Sam Altman warns the “dead internet theory” may soon come true — bots and AI like ChatGPT could kill the web in 3 years

by

The advancements in generative AI have moved past being just tools for creating visuals from text instructions; they are now making tangible differences in various sectors like healthcare, education, technology, entertainment, and beyond. Yet, these advancements can be seen as a double-edged sword, as they bring about both constructive and potentially concerning outcomes.

On one side, this advanced tech holds immense potential in healthcare, particularly in identifying early symptoms of conditions like dementia and cancer, allowing for earlier intervention and treatment before the diseases progress. However, there’s also a concern that this technology could pose an existential risk to humanity, given a high likelihood (approximately 1 out of every 100,000 instances) of leading to human extinction.

Demis Hassabis, the CEO of Google DeepMind, believes that Artificial General Intelligence (AGI) might become a reality in the near future. However, he expresses concern that our society may not be adequately prepared to manage all its implications. He confides that these possibilities often keep him awake at night.

In essence, predictions indicate that the ‘dead internet’ hypothesis may materialize within the following three years, given the increasing dominance of AI-produced content over user-generated content. This theory postulates that a substantial portion of online activities is automated by bots or AI, and this manipulated content is often curated algorithmically to exert control over populations and diminish natural human engagement. (Paraphrased from the original article in Popular Mechanics)

It’s quite troubling to note that a report titled “Bad Bot” published by cybersecurity firm Imperva in 2024 estimated that roughly half of all web traffic is artificially generated using AI. This percentage has significantly increased since 2021, with AI-generated internet traffic jumping from 42.35% to 49.6%.

According to current patterns, it’s anticipated that a significant portion of internet traffic and content will be generated by bots and AI. Over the last few years, prominent news outlets have significantly reduced their workforce and opted for AI replacements instead. Notably, a study from Pew Research Center indicates that approximately 38% of websites created by humans in 2014 no longer function due to an issue known as “link decay.

In the previous year, researchers from Amazon Web Services (AWS) proposed that approximately half of all content found online is either artificially produced or automatically translated by AI technology, potentially lowering the standard of search findings’ quality.

Microsoft and OpenAI are facing multiple copyright infringement cases within the court system, due to the fact that AI-driven tools such as Copilot and ChatGPT primarily rely on internet resources for their learning process.

Although a different study indicated that progress in the technology might be hindered because of insufficient high-grade material for training purposes. This predicament appears to impede leading AI research institutions such as OpenAI, Google, and Anthropic from advancing their AI models significantly.

Even OpenAI’s CEO thinks the dead internet theory is coming true

Sam Altman, the CEO of OpenAI, recently stated on X, ‘Although I’ve never given the idea of the dead internet theory much credence, it appears that there are an unprecedented number of Twitter accounts controlled by Language Model (LLM) these days.’

While I haven’t given much credence to the idea of a “dead internet,” it appears that an increasing number of Twitter accounts are indeed being managed by Language Model (LLM) entities these days.

It appears that AI-powered search tools such as ChatGPT are becoming increasingly preferred by users over traditional search engines like Google and Bing, with many individuals expressing a growing preference towards these AI-driven search options when conducting searches.

Sam Altman, CEO of OpenAI, has shared that he seldom uses Google for searches these days and instead leans on ChatGPT more often. Despite this shift, he expressed some reservations about whether ChatGPT could truly displace Google as the search giant, given its long-standing dominance in this field.

Google is additionally incorporating AI into its search functions through tools such as AI Mode and AI Summaries. However, these tools have largely failed to meet the mark when it comes to dependability. While AI Summaries claim to provide concise, web-scraped responses, they’ve fallen short by often producing inaccurate information. What’s more troubling is that they’ve even suggested dangerous actions like eating rocks, applying glue, and in extreme cases, encouraging users to contemplate suicide.

Google argues that in cases where its search function produced misleading results, this was largely due to insufficient data available on specific topics online. They contend that the scarcity of comprehensive information on these subjects may have led to such instances. Furthermore, they claim that screenshots showing their system generating misleading responses were potentially falsified by users.

In 2024, Jack Dorsey, a former CEO and co-founder of Twitter, predicted that distinguishing reality from fabrication could become challenging due to advanced image creation techniques, deepfakes, and videos. He advised users to exercise caution and personally verify the authenticity of information for themselves.

Is it possible that advanced AI and automated systems could eventually dominate the web, degrading the quality of search results? I’d be interested to hear your perspectives on this topic. Feel free to share your thoughts below.

Read More

2025-09-11 16:11