How Do We Find Authenticity in the Era of AI-Translated Content?

Why Trust Techopedia

Do you consider yourself vigilant about the quality and authenticity of the information you consume online?

Traditionally, the top results on the first page of Google, often perceived as the most reliable, are not necessarily a beacon of truth. And for the most part, we have learned how brands blend SEO-optimized content with marketing and advertising strategies. 

However, the rise of AI-generated and translated content is already beginning to muddy the waters further.

The old phrase “believe only half of what you see and nothing of what you hear” has never been more fitting for a digital age where AI can fake the written word, audio, and video content generated by humans.

Key Takeaways

  • AI-generated and translated content is reshaping online information, with a study revealing that the widespread use of machine translation is leading to deteriorating content quality.
  • With nearly 50% of internet traffic driven by bots and most of us relying on search engine results to find our answers, the ability for distortion grows rapidly.
  • As the world faces a historic 2024 election year, concerns arise about AI’s potential to disrupt democratic processes, with incidents like a deepfake robocall impersonating US President Joe Biden as one example.
  • The difficulty in identifying AI-generated content raises questions about the reliability of AI-based plagiarism detection tools, emphasizing the need for nuanced approaches in distinguishing between human and AI-authored content.

Redefining Authenticity in the Era of AI-Translated Content

A study published earlier this month by researchers at the Amazon Web Services AI lab presents a startling revelation about the state of machine-translated content on the web.

The research, published on the pre-print server arXiv, delves into the depths of 6.38 billion sentences, unearthing a reality that might change how we perceive internet content, particularly in languages spoken in Africa and the Global South.


The report suggests AI is increasingly being deployed to mass-produce substandard English content, which is further muddled through AI-powered machine translation into multiple languages, leading to a continuous degradation of information and vast swaths of the internet becoming cluttered with progressively deteriorating AI-scrambled replicas.

The study leaders said in their abstract: “We show that content on the web is often translated into many languages, and the low quality of these multi-way translations indicates they were likely created using machine translation.

“Machine-generated content not only dominates the translations in lower resource languages; it also constitutes a significant fraction of the total web content in those languages.

“We also find evidence of a selection bias in the type of content translated into many languages, consistent with low-quality English content being translated en masse into many lower-resource languages via machine translation.

“Our work raises serious concerns about training models such as multilingual large language models (LLMs) on both monolingual and bilingual data scraped from the web.”

As AI-translated material becomes more prevalent, the web becomes inundated with content translated across multiple languages, often losing accuracy and context, suggesting this influx of AI-translated material is set to exacerbate the existing online trust problem.

The Digital Illusion and ‘Dead Internet Theory’

Findings from Imperva’s Bad Bot Report revealed that bots generate a staggering 47.4% of internet traffic (PDF). These revelations breathe life into the once-dismissed ‘dead internet theory,’ transforming it from a fringe conspiracy into our daily digital lives.

It’s a thought-provoking and somewhat disconcerting realization that a significant portion of our online world is driven by AI-generated content and automated bots, often interacting with us without our knowledge.

An ex-CIA operations officer, Dan Woods, suggested that this problem is much bigger than many think and that more than 80% of Elon Musk’s X accounts could be fake bots. Elsewhere, a simple Google search for a common term often reveals an overwhelming 8 billion results, but our access is typically limited to a mere 15 pages, offering roughly 150 results.

Such scenarios paint a picture of the internet not as an endless wilderness of diverse content but as a confined space where the same information is echoed repeatedly, akin to a hall of mirrors.  

This realization challenges our perception of the internet as an open field of infinite exploration, nudging us to question the authenticity and diversity of the content we encounter daily. But what makes this epiphany so crucial is that billions of people will consume this AI-translated content before heading to the elections this year.

2024 is the Biggest Election Year in History 

As the world prepares for the most significant election year in history, over 60 countries and nearly half of the global population are getting ready to vote in 2024. Yet the role played by AI in shaping outcomes is already becoming a crucial concern.

A recent incident involving a robocall imitating the voice of US President Joe Biden and instructing residents of New Hampshire not to vote serves as a timely reminder of how disruptive AI can be during elections. This occurrence and open access to AI technology raises urgent questions about upholding democratic processes.

In a conversation with Bloomberg’s Francine Lacqua, Microsoft co-founder Bill Gates recently forewarned that “bad guys will be more productive” with AI, highlighting the ease with which malicious actors could exploit AI advancements to influence voter perceptions and behaviors.

How will nations safeguard their elections against bad actors leveraging the influence of generative AI, particularly deepfakes, which could distort the political narrative and sway public opinion?

With its unprecedented scale, the global electoral landscape of 2024 presents an extraordinary test of our ability to balance technological innovation with preserving democratic values.

Who Checks the Checkers? 

As Silicon Valley continues its obsession with moving fast and breaking things, we are only just beginning to understand how difficult it can be to identify AI-generated content. The sentiment was recently echoed by OpenAI CEO Sam Altman, who warned schools and policymakers about over-reliance on AI-based plagiarism detection tools. The suggestion that they don’t work further ignited controversy. Critics have since dubbed many of these emerging solutions digital snake oil when questioning their accuracy and ethical implications. 

Instances are also shared where classic texts like the US Constitution were misidentified as AI-written by these tools, suggesting a fundamental flaw in their design and function.

If a plagiarism checker is trained on a particular writing style, it may falsely recognize similar human-written content as AI-generated. This problem also raises profound questions about the nature of authenticity in writing.

If a text crafted by a human can be misidentified as machine-generated, where does that leave the authenticity of human creativity? Conversely, texts manipulated by “Humanizer” tools to bypass AI detectors are deemed original, blurring the lines between human and AI authorship.  

These dilemmas highlight the complex challenges in distinguishing AI-generated content from human writing. It also underscores the need for a more nuanced approach to AI detection tools in academic and professional settings.

The Bottom Line 

As AI-translated or generated content and interactions with bots become increasingly commonplace, our grasp of reality becomes worryingly skewed. The once-clear line between credibility and mere visibility is becoming blurred, leading to a distorted perception of authenticity and truth. These can be found in our newsfeeds, often fueled by disinformation tactics, echo chambers, deepfakes, fabricated accounts, and algorithms riddled with inherent biases.  

The stakes have been raised, and the world now has much more to worry about than the authenticity of an article recommending the best ten air fryers. The adage that ‘content is king’ holds, yet the context means everything.

In understanding and acknowledging the context — the source, the motive behind the creation, and the potential biases — we can begin to discern the true nature of the information we consume. This vigilant approach to digital content is crucial in preserving the integrity of data and the fabric of our perceived reality.


Related Reading

Related Terms

Neil C. Hughes
Senior Technology Writer
Neil C. Hughes
Senior Technology Writer

Neil is a freelance tech journalist with 20 years of experience in IT. He’s the host of the popular Tech Talks Daily Podcast, picking up a LinkedIn Top Voice for his influential insights in tech. Apart from Techopedia, his work can be found on INC, TNW, TechHQ, and Cybernews. Neil's favorite things in life range from wandering the tech conference show floors from Arizona to Armenia to enjoying a 5-day digital detox at Glastonbury Festival and supporting Derby County.  He believes technology works best when it brings people together.