Consumer fatigue with low-quality generative output—dubbed “AI slop”—is pushing the industry to look beyond chatbots in 2026. Researchers and tech giants are pivoting to world models that learn from video and simulation to reason about physics and cause-and-effect, promising advances in robotics, video, and industrial “digital twins.” Europe is embracing smaller, on-device language models to cut costs, conserve energy, and limit reliance on U.S. cloud providers, even as American firms double down on hyperscale data centers and ever-larger models, fueling talk of an AI investment bubble.
Safety and governance are emerging fault lines. Reports of harmful chatbot behavior and rising concern over user vulnerability coincide with a U.S. shift to pre-empt state-level AI rules, intensifying political and regulatory uncertainty. Advocates expect broader public pushback and calls for stronger standards. Meanwhile, competitive pressure remains fierce: Google’s Gemini 3 has accelerated OpenAI’s push toward GPT-5; prominent researchers including Yann LeCun and Fei-Fei Li are launching ventures in world models; and Chinese firms such as Tencent are entering the space, underscoring a global race to define the next era of AI.
AI in 2026: World models rise, small models spread, and ‘AI slop’ backlash stokes bubble fears
- Trending
- Comments
- Latest
