Greene described the problem as a dangerous feedback loop in which large language models absorb machine-generated material and reproduce it as authoritative information. “LLMs are scanning this junk, and then that becomes truth,” he said during a public discussion about artificial intelligence and digital culture. The phenomenon, he added, resembles “a race to the middle of sh*t,” where originality and accuracy are replaced by repetitive algorithmic output.
His comments reflect widening unease across the technology sector about the reliability of generative AI systems that rely heavily on scraping the open web for training data. Tools capable of producing convincing articles, code and imagery have expanded rapidly since the release of large-scale models by companies such as OpenAI, Google and Anthropic. While these systems promise productivity gains and creative experimentation, critics argue they may accelerate the spread of misinformation and dilute the value of authentic human-produced material.
Greene questioned how users can trust information generated by systems that openly acknowledge the need for verification. “How can you trust stuff that says at the bottom you need to fact-check all the answers I’m giving you?” he asked, highlighting what he sees as a paradox in the current generation of AI tools.
Concerns about the quality of machine-generated text have become more prominent as AI-assisted writing spreads across blogs, marketing websites and social media platforms. Researchers studying the online information ecosystem warn that automated content can quickly multiply, creating what some analysts describe as “model collapse,” a situation where AI systems trained on synthetic data gradually lose accuracy and diversity.
Studies in machine learning suggest that large language models perform best when trained on carefully curated human-produced datasets. When models are repeatedly retrained on content generated by earlier models, statistical patterns can become distorted. Over time this may lead to degraded outputs, reinforcing errors and narrowing the range of ideas present in the training data.
Technology companies have begun experimenting with methods to mitigate the risk. Some firms are exploring techniques to watermark AI-generated content or detect synthetic text in order to preserve transparency. Others are investing in curated training datasets that prioritise high-quality sources such as academic literature, licensed media archives and verified public records.
Greene’s remarks arrive amid an intense global debate about the broader social impact of generative AI. Governments across Europe, North America and Asia are developing regulatory frameworks aimed at improving transparency and accountability in automated systems. The European Union’s Artificial Intelligence Act, for example, sets disclosure requirements for AI-generated content and establishes oversight rules for high-risk applications.
Media organisations have also responded to the surge in AI-produced material. Many newsrooms now restrict the use of generative tools in editorial workflows or require journalists to verify any automated output before publication. Editors and researchers argue that editorial standards developed over decades of reporting remain essential in an era where algorithms can generate text at scale.
Digital platforms face similar challenges. Social media networks are experimenting with labelling systems that identify AI-generated images or posts, while search engines are attempting to reduce the visibility of low-quality automated pages designed primarily to manipulate rankings.
Despite his concerns, Greene acknowledged that artificial intelligence can still play a constructive role in creative industries when used responsibly. Developers and artists increasingly employ AI tools for tasks such as concept design, data analysis and game development, often blending algorithmic assistance with human decision-making.
Greene himself has spent years experimenting with ambitious digital worlds through his studio PlayerUnknown Productions, including research into large-scale virtual environments and simulation technology. His work on PlayerUnknown’s Battlegrounds helped popularise the battle-royale genre and influenced a generation of multiplayer games.
Follow Arabian Post
Select Arabian Post as your preferred source on Google and MSN News for trusted business news and Arab politics and updates.
