What Might Come Next

Predicting AI's future is a mug's game - the field has consistently surprised experts in both directions, advancing faster than expected in some areas and slower in others. That said, several trends seem likely in the near term. Models are getting more capable at reasoning, not just generating fluent text. Multimodal systems that work across text, images, audio and video are becoming standard. AI agents - systems that can take actions in the world, not just produce text - are an active area of development. On the hardware side, the race to build more efficient AI chips is intensifying, which could make powerful AI more accessible and cheaper to run. Regulation is coming, with the EU's AI Act already in force and other jurisdictions developing their own frameworks. The biggest uncertainty is whether current approaches hit a ceiling or continue to improve. Some researchers believe we need fundamentally new ideas to progress further; others think scaling existing methods will keep working. For most people's purposes, the practical advice is straightforward: pay attention to what AI can do today, experiment with it, but plan for a world where the specific tools and capabilities change significantly every twelve to eighteen months.