{
  "id": 14884281,
  "title": "Will physics be the next field to be turned upside down by AI?",
  "url": "https://urgent.news/2026/10/16/will-physics-be-the-next-field-to-be-turned-upside-down-by-ai",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-16T06:00:00.000Z",
  "source": {
    "name": "New Scientist",
    "slug": "new-scientist",
    "url": "https://www.newscientist.com/article/2593273-will-physics-be-the-next-field-to-get-turned-upside-down-by-ai/"
  },
  "original_language": "en",
  "account": "Recent advancements in artificial intelligence have raised concerns that the field of physics may soon be disrupted by AI, much like how it has impacted mathematics. John Sous, a physicist at Yale University, believes that AI models' reputation for struggling in physics may be unjustified due to flawed testing methods. When Sous and his colleagues addressed these issues, their AI models achieved impressive results, scoring up to 94% on tests designed to evaluate their performance in physics problems.\n\nWhile some aspects of physics can be well-defined and amenable to AI analysis, many areas remain ambiguous and require creative problem-solving. As physicist Vladimir Arnold once stated, mathematical physics is the domain where experiments are inexpensive, but other areas of physics are more complex and less well-defined. It is in these uncharted territories that AI may encounter challenges.\n\nSome experts argue that AI could potentially take over the entire process of theoretical physics, from defining problems to creating models, exploring, testing, and validating results. However, the question remains whether AI can effectively handle the artistic and creative aspects of physics research. Currently, AI is already being employed in physics for various tasks, such as writing computer code, performing mathematical calculations, and even designing and conducting experiments.\n\nYale University's Logan Wright is leading the charge in integrating AI into physics research. His course on using AI in physics research is constantly evolving to keep pace with new developments. Wright emphasizes the importance of creating computer-controlled experiments, where inputs and outputs can be easily managed, allowing AI to explore the problem space and generate meaningful results. Wright acknowledges that there may be domains where AI-designed experiments may exceed human understanding, but he believes that this could be the next frontier in physics research.",
  "summary": "AI has taken huge strides in computer programming and mathematics – physics could be next",
  "key_points": [
    "AI models initially struggled in physics due to flawed testing methods.",
    "Yale University researchers improved AI performance to 94% on physics tests.",
    "AI may handle theoretical physics process but challenges remain in creative aspects."
  ],
  "editors_take": "Refined testing methods suggest that artificial intelligence is moving closer to automating theoretical research, potentially shifting the physicist's role from manual calculation to managing AI-designed experiments.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}