In an era where artificial intelligence (AI) promises groundbreaking advancements, it becomes increasingly crucial to dissect the potential pitfalls that lurk beneath the surface. Michael P. Ferguson’s recent article, “Entering the Age of Artificial Truth,” offers a comprehensive exploration of the concerns surrounding AI, specifically focusing on generative AI and large language models. These technologies, exemplified by OpenAI’s ChatGPT or Googles BARD, raise profound questions about the reliability of information in our digital landscape.
Ferguson’s critique goes beyond the mere acknowledgment of AI’s capabilities. He unveils the unsolved issues at the core of these powerful tools, emphasizing their propensity to create self-amplifying echo chambers. These echo chambers, fueled by vast datasets, become breeding grounds for flawed or fabricated information, painting a concerning picture of a distorted online environment. The author aptly questions the reliability of AI-driven internet searches, using misattributed quotes as a poignant example to underscore the potential distortion of historical facts.
The contemporary digital landscape can be characterized as a distorted reflection of the internet’s evolution over the past two decades. Within this realm, self-reinforcing echo chambers perpetuate flawed or even concocted information, a phenomenon aptly termed ‘Habsburg AI.’ This occurs when AI-generated content is continuously fed back into another AI system, resulting in a self-amplifying loop of information ‘inbreeding’ that produces distorted and unreliable data.
The central argument presented by the author is that the true peril of generative AI lies in the dissemination of inaccurate information. The reliance of AI on a “warped mirror of what’s on the internet for the last 20 years” underscores the challenges posed by information inbreeding, often referred to as the Habsburg AI effect. This phenomenon gives rise to false artificial historical facts, creating a complex environment where sifting through information for solid and grounded facts becomes a formidable task.
The shift in America away from traditional sources like sourced news, peer-reviewed journals, and established checks and balances such as expert knowledge and reputable publishing houses amplifies the risks. Advanced search tools, driven by AI, are cautioned to reinforce existing prejudices and accelerate human biases, turning into self-amplifying echo chambers that perpetuate flawed or fabricated information. The reliability of data becomes vulnerable to online practices like search-engine poisoning, keyword stuffing, and information inbreeding, leading to potential misattributed quotes and distortion of historical facts.
Manipulative tactics, such as search-engine poisoning, keyword stuffing, and spamdexing, further exacerbate these issues, allowing AI to fuel deceptive schemes. Notably, this manipulation extends to the creation of artificial historical narratives, as witnessed in instances where disinformation infiltrated trusted sources, influencing the consensus that led to the invasion of Iraq in 2003.
The ramifications of AI-generated “artificial truth” extend far beyond misattributed quotes. Ferguson explores the broader societal implications, emphasizing the risk of a distorted online environment. He warns that practices like search-engine poisoning and keyword stuffing, once manipulated by programmers, can be supercharged by AI, leading to a flood of unreliable information. The author draws attention to the potential consequences for critical aspects of society, including the shaping of public opinion, influencing news consumption, and the rise of questionable AI-authored literature.
Ferguson’s exploration extends beyond the immediate concerns of misattributed quotes. The societal implications of AI-generated “artificial truth” are vast, influencing public opinion, shaping news consumption patterns, and even contributing to the emergence of questionable AI-authored literature. He warns against the manipulation of online practices like search-engine poisoning and keyword stuffing, which, when combined with AI, can lead to an overwhelming flood of unreliable information. The consequence is a distorted information environment that challenges meaningful communication and reinforces existing biases.
The COVID-19 pandemic and the 2020 presidential election marked critical junctures where social media platforms, in an effort to curb misinformation, inadvertently censored reports that later proved to be true. This paradoxical situation contributes to a ‘Tower of Babel” effect, fostering an online ecosystem rife with self-replicating fictions.
The societal implications of AI-generated “artificial truth” become even more significant in a world that already embraces and amplifies alternative facts and fake news. It goes beyond the concern that disinformation could infiltrate the datasets propagated by generative AI. Instead, it creates a distorted information landscape where accessing meaningful information becomes challenging, and people’s biases are further entrenched.
Societal shifts, including a decline in book reading, diminished trust in traditional news sources, a reduced emphasis on higher education, and an increased reliance on platforms like TikTok for news consumption, create fertile ground for algorithmic manipulation. In this environment, conventional safeguards against misinformation, such as expert knowledge, reputable publishing agencies, and reliable news sources, have seen a decline in influence.
The article explores the concept of a Tower of Babel effect, where self-replicating fictions can influence the shaping of public opinion. When disinformation infiltrates trusted sources, it gives rise to artificial historical facts or a distorted version of truth, representing biased propaganda that poses inherent dangers. The author illustrates how this phenomenon unfolded during critical events like the Iraq war, the Covid pandemic, and the 2020 Presidential election.
Advanced search tools, rather than refining perspectives, have the potential to reinforce existing prejudices and biases, thus hyper-accelerating human bias in the online sphere. Without a clear strategy to manage the risks associated with the propagation of artificial truth, there is a looming danger that America could find itself submerged in a sea of incoherence.
Drawing a parallel with George Orwell’s dystopian masterpiece, 1984, Ferguson emphasizes the potential consequences of living in a world where historical memory is entirely untrustworthy. Orwell’s vision of a society where the past is continuously rewritten resonates with concerns about the reliability of historical information in the age of artificial truth. The quote from 1984 serves as a stark reminder of the dangers of a manipulated historical narrative, echoing the potential chaos AI-generated misinformation could unleash.
“Do you realize that the past, starting from yesterday, has been actually abolished? If it survives anywhere, it’s in a few solid objects with no words attached to them, like that lump of glass there. Already we know almost literally nothing about the Revolution and the years before the Revolution. Every record has been destroyed or falsified, every book has been rewritten, every picture has been re-painted, every statue and street and building has been renamed, every date has been altered. And that process is continuing day by day and minute by minute. History has stopped. Nothing exists except an endless present in which the Party is always right. I know, of course, that the past is falsified, but it would never be possible for me to prove it, even when I did the falsification myself. After the thing is done, no evidence ever remains.” (George Orwell, 1984, pg 103)
Orwell’s quote underscores the repercussions of navigating an entirely unreliable historical memory, manipulated by entities seeking control over the narrative. This manipulation poses a risk of further fragmenting society, steering it away from grounded, observable facts and deeper into the labyrinth of alternative facts and fake news.
Personally, my apprehension about the perils of digitally manipulating text, akin to Orwell’s 1984, intensified with the shift from printed books to PDFs. The digital nature of these documents makes them susceptible to alteration if accessed by malicious actors. In the online realm, where data exists in a non-physical form, there is a significant risk of the internet becoming a platform for rewriting history and manipulating text to align with the agendas of those aiming to control and influence public opinions. This highlights the inherent dangers of embracing an “artificial truth” rooted in distorted or unreliable information.
As the author highlights the risks posed by AI, the need for alternative sources of reliable information becomes evident. While the article doesn’t explicitly provide solutions, it implies a return to traditional checks on information, such as expert knowledge, reputable publishing agencies, and hard news sources, which have lost influence in the face of the AI onslaught.
Again the dangers of alternative facts, fake news are a major concern, however if the internet searches become unreliable for well sources information regarding history, we need to go to peer reviewed journals, reputable publishing agencies, hard news, and expert knowledge. It is important to combat opinion news based in alternative facts and fake news with sourced news based and grounded in the critical method of research that involves peer review and source criticism to find grounded truth. We need to rely on reliable means to verify information and facts. The need for fact checking is at an all time high with the potential dangers of AI propagating bad information.
The introduction of the “Habsburg AI” effect serves as a striking metaphor for a concerning phenomenon within AI systems. The continuous feedback loop, akin to the historical concerns of inbreeding in royal families, paints a vivid picture of information repetition within AI systems. The negative consequences of this repetition manifest in the form of distorted information, further complicating the challenge of distinguishing fact from fiction.
Named after the Habsburg dynasty notorious for inbreeding, the term “Habsburg AI” is intended to highlight the self-amplifying issue of information “inbreeding” within generative AI models. This phenomenon occurs when these models rely on datasets from other AI models, creating an echo chamber filled with flawed or fabricated information.
Through personal experimentation with generative AI in historical text, I have witnessed this problem firsthand, observing the proliferation of artificial historical facts that are, in reality, false. The feedback loop, where AI-generated information is continually fed back into another AI program, exacerbates the issue, resulting in the generation of more inaccurate information. This poses a significant risk as the information produced by generative AI can be flawed, leading to the generation of nonsensical content, false details, or misattributed quotes.
Ferguson’s call to action is clear – a return to traditional checks on information. Expert knowledge, reputable publishing agencies, and hard news sources, which have somewhat lost their influence, are critical components in combatting the rise of alternative facts and fake news. The article implies that, amid the dominance of AI, these traditional pillars of information verification must be revived to ensure a reliable foundation for knowledge.
Ferguson’s article is a poignant reminder that as we step into the age of artificial truth, caution is paramount. In a world increasingly intertwined with AI, an unwavering examination of its impact on information reliability becomes imperative. Exploring alternative avenues rooted in traditional information verification methods is not just a suggestion but a necessity to navigate the complexities of the digital era without succumbing to a web of distorted truths. As the digital age unfolds, our collective responsibility to uphold the integrity of information becomes more critical than ever.
The Pulling the Thread Podcast 
Leave a comment