and when it should connect people to human experts or institutions.

AI use in the military is also adding to concerns. Large language models (LLMs) can be used beyond simple analysis as decision-support tools. Observers say if military operations move to a stage where AI organises information and provides advice, questions of accountability and control mechanisms become more important.

Social backlash is also growing. In London, anti-AI protests have become more organised, and film and game fans have continued to oppose the adoption of generative AI. Some cases have shown controversies growing simply because AI was used in part of the creative process. The trend reflects concerns that AI could undermine creators’ rights and the value of labour.

Environmental burdens from expanding AI infrastructure are also an issue. In the United States, thousands of data centres are in operation, and rising AI demand is increasing electricity use. Complaints are also growing about higher local power bills, strain on power grids and environmental impacts.

AI use in scientific research is creating both expectations and concerns. Google DeepMind developed an AI scientific tool that compares existing research results, generates hypotheses and helps with experimental design. OpenAI has also set out a goal of creating a fully automated researcher by 2028.

Some scientists, however, warn that reliance on AI could narrow research topics or churn out inaccurate results in large volumes. AI can speed up research, but there are concerns that rapidly spreading unverified results could damage scientific trust.

AI has already spread through daily life and across industries, but its direction is still not clear. AI companies are holding up artificial general intelligence (AGI) and signalling massive change, but the speed and form of real change have not been determined.

The outlet stressed that even if AI produces human-like results, it remains a technology. It said AI could create transformative change like electricity or the internet, but it could take time to take hold across society.

Ultimately, the core of the current AI debate is not confined to either expectation or fear. The jobs effect remains uncertain, but safety issues, social backlash and use in scientific research have already become concrete realities. Whether AI becomes the next general-purpose technology will depend not only on the speed of technological advances but also on how society sets up controls, regulation and accountability structures.