Trust AI At Your Own Risk
AI (Artificial Intelligence) programs live up to the belief that computer programs are a matter of "Garbage in and garbage out." In fact, AI (Artificial Intelligence) relies directly on the principle of "Garbage In, Garbage Out" or "GIGO," meaning that flawed data, supposition, conjecture, Leftist biased, or incomplete inputs inevitably lead to flawed or useless outputs.
Why GIGO Applies to AI
Let's talk about training data quality. AI and machine learning models are statistical systems trained on massive sets of historical data selectively input by biased programmers. If that data contains errors, missing context, or systemic biases, the AI will learn and replicate those exact flaws in its output.
Of course, there's the problem of "prompt and input ambiguity." What's that about? Prompt and input ambiguity means that a request, question, or instruction given to an AI or computer system can be interpreted in more than one way, or lacks the specific details needed to know what the user actually wants.
So, if a question is vague, poorly worded, or factually incorrect, that will prompt AI to give results that are irrelevant or in some cases give you a nonsensical response. A great example of "Garbage prompts in, garbage answers out".
Instead of asking a clarifying question, an AI will often "fill in the gaps on its own" with its own "plausible defaults," which can lead to unwanted or inaccurate results.
Using a word that has multiple meanings, for example: asking about "apple" could mean the tech company or the fruit. The phrasing matches multiple possible actions or goals, for example: "handle this" could mean reply, summarize, or escalate.
Using vague verbs that do not specify the operation can be a problem, for example: "look at," "help," or "improve" without saying how. And using vague references like "it," "that," or "do it again" without previous conversation history to explain what "it" refers to can create problems.
Why It Matters With AI?
Believe it or not, AI programs have "Hallucinations." When a prompt is too broad or vague, the AI guesses your intent and may generate false or misleading facts. If you know your subject, then you can hopefully catch the errors that AI produced. Of course, if you don't know your subject and are literally trying to learn some of the nuances of a subject, my recommendation is that you shouldn't do that using AI because AI is not reliable.
If I recall correctly, a widely used AI-detection software program actually flagged the Declaration of Independence as being over 95% "AI-generated." Yes, it's true. an AI program mistook its 1776 formal, symmetrical, and balanced 18th-century writing style for some sort of modern machine-learning text patterns.
How does any of this apply to my research and this blog, the American Cowboy Chronicles?
Well, let's talk about how AI can produce errors regarding the history of the Old West which I love researching.
The reasons that I don't use AI are many. First, I've learned that AI errors take place because AI likes to repeat Hollywood myths instead of referencing archival newspapers which I research a lot. And yes, AI invents local history and lore.
When asked about some obscure Ghost Town or a Frontier town, possibly a regional landmark, or maybe a local figure, I've found that AI programs will frequently invent plausible-sounding events or origin stories to satisfy my question. It's as if it sees my prompt as a challenge to see if I accept its fiction writing. Also, I've found that AI has generated history pages fabricating entire 19th-century tragedies, mines, and town legends that never happened.
As for what AI uses for its sources? It conflates Dime-Novel Fiction with Fact.
Because 19th-century dime novels, Wild West show programs, and modern Western scripts make up a massive portion of online text, AI frequently treats fictionalized folklore as "verified biography." This leads to text errors such as attributing fictional gunfights to real lawmen or misstating basic dates and locations of famous events like the Gunfight at the O.K. Corral.
AI also like to flatten the complex histories of the American Pioneers, the Settlers, Homesteaders, Lawmen, Outlaws, Native Americans, what it took to survive the West into generic Hollywood tropes. And worst, I've learned that AI has even fabricated period photography of famous figures.
AI models tend to flatten history into broad, generic tropes because they are trained on massive amounts of internet text, pop culture, and entertainment media where Hollywood clichés vastly outnumber nuanced historical research.
When prompted simply about the "Old West," an AI defaults to the highest-probability patterns in its training data. That would be all that bad if it were historically accurate films like The Civil War series by Ken Burns, but instead it will source Dime-novel myths.
Of course, it may also negate the rugged, self-reliant individualism that conquered land that never tasted a plow, or what it took to come West in a covered wagon, and instead attempt to devalue the sacrifice of American Pioneers and Settlers by falsely stating that their "survival depended primarily on interdependence, government land policies, and collectivism." Yes, as if promoting a Soviet Union propaganda film.
Other false claims include stating how most Western towns had strict gun-control ordinances, how "violence in the Old West was tied to corrupt corporations, labor disputes, or fights against capitalists cattle barons." Once you read answers like this, you know that its programmer was a Leftist. But we'll talk about that in a few minutes.
AI fabricated what it called "Archival Portraits & Photos" of Crazy Horse. What AI did not know is the Oglala Lakota leader Crazy Horse famously refused to be photographed during his lifetime. While that's a fact, that hasn't stopped AI from flooding social media sepia-toned AI-generated "tintypes" claiming to be newly discovered portraits of him.
As for fabricated photo of outlaws and other Old West characters? AI images frequently depict figures like Billy the Kid, Jesse James, and Wyatt Earp in crisp, cinematic poses that contradict the handful of verified photographs held by the Library of Congress.
As for erasing secondary figures from photographs? AI actually reworks versions of famous frontier photographs, such as portraits of Buffalo Bill Cody and Wild Bill Hickok, to crop out, distort, or completely erase people like Texas Jack Omohundro. And, because AI programs rely heavily on 20th-century Hollywood movies instead of 19th-century archives, AI routinely puts out photos people wearing modern Stetsons, low-slung Hollywood "Buscadero" gun belts, and shiny double-action revolvers into the 1860s and 1870s.
AI like repeating "Pop-Culture Myths" and telling us that they are "Facts." When asked to describe frontier violence or town life, text models frequently treat Hollywood-style main-street quick-draw duels as routine daily occurrences instead of being the rare, localized, sensationalized events that they were. And yes, AI essays often stitch two separate frontier facts together with "invented causation."
As for gunfights and gun laws in the Old West? AI has made the claim that the 1881 "Gunfight at the O.K. Corral" caused Western towns to pass gun-control ordinances. Actually, the city of Tombstone already had strict gun laws on the books before the shootout happened.
It's true. The AI claim that the Gunfight at the O.K. Corral caused Western towns to pass gun-control ordinances flips the historical cause and effect. In reality, Tombstone already had strict gun control on the books, and enforcing that exact law is what triggered the shootout.
Before the October 26, 1881 gunfight, Tombstone had passed Ordinance No. 9, which outlawed carrying "any deadly weapon" within city limits without a permit.
Visitors entering Tombstone were legally required to check their firearms at a saloon, hotel, horse stable, or lawman's office upon arrival.
On the day of the shootout, City Marshal Virgil Earp deputized his brothers Morgan and Wyatt, along with Doc Holliday, specifically to disarm the Billy Clanton and Frank McLaury. While Billy Clanton was openly carrying a revolver, and Frank McLaury was openly carrying a revolver and rifles in their saddles in defiance of the local ban, Billy's brother Ike Clanton was unarmed and Frank's brother Tom McLaury checked his gun in when he arrived in town as required.
As for AI oversimplifying Native American history? AI makes Native American History sound like all of the individual Indian Nations got along, spoke the same languages, had the same customs, were all the same people, all lived in peace, never waged war against each other, and never displaced another tribe's people or made a their enemies slaves.
That distortion of the truth is the fault of whoever programmed that AI program. AI-generated information on the American West routinely blurs how distinct Indigenous nations were and still are — such as the Shoshone, Crow, and Blackfeet—into a single monolithic group, overlooking shifting alliances, borders, and conflicts.
In reality, Indigenous nations across North America were, and remain, distinct sovereign societies with complex histories that included diplomacy, trade, warfare, territorial expansion, and captivity.
As for their cultural and linguistic diversity? North American tribes had hundreds of distinct languages. In fact, according to the Smithsonian National Museum of the American Indian, roughly 500 distinct Native languages were historically spoken across North America. These languages belonged to dozens of completely separate language families, making them as different from one another as English is from Mandarin or Arabic.
As for how different were their customs and systems of government? For "Native Americans," their tribes geography shaped the ways they lived entirely differently. Coastal nations in the Pacific Northwest built permanent cedar plank houses and relied on salmon and whaling. Nomadic nations on the Great Plains followed buffalo herds and lived in portable tipis. Believe it or not, their societies ranged from nomadic hunter-gatherer bands on the Great Plains to massive agricultural urban centers like Cahokia.
Bcause nations were so culturally distinct, groups developed specialized trade jargons like Chinook Wawa and Plains Indian Sign Language to communicate across borders. And yes, their political systems varied wildly across the continent. For example, the Haudenosaunee (Iroquois) Confederacy did in fact operate under a complex Constitutional democracy, while the Natchez people of the lower Mississippi Valley had a strict, hierarchical monarchy.
When I was much younger and became very aware of the differences between tribes, including once being told that the term "Native American" was a generic "umbrella term" that meant nothing simply because the tribes were as different from each other as France is from Germany even though they share a common border. The term "Native American" is a modern umbrella label, rather than an identity that existed before European contact.
The term "Native American" gained widespread use in the 1960s and 1970s. During the civil rights era, activists from different tribes united to fight for shared legal rights, sovereignty, and federal recognition, making a collective term politically useful. Today, most Indigenous people prefer to be identified by their specific tribal nation first, such as Navajo, Cherokee, Lakota, or Ojibwe, and so on
I found out that warfare and territorial displacement and inter-tribal warfare in North America was no different than what took place in Europe, Asia, Africa, and South America. And yes, they formed alliances for a millennia of more no differently than the Europeans did for a millennia of more.
My comparing Indigenous nations to European nations is well founded in historical reality. Before colonization, the Americas were home to hundreds of distinct, sovereign nations. Grouping them all together often hides the vast differences in how they lived, governed, and viewed the world.
When Europeans arrived, Indigenous nations frequently formed strategic military alliances with European powers, such as the French, British, or Spanish, to gain an advantage against their own longstanding Indigenous enemies. And no, they did not see themselves as a single unified group fighting a common enemy when Europeans arrived.
As for taking captives during warfare? That was a widespread practice across many Indigenous societies. Depending on the nation's customs, captives might be adopted into families to replace fallen relatives, held for ransom, used in diplomatic exchanges, or subjected to forced labor. It all depended on what tribe took you captive. As for the Comanche empire in the Southwest and Southern Plains, the Comanche built a powerful empire driven largely by raiding, the horse trade, and a vast network of captive trading across borders.
As for owning African Slaves? Later, in the Southeast, some wealthy members of nations like the Cherokee, Chickasaw, Choctaw, Creek, and Seminole adopted European-style chattel slavery. Yes, they held enslaved people of African descent. By the late 18th and early 19th centuries, a wealthy, slaveholding elite emerged within these five nations, adopting plantation agriculture and strict racial slave codes similar to those in neighboring Southern states. When the U.S. government forcibly removed these nations to Indian Territory (modern-day Oklahoma) in the 1830s on the Trail of Tears, Native American slaveholders brought thousands of enslaved Black people with them.
As I've said over the years, just as French and German belong to different language families, Indigenous nations spoke hundreds of unrelated languages. Across North America alone, there were dozens of distinct language families, meaning neighboring tribes often could not understand each other without sign language or trade jargons.
Like European countries, Indigenous nations had long histories of diplomacy, trade networks, warfare, and shifting alliances. In fact, Indigenous nations fought wars over hunting grounds, trade routes, resources, and political dominance. Territories shifted and displacement took place. Boundaries were never static.
For example, during the 17th-century Beaver Wars, the Haudenosaunee expanded their territory and displaced several neighboring nations around the Great Lakes. Similarly, the Lakota migrated onto the Great Plains and pushed out rival groups like the Crow and Pawnee.
When Europeans arrived, Indigenous nations frequently formed strategic military alliances with European powers, such as the French, British, or Spanish, to gain an advantage against their own longstanding Indigenous enemies. And no, they did not see themselves as a single unified group fighting a common enemy when Europeans arrived.
As for taking captives during warfare? That was a widespread practice across many Indigenous societies. Depending on the nation's customs, captives might be adopted into families to replace fallen relatives, held for ransom, used in diplomatic exchanges, or subjected to forced labor. It all depended on what tribe took you captive. As for the Comanche empire in the Southwest and Southern Plains, the Comanche built a powerful empire driven largely by raiding, the horse trade, and a vast network of captive trading across borders.
And no, those slaves were not latter emancipated by Lincoln's proclamation. The fact is that the Emancipation Proclamation did not free the people they enslaved. Because these Indigenous nations held sovereign status through federal treaties, U.S. wartime executive orders, which is what the Emancipation Proclamation was, and U.S. Constitutional Amendments did not automatically apply within their territories.
The 1863 Emancipation Proclamation applied only to enslaved people in Confederate states in rebellion against the United States. Indian Territory was legally distinct from U.S. states and governed by tribal treaties. When the 13th Amendment was ratified in December 1865 to abolish slavery throughout the United States, it still did not immediately end slavery in Indian Territory because of tribal sovereignty.
Because factions within these five nations had allied with the Confederacy during the Civil War, the U.S. government required them to negotiate new treaties in 1866.
Abolition and citizenship: These treaties formally abolished slavery within the five nations. They also required the nations to grant tribal citizenship and rights to the formerly enslaved people, known as "Freedmen," though the implementation of those citizenship rights led to legal and political disputes that lasted for over a century.
Because factions within these five nations had allied with the Confederacy during the Civil War, the U.S. government required them to negotiate new treaties in 1866.
Abolition and citizenship: These treaties formally abolished slavery within the five nations. They also required the nations to grant tribal citizenship and rights to the formerly enslaved people, known as "Freedmen," though the implementation of those citizenship rights led to legal and political disputes that lasted for over a century.
So Why Does AI Over-Simplify History?
AI doesn't "know" history. AI does not source diaries, treaties, or academic histories, an AI program treats the myth as the default truth unless specifically steered away from it. Add that to the fact that Left-leaning revisionists, anti-conformists, programmers input information to deliberately try to subvert the traditional ethos of Westerners of the frontier, and you get answers that don't align with what took place in reality in the Old West.
Instead of emphasizing historically accurate themes of resilience, self-reliance, agricultural innovation, and the building of modern American infrastructure and food systems, AI examines how "plowing native prairies and introducing non-native livestock disrupted soil health, native grasses, and local wildlife populations, contributing to events like the Dust Bowl."
AI programs generate text based on its input. The perspective of its programmer matters. And while intended to sound respectful and as "politically correct" as possible, the generalizations that AI produces actually minimizes the efforts of Americans Pioneers, Settlers, Homesteaders, Cattlemen, Cowboys, Lawmen, Businessmen and women who risked everything they owned. And yes, AI strips Indigenous nations of their individual identity and their true history.
As for AI imposing a "Romanticized Harmony" trope to Native Americans? Rather than citing specific treaties, migrations, or resource disputes, AI uses summaries that often default to vague historical generalizations about early peoples living in timeless harmony with nature.
The "Confident Wrongness" Problem
AI has "Hallucinations." In fact, AI frequently invents facts, historical dates, statistics, or academic citations that look real but do not exist. And when AI comes up against a gap in its data, it makes a statistical "guess" rather than admitting it lacks the information. Human researchers often fail to catch AI mistakes because they defer to the tool's confident presentation — a problem documented across medical, legal, and academic fields.
Concerns about political slant and factual accuracy are central to current debates over AI reliability, with independent research confirming measurable left-leaning tendencies. This is the result of deliberate programming, and that's why reliability of getting unbiased answers varies widely depending on the model and task.
Evidence on Political Bias
AI has a measurable "Left-Leaning Skew." Multiple academic and media evaluations, including a 2025 study from the Stanford Graduate School of Business and 2026 testing by The Washington Post, found that popular chatbots frequently favor left-leaning arguments on political topics.
So How Often Is AI Wrong?
Studies on AI (Artificial Intelligence) show error and hallucination rates ranging from 37% to as high as 60% to 95%, depending on the task, model, and source material. A study by the BBC and the European Broadcasting Union found that 45% of AI news queries across major platforms (ChatGPT, MS Copilot, Gemini, and Perplexity) produced factual errors.
A 2026 report highlighted in The Atlantic found that 40% of responses from commonly used chatbots contained at least one major factual error when answering questions about high-stakes topics like elections. The widely cited 95% failure rate comes from a 2025 report by MIT Media Lab's Project NANDA, which found that 95% of corporate generative AI pilots failed to generate rapid financial returns. That figure measured business return on investment and workflow integration rather than basic factual accuracy.
Research highlighted by the Columbia Journalism Review noted that AI-powered search tools provided wrong information 60% of the time on average, and leading chatbots incorrectly cited their sources 60% of the time (including fabricating fake academic papers and URLs).
The "Confident Wrongness" Problem
AI models do not "know" when they are missing data. So, instead, they simply use information to generate what AI thinks is a plausible-sounding narrative. That means they are often stating false information with total confidence.
Why AI Fails At Research?
Why AI Fails At Research?
AI has "Hallucinations." In fact, AI frequently invents facts, historical dates, statistics, or academic citations that look real but do not exist. And when AI comes up against a gap in its data, it makes a statistical "guess" rather than admitting it lacks the information. Human researchers often fail to catch AI mistakes because they defer to the tool's confident presentation — a problem documented across medical, legal, and academic fields.
As For AI Leftist Bias?
Concerns about political slant and factual accuracy are central to current debates over AI reliability, with independent research confirming measurable left-leaning tendencies. This is the result of deliberate programming, and that's why reliability of getting unbiased answers varies widely depending on the model and task.
Evidence on Political Bias
AI has a measurable "Left-Leaning Skew." Multiple academic and media evaluations, including a 2025 study from the Stanford Graduate School of Business and 2026 testing by The Washington Post, found that popular chatbots frequently favor left-leaning arguments on political topics.
In the Washington Post analysis, OpenAI's model presented left-leaning arguments 80% of the time, while Google's Gemini provided both perspectives in over 90% of its answers.
Why The Leftist Democrat Slant?
Researchers attribute this lean primarily to a few factors: 1) The input of explicit partisan code; 2) AI searches text from academic papers and the Mainstream Democrat-controlled news; and 3) the demographics of the AI programmers is overwhelmingly young people who are Democrats who lean to the Left, and extreme Left.
The good news is that studies from Stanford and Yale University show that while default outputs for AI programs often carry a Leftist Bias framing, AI programs can be prompted or fine-tuned to adopt a neutral, balanced, or conservative stance. Let's hope they are in the future. Until then, AI is unreliable at best.
Tom Correa
No comments:
Post a Comment
Thank you for your comment.