Large models, although impressive, are losing efficiency, and progress will come from smarter design and better training approaches. Companies are moving towards smaller, specialized models that are cheaper and faster, facilitating the use of local AI on devices. AI agents, which promised autonomy, are starting to become functional through standards that link them to real tools. Additionally, AI will learn not only from text but also from interactions with the world, having a significant impact in virtual environments. Instead of taking jobs, AI will support people, generating new roles around oversight and data governance. Advances in small models and edge computing will bring AI into everyday life, but progress in robotics will be slower.
In conclusion, 2026 will bring a more integrated, reliable, and truly useful AI.