There is no Pirates of the Caribbean 6: How long is the internet going to work?

Generative AI is improving, and its use is increasing. AI tools have rapidly accelerated, and, as a result, AI is now widely used to generate lesson plans, find recipe measurements, and draft lists of potential client companies. It can bring fan ideas to life, create high-quality art at the push of a button, and superpower creative ideas (in theory). 

Here is the problem: AI makes it so easy for fans to bring ideas to life, it itself cannot actually tell the difference between fan fiction and reality.

Take Pirates of the Caribbean 6. 

This movie does not exist. There is no cast, no producer, no script, and no trailer. Yet Google AI believes otherwise, even giving it a title – Pirates of the Caribbean: A Day At The Sea. 

There are fan trailers and AI generated screenshots from the film, a listed director and producer, cast predictions, and the ability to mark the film as ‘already watched’. 

A specific search for one of the other made-up titles, “Pirates of the Caribbean 6: The Return of Davy Jones”, sees the AI overview treating it as a “planned sequel”, featuring Jenna Ortega and Margot Robbie as new characters. None of this is true either. According to The Direct, it is based entirely on a fan-made trailer posted to YouTube in 2024. 

This goes beyond silly hallucinations 

The issue does not begin or end with entertainment fandom. Anything from fantasy medical conditions (brought to life with AI images) to made-up historical events, complete with AI-generated news articles, are possible. Most may not even be malicious: a school project posted online can be treated by the LLM as fact, with few – if any – practical tells to differentiate it. 

There is no Pirates of the Caribbean 6. I know this, you know this, Disney knows this. But take something more obscure, that I didn’t know, and I might believe it. The internet lies – this is not new. Yet our sources of truth and authority are becoming ever more fallible as they cut budgets and incorporate chatbots. Our arbiters of reality now exist in Reddit threads, Wikipedia source links, and expert YouTube videos, none of which would have passed muster in a 12-year olds’ history assignment fifteen years ago. Traditional media that must use clickbait to compete with the content flood, and whose reporters do much of their time-stretched research on social platforms, are prone to mistakes or misrepresentations. As reports by The New Yorker and Associated Press highlight, government agencies in the US are increasingly caught giving conflicting statements and removing information for political reasons. The internet has always lied – but does it ever tell the truth anymore? 

AI is improving, but it is trained on everything that already exists on the internet – which can theoretically be found by simply by using a search engine. However, the search engines are deteriorating. A few years ago, a simple search could reveal all the human-made lesson plans, recipe conversions, and company classifications ever needed; now they are obscured by pages of ad placements and SEO-optimised filler. It simply becomes easier to ask ChatGPT or Google AI instead and then filter through the results. 

This comes with inherent downsides, as AI does not just supply one half of the equation. It can now pollute its own datasets. It is trained on the same internet it is now populating at unimaginable speed. As outlined by The Byte, new data estimates that over 90% of content online will be AI generated by 2026. Nothing fools AI as well as AI itself, so every improvement we make on one side also heightens the challenges on the other. 

What does that mean for the internet long-term? 

Everything online is becoming increasingly questionable in merit and value, as AI use dilutes both the reliability of information and content, and exponentially reduces the effort required to make it. This is a long-term problem, and without addressing it, we cannot expect to solve it. 

In the short term, having real-world roots could be the best way of both differentiating and establishing reliability. When everything digital can be faked or called into question, grounding in reality may only be truly possible offline. It is unlikely that users will quit the internet in droves. Trust is undeniably higher in a brand with a store on a high street than one that only exists in Instagram ads with AI models, and durable fandom is built better through live shows than TikToks. 

In the long term, we appear to be seeing the trend of quality decline infecting the internet itself, rather than simply some platforms on it. To solve it, a bigger shift in mindset is required – one that reevaluates the purpose and value of digital content, the role and importance of its creators... and why, whether, and how much the online rat race is worth participating in.