Big Short Prototype: Trillion-Dollar AI Investment Went Wrong from the Start

Large language models place language before true rational ability and can never reach understanding.

Author: Michael Burry

Translation: Deep潮 TechFlow

The New York Times, Saturday, June 19, 1880

Welcome to the “History Always Rhymes” series. In this series, I start from key perspectives in the distant past to shed light on current events.

One quiet Saturday, I was browsing old newspapers—as I love to do—when I unexpectedly found a report from June 19, 1880, that astonishingly relates to today’s anxieties about AI.

This is the story of Melville Ballard. He had no language from childhood but stared at a tree stump, asking himself: Is the first human being born from here?

This case from 144 years ago—published officially by the Smithsonian Institution—raises a potentially fatal question for today’s large language models and the huge investments behind them. Through an ordinary person’s story, it boldly declares: complex thought is born in silence before language.

Today, deep into the 21st century, placing language before rational ability is not about building intelligence—we are merely creating an increasingly refined mirror.

In that old newspaper, there are two articles worth noting. Starting with the middle article on the third page, titled: “Thought Without Language.”

Of course, the hottest topics today are large language models, small language models, and reasoning abilities.

The full title of that article is: “Thought Without Language—A Deaf-Mute’s Account: His Initial Thoughts and Experiences.” It was first published on June 12, 1880, in the Washington Star.

The protagonist is Professor Samuel Porter of Kendall Green State School for the Deaf and Dumb, who published a paper at the Smithsonian titled “Can There Be Thought Without Language? A Case of a Deaf-Mute.”

The paper begins by discussing the mental activities of deaf-mutes and children without language forms. Its wording and ideas are far behind today’s understanding, so I initially planned to skip it.

But the case’s main character is Melville Ballard, a teacher at Columbia School for the Deaf, who was himself deaf-mute and a graduate of the National Deaf-Mute University.

Ballard said he communicated with his parents and siblings through natural gestures or pantomime in childhood. His father believed observation could develop his intelligence and often took him out for rides.

He continued: just a few years before he was formally introduced to written language, during a ride, he began to ask himself: “How did the world come into being?” He was deeply curious about the origin of human life, the first appearance, and why the Earth, Sun, Moon, and stars exist.

Once, he saw a tree stump and wondered: “Is it possible that the first person to come into this world grew out of that stump?” But he quickly thought that the stump was just the remains of a once-mighty tree; where did that tree come from? It grew slowly from the soil, just like the small saplings before him—then he thought it was absurd to connect human origins with a decayed old tree stump, and dismissed the idea.

He didn’t know what triggered his questions about the origins of all things, but he had already formed concepts of inheritance, animal reproduction, and plant growth from seeds.

His real concern was: at the earliest point in time, before there were humans, animals, or plants, where did the first human, the first animal, the first plant come from? He thought mostly about humans and the Earth, believing that humans would eventually perish and that there was no resurrection after death.

Around age five, he began to understand the concept of inheritance; by 8 or 9, he started asking about the origin of the universe. Regarding the shape of the Earth, he inferred from a map with two hemispheres that it was two large, adjacent disks of matter; the Sun and Moon were two round luminous plates. He felt a certain reverence for them and deduced from their rising and setting that some powerful force must govern their paths.

He believed the Sun entered a hole in the west, emerged from another in the east, passing through a huge pipe inside the Earth, following the same arc across the sky. To him, stars were tiny points embedded in the night sky. He described how he vainly pondered all this until he entered school at age 11.

Before that, his mother told him there was a mysterious being in the sky, but when she couldn’t answer his questions, he despaired and felt sad because he couldn’t gain any definite knowledge about that celestial divine life.

In his first year of school, he learned only a few sentences each Sunday. Although he studied these simple words, he never truly understood their meaning. He attended church but, due to limited sign language skills, understood almost nothing. The second year, he had a small catechism with a series of questions and answers.

Language combined with rational ability to promote the development of understanding.

Later, he could understand the sign language used by teachers. Perhaps some would say his curiosity was satisfied. But that was not the case—when he learned that the universe was created by the great spirit of the Lord, he began to ask: Where does the Creator come from? He continued to seek the essence and origin of that Lord. When pondering this, he asked himself: “After entering the Lord’s kingdom, can we understand God’s essence and comprehend His infinity?” Should he, like his ancestors, say: “Can you search and penetrate God’s nature?”

Professor Porter then presented his core argument to the Smithsonian audience in 1880.

He said that animals might understand some words and distinguish certain objects. But he pointed out:

“Even if we consider all the possibilities animals possess, isn’t it obvious—that humans have certain abilities that we cannot imagine developing from what humans and lower animals share, nor can we think they are merely an elevation of those common traits?”

“…No matter how similar the impressions or organs of sense are, no matter how dependent on organic activity—meaning, no matter how close physiologically—the perception of the eye is different from that of the ear, the head, or the tongue, and indicates a special gift or faculty that is not included in the latter. Rational action and the operation of lower faculties are not the same.”

“…Having some elements in common does not prove they belong to the same order, nor does it make one develop into the other. If the soul’s eye—that higher rationality that allows us to perceive the universe—cannot introspect itself, clearly discern its own nature and operation, then we should not forget its function, deny its inherent superiority, or equate it with the lower faculties we can observe.”

The thing that enables us to understand all things must be inherently superior to anything it understands.

One audience member particularly noted that Ballard’s gaze conveyed meaning perfectly, without misunderstanding:

“The most interesting moment at this meeting was Mr. Ballard’s use of gestures to describe how his mother told him he was going to a distant place to study, where he would read books, write letters, and send them back; and pantomime of a hunter shooting a squirrel, accidentally shooting himself. Mr. Ballard’s gestures, along with his facial expressions and eye contact, conveyed his meaning flawlessly. As one member said, the expression of the eyes is a language that cannot be misunderstood.”

Consider these two sentences:

• “The thing that enables us to understand all things must be inherently superior to anything it understands.”
• “The expression of the eyes is a language that cannot be misunderstood.”

In summary:

• Language without rational ability cannot achieve understanding.
• Only when rational ability exists can language unlock understanding.
• Fully realized understanding surpasses language itself.

Large language models place language at the forefront, establishing a primitive form of rationality solely through logical inference. But this rationality has proven flawed, prone to hallucinations at many rough edges of knowledge.

Rational ability has never truly existed. Therefore, language cannot elevate to understanding through rationality.

In his work with deaf-mutes, Professor Porter discovered that genuine rational ability must precede language; only then can language unlock understanding—understanding is the product of true rational ability and language working together.

“The expression of the eyes is a language that cannot be misunderstood.”

In other words, the expression of the eyes is the perfect form of understanding—no language needed.

Large language models, by placing language before true rational ability, can never truly reach understanding.

If understanding truly surpasses language—as revealed in the Smithsonian speech 144 years ago—then it should be easy to find evidence today.

I myself have experienced this through medical study and practice. Throughout undergraduate pre-med courses and most of medical school, deductive logic was the tool organizing vast medical knowledge. In clinical stages, the art of medicine—signs, emotions, humanistic knowledge—began to develop. Later, during residency or early practice, with the accumulation of extensive experience, understanding finally arrived. All parts interconnected in a vast, complex network, allowing experienced doctors to provide comprehensive patient care.

When two surgeons handle a complex head and neck cancer surgery or trauma, or when nurses working with them communicate with just a glance—full understanding is conveyed, actions triggered—because everyone present has achieved understanding, surpassing logical deduction and primitive reasoning in early medical education.

Thus, the eyes provide an intuitive grasp of reality, built on shared understanding, which in turn arises from rationality present when language is involved.

Large language models—and small ones—remain stuck in the middle ground. They can simulate reasoning but lack true rational ability, eyes, or understanding.

Ballard’s test: an entity must demonstrate rationality without language to truly possess understanding.

This is a known flaw, a poor starting point. The initial goal of AI research was to produce genuine rational ability, but that has never been achieved. The field then shifted to language-first approaches—because they are easier.

This “poor starting point” has led to a “parameter trap”: brute-force language processing driven by countless energy-consuming chips, which has become an extremely ironic bottleneck.

As I emphasized in my conversation with Sebastian Siemiatkowski, founder of Klarna, the future lies in compression—prioritizing “System 2” reasoning, digesting redundant information and the relatively limited set of human queries, thereby greatly reducing computational demands.

This new approach rejects the pursuit of singularity through mutual conversation of language models in an infinite mirror—an aim of resource waste with no clear direction, and impossible without economic feasibility.

Frontier research like Google’s AlphaGeometry and Meta’s Coconut is shifting toward this “rationality-first” architecture, but fundamentally, they are rediscovering what the Smithsonian already presented 144 years ago: Language is the output of understanding, not the engine of rationality.

This trillion-dollar myth of “computing power” may be broken by a return—a return to the silence of pre-linguistic rationality—like the full-spectrum rational mind of the deaf-mute, whose silent thoughts reached the stars long before they found words to express them.

Silicon Valley

Earlier, I mentioned another article on the same page. Its relevance probably exceeds anything imagined in the 1880s.

The article is titled: “Wealth in San Francisco: A City Full of Speculators and Rapid Riches.”

It was written in San Francisco on June 1, 1880, and published in The New York Times only on June 19.

There’s a French saying: “The more things change, the more they stay the same.” It comes to mind now.

“San Francisco’s ‘hard times’ might mean ‘fairly prosperous days’ in eastern cities—referring not to extravagance and waste, but to a lack of reckless spending and poverty.”

At that time, California was a paradise for small investors. To satisfy speculative desires, a unique open bidding system emerged: for just $50, you could buy a share of a mine, at $1 per share, or two nickels, or any amount at different prices.

When a stock “boomed,” it seemed to only ignite the impulse to “try again.” It fueled the same speculative frenzy in San Francisco—people chasing after the opportunities lost by the wealthy; “prosperity” came with market losses, and when it faded, prices returned to normal.

The article’s ending hits hard against today’s reality:

It seems San Francisco has become accustomed to the idea that wealth must be gained instantly. After the big bubble burst in Virginia City, they seem unwilling to recover and seek wealth through manufacturing, trade, or agriculture. Almost the entire city is filled with speculation; if a new gold mine as large as Nevada were discovered nearby, stock prices would soar again to absurd heights, and San Francisco would relive those days of rapid wealth—and then suffer all the consequences of the past two years again.

In “The Core Mark of a Bubble: Greed on the Supply Side,” I outlined this alarming tendency originating from the San Francisco Bay Area: relentless speculation driving investments far beyond what any reasonable demand could absorb over any realistic timeframe.

Reading such old newspapers allows us to interpret today’s events from a different perspective. Will Silicon Valley “experience those days of rapid wealth again and then suffer everything,” as it has repeatedly? Or will it break the pattern—no one can say for sure. I hope this article has been helpful to you.

Finally, I recommend Midjourney, a tool for generating images and videos.

It’s incredibly fun and thought-provoking. Unleash your creativity!