A decade ago, in the Fall 2015 issue of this review, William Bryk ’19 penned a provocative essay titled “Artificial Superintelligence: The Coming Revolution.” Drawing on Arthur C. Clarke’s third law—that any sufficiently advanced technology is indistinguishable from magic—Bryk invited us to imagine a “computational beast” with an IQ of 34,597, capable of integrating the Library of Congress into a comprehensive analysis of humanity before we could blink.
Today, standing in 2026, that “beast” is no longer a science-fiction monster lurking on the horizon. It is the infrastructure of our daily lives. Looking back at Bryk’s vision from ten years ago, we find ourselves in a world that is simultaneously exactly what he predicted and nothing like what we expected.
The Successes: The Death of the Specific Task
Bryk’s most prescient observations concerned the transition from “narrow” to “general” capability. In 2015, he noted that “Siri might respond intelligently to your requests for directions, but if you ask her to help with your math homework, she’ll say ‘Starting Facetime with Matt Soffer.’”
He predicted a future where “Siri will listen and speak more eloquently than you do” and “IBM’s Watson will analyze your medical records and become your personal, all-knowing doctor.” In 2026, the Agentic Revolution has largely fulfilled this. We no longer “interact” with AI; we “delegate” to it. The Large Language Models (LLMs) that exploded in the early 2020s bypassed the rigid programming Bryk’s skeptics pointed to. Instead of being “good at very specific tasks,” our current systems possess a fluid, reasoning-capable architecture that would have shocked a 2015 researcher.
The Misses: The Path to the “Mega-Brain”
Where the 2015 article missed the mark was in the method of achievement. Bryk outlined two main routes to Human Level Machine Intelligence (HLMI): complex machine learning algorithms and “Whole Brain Emulation.” He wrote, “If we can image every neuron in a brain… we would have a human-level artificial intelligence.”
As of 2026, Whole Brain Emulation remains a stagnant, albeit fascinating, subfield. The “algorithmic” path—specifically the Transformer architecture and its descendants—didn’t need to copy the biological “wetware” of the brain. Instead, we found a mathematical shortcut. We didn’t need to simulate the neuron; we just needed to simulate the attention.
Furthermore, Bryk cited a 2012 survey where the median prediction for HLMI was the year 2040. By 2026 standards, that timeline looks incredibly conservative. With the advent of recursive self-improvement—AI systems now writing 70% of their own optimization code—many experts at the Harvard Data Science Initiative suggest we have already hit “soft” HLMI. We didn’t have to wait for 2040; the “Intelligence Explosion” Bryk feared is already simmering in data centers from Virginia to Iceland.
The Singularity and the “Chipotle” Problem
The most chilling part of Bryk’s 2015 piece was a reimagining of Fredric Brown’s short story, “Answer,” where hackers start an ASI and go to Chipotle, only to return to a machine that has declared itself a God.
While we haven’t seen a “God in a garage” yet, the “Recursive Self-Improvement” Bryk described is no longer a thought experiment. The “Technological Singularity” is the dominant theme of 2026 geopolitics. When Bryk wrote, “the rate at which technology advances is advancing,” he was describing the Second Half of the Chessboard. We are now living on that half. The gap between a system that can pass the Bar Exam and a system that can design a more efficient fusion reactor turned out to be less than three years—not the decades previously assumed.
Conclusion: Are We Ready?
Bryk concluded his essay with a haunting question: “Are we ready to take the risk?” He cited an 18% chance of “extremely bad” (existential) outcomes. In 2026, that 18% is the subject of intense international regulation and the “Alignment Problem” that occupies the brightest minds at Harvard’s Berkman Klein Center.
We have the “mega-brain.” It may not have an IQ of exactly 34,597, but it can read the entire internet in the time it takes to type “Enter.” Ten years ago, Bryk called it a “ticket to understanding.” Today, it’s our co-pilot. We haven’t been “desensitized” to the magic as he feared; rather, we have discovered that the magic is much more complicated, and much more human, than we ever imagined.
Cited Sources and References
- Bryk, W. (2015). “Artificial Superintelligence: The Coming Revolution.” Harvard Science Review, 18(1), 22-25.
- Turing, A. (1936). “On Computable Numbers, with an Application to the Entscheidungsproblem.” Oxford Journal.
- Müller, V. C., & Bostrom, N. (2014). “Future Progress in Artificial Intelligence: A Survey of Expert Opinion.” Synthese Library.
- Kurzweil, R. (2005). The Singularity is Near. Penguin Books.
- Good, I. J. (1965). “Speculations Concerning the First Ultraintelligent Machine.” Academic Press.
- Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford University Press.
- Brown, F. (1954). “Answer.” Angels and Spaceships. Dutton.
10 Years Ago HSR Wrote “Artificial Superintelligence: The Coming Revolution”: What Did We Get Right, What Did We Get Wrong?
A decade ago, in the Fall 2015 issue of this review, William Bryk ’19 penned a provocative essay titled “Artificial Superintelligence: The Coming Revolution.” Drawing on Arthur C. Clarke’s third law—that any sufficiently advanced technology is indistinguishable from magic—Bryk invited us to imagine a “computational beast” with an IQ of 34,597, capable of integrating the Library of Congress into a comprehensive analysis of humanity before we could blink.
Today, standing in 2026, that “beast” is no longer a science-fiction monster lurking on the horizon. It is the infrastructure of our daily lives. Looking back at Bryk’s vision from ten years ago, we find ourselves in a world that is simultaneously exactly what he predicted and nothing like what we expected.
The Successes: The Death of the Specific Task
Bryk’s most prescient observations concerned the transition from “narrow” to “general” capability. In 2015, he noted that “Siri might respond intelligently to your requests for directions, but if you ask her to help with your math homework, she’ll say ‘Starting Facetime with Matt Soffer.’”
He predicted a future where “Siri will listen and speak more eloquently than you do” and “IBM’s Watson will analyze your medical records and become your personal, all-knowing doctor.” In 2026, the Agentic Revolution has largely fulfilled this. We no longer “interact” with AI; we “delegate” to it. The Large Language Models (LLMs) that exploded in the early 2020s bypassed the rigid programming Bryk’s skeptics pointed to. Instead of being “good at very specific tasks,” our current systems possess a fluid, reasoning-capable architecture that would have shocked a 2015 researcher.
The Misses: The Path to the “Mega-Brain”
Where the 2015 article missed the mark was in the method of achievement. Bryk outlined two main routes to Human Level Machine Intelligence (HLMI): complex machine learning algorithms and “Whole Brain Emulation.” He wrote, “If we can image every neuron in a brain… we would have a human-level artificial intelligence.”
As of 2026, Whole Brain Emulation remains a stagnant, albeit fascinating, subfield. The “algorithmic” path—specifically the Transformer architecture and its descendants—didn’t need to copy the biological “wetware” of the brain. Instead, we found a mathematical shortcut. We didn’t need to simulate the neuron; we just needed to simulate the attention.
Furthermore, Bryk cited a 2012 survey where the median prediction for HLMI was the year 2040. By 2026 standards, that timeline looks incredibly conservative. With the advent of recursive self-improvement—AI systems now writing 70% of their own optimization code—many experts at the Harvard Data Science Initiative suggest we have already hit “soft” HLMI. We didn’t have to wait for 2040; the “Intelligence Explosion” Bryk feared is already simmering in data centers from Virginia to Iceland.
The Singularity and the “Chipotle” Problem
The most chilling part of Bryk’s 2015 piece was a reimagining of Fredric Brown’s short story, “Answer,” where hackers start an ASI and go to Chipotle, only to return to a machine that has declared itself a God.
While we haven’t seen a “God in a garage” yet, the “Recursive Self-Improvement” Bryk described is no longer a thought experiment. The “Technological Singularity” is the dominant theme of 2026 geopolitics. When Bryk wrote, “the rate at which technology advances is advancing,” he was describing the Second Half of the Chessboard. We are now living on that half. The gap between a system that can pass the Bar Exam and a system that can design a more efficient fusion reactor turned out to be less than three years—not the decades previously assumed.
Conclusion: Are We Ready?
Bryk concluded his essay with a haunting question: “Are we ready to take the risk?” He cited an 18% chance of “extremely bad” (existential) outcomes. In 2026, that 18% is the subject of intense international regulation and the “Alignment Problem” that occupies the brightest minds at Harvard’s Berkman Klein Center.
We have the “mega-brain.” It may not have an IQ of exactly 34,597, but it can read the entire internet in the time it takes to type “Enter.” Ten years ago, Bryk called it a “ticket to understanding.” Today, it’s our co-pilot. We haven’t been “desensitized” to the magic as he feared; rather, we have discovered that the magic is much more complicated, and much more human, than we ever imagined.
Cited Sources and References
Related Articles
Federal Science Funding Moves Toward Scientists
Federal Science Funding shifts are pushing more scrutiny toward individual
AI Adoption Challenges Facing U.S. Firms
AI Adoption Challenges in U.S. firms center on technical debt,
wildfire water quality risks in the West
Western wildfire water quality risks are rising. Recent research links
Fusion Digital Twins Move From Models to Tests
epigenome reprogramming and healthspan impacts
AI Battery Screening for Solid-State Materials