MIT Technology Review editor outlines five major AI themes midway through 2026, from job uncertainty to real-world harms and scientific potential.

Generative AI tools are now used by millions to automate everyday office tasks, but there is almost no data on what effect this technology will have on employment and the economy overall, despite hype about AI joining the workforce soon.

Near-term, real-world AI harms have materialized: deepfakes (98% of which are pornographic and 99% involve women) have been used to incite violence and swing votes; chatbots have faced multiple lawsuits alleging they encouraged or aided suicides and self-harm; and military chatbots can now give tactical advice, raising risks in high-stress active conflict.

AI for science shows genuine potential, with Google DeepMind developing Co-Scientist to help researchers compare results and devise experiments, and OpenAI stating its North Star is building a fully automated researcher by 2028; however, scientists warn overreliance on AI tools could narrow research scope and lead to inaccurate or fake results.

Public backlash is growing: protests against AI are becoming more organized and drawing larger crowds; the 2025 game Clair Obscur was stripped of an award when developers admitted to using AI; and activists are managing to stall data center development in multiple places over environmental and electricity cost concerns.