Can’t tech a joke: AI does not understand puns, study finds

The Guardian — Artificial IntelligenceMonday, November 24, 2025 at 7:59:34 AM
Can’t tech a joke: AI does not understand puns, study finds
  • Researchers from universities in the UK and Italy have found that large language models (LLMs) struggle to understand puns, highlighting their limitations in grasping humor, empathy, and cultural nuances. This study suggests that AI's capabilities in comprehending clever wordplay are significantly lacking, providing some reassurance to comedians and writers who rely on such skills.
  • The findings underscore the ongoing challenges faced by AI technologies in replicating complex human cognitive functions, particularly in creative fields where nuanced understanding is crucial. This limitation may impact the development and acceptance of AI in areas traditionally dominated by human creativity.
  • The study reflects broader concerns regarding the reliability of AI systems, as other research indicates that LLMs often fail to generate outputs that align with desired probability distributions and can struggle with logical reasoning. These issues raise questions about the future role of AI in creative industries and the potential implications for professionals in those fields.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps
Continue Readings
Erase and Rewind: Surgically Removing Bias from AI Models
PositiveArtificial Intelligence
A novel technique called Geometric-Disentanglement Unlearning (GDU) has been introduced to surgically remove biases from AI models without the need for complete retraining. This method allows developers to isolate and eliminate the influence of problematic data while preserving the model's overall integrity. The approach treats model updates as movements in a high-dimensional space, effectively enabling targeted adjustments to the model's learning landscape.
AI could replace 3m low-skilled jobs in the UK by 2035, research finds
NegativeArtificial Intelligence
A report by the National Foundation for Educational Research indicates that up to 3 million low-skilled jobs in the UK could be lost to automation and artificial intelligence by 2035, with trades, machine operations, and administrative roles being the most vulnerable.
Pro-AI super PAC Leading the Future launches a $10M media campaign to push Congress to craft a national AI policy that will override a patchwork of state laws (CNBC)
PositiveArtificial Intelligence
A pro-AI super PAC named Leading the Future has initiated a $10 million media campaign aimed at persuading Congress to establish a national AI policy that would supersede the current inconsistent state laws. This initiative reflects the growing influence of the AI industry in shaping legislative frameworks around technology.
Trump orders wide-ranging "Genesis Mission" to boost AI research (Axios)
PositiveArtificial Intelligence
President Trump signed an executive order establishing the 'Genesis Mission' aimed at enhancing artificial intelligence (AI) research and development, with a focus on reducing energy costs for Americans. This initiative represents a strategic move by the administration to promote innovation in AI technology.
How artificial intelligence can help achieve a clean energy future
PositiveArtificial Intelligence
Artificial intelligence (AI) is playing a pivotal role in the transition to clean energy by optimizing power grid operations, guiding infrastructure investments, and aiding in the development of innovative materials. This integration of AI technologies is essential for achieving a sustainable energy future.
AWS to invest up to $50 billion in U.S. AI and supercomputing for government agencies
PositiveArtificial Intelligence
Amazon has announced a substantial investment of up to $50 billion aimed at enhancing AI and supercomputing infrastructure for U.S. government agencies. This initiative is part of a broader strategy to expand its capabilities in artificial intelligence, reflecting the growing demand for advanced technological solutions in federal operations.
AI's Paradoxical Path to New Math: To Find Better Answers, It Needs Less Data and a "Dumber" Brain
NeutralArtificial Intelligence
Recent discussions in artificial intelligence (AI) suggest that to achieve better mathematical solutions, AI systems may require less data and simpler processing capabilities. This paradoxical approach challenges conventional wisdom about data quantity and complexity in AI development.
More than half of new articles on the internet are being written by AI. Is human writing headed for extinction?
NeutralArtificial Intelligence
More than half of new articles on the internet are now being generated by artificial intelligence (AI), raising concerns about the future of human authorship in writing. The increasing sophistication of AI technology has blurred the lines between human and machine-generated content, making it challenging to discern the source of written material.