AI mirrors the essence of our consciousness through a concept known as temperature. This “temperature” setting governs the likelihood of specific responses to inquiries. When the temperature is set to its highest, the reply to a question remains consistent. Conversely, the response becomes entirely random at zero, drawing from a pool of options, each with equal chances of being chosen.
Consider the difference between a simple chat response to “Why does an apple fall from a tree?” and Newton’s profound realization of gravity sparked by observing an apple’s descent. This distinction highlights the depth of human insight that AI endeavors to capture.
In the case of LLMs, one task has been tried that might be thought of as weakly satisfying; something that could be called creative is generating data from its human-produced database that could be used to train another computer on tasks with fewer ontologies that those represented by its enormous database.
In a recent cover story in Nature, LLMs were asked to generate data relevant to dogs.
I wondered if this was a joke! After many iterations, they had multiple pictures of one breed: Golden Retrievers. This data is inherently flawed… as we all know, there are hundreds of dog breeds the LLMs could have presented.
Another ludicrous example: after several iterations, it created a list of jackrabbits when asked to develop data pertinent to medieval architecture. Hmmm… that’s a head-scratcher!
While it may appear humorous on the surface, serious underlying issues are at play. We have invested trillions of dollars in AI-related companies, driven by the anticipation of significant returns, underscoring how much is at stake. Large Language Models (LLMs) and AI should transcend routine tasks relying on human oversight. At the very least, AI has the potential to provide solutions to critical challenges that could otherwise hinder its advancement.
False Information Breeds False Narratives
In a recent YouTube interview, Eric Schmidt, the technologist and former CEO of Google, humorously pointed out the significant resources that AI requires, suggesting we maintain a strong relationship with our Canadian neighbors. The rationale behind this is their abundant oil and gas reserves, which play a crucial role in powering AI advancements. Alongside Schmidt, his co-authors Henry Kissinger and Daniel Huttenlocher, both esteemed AI researchers and administrators at MIT, underscore the pressing need for collaboration in this field, particularly given Schmidt’s academic background with a Ph.D. from MIT.
The central argument in Schmidt’s book requires significant reevaluation for several compelling reasons. Contrary to its main thesis, AI has already surpassed human capabilities in specific domains, such as gaming, and is poised to drive one of the most remarkable transformations in human history. This perspective is particularly surprising coming from Schmidt, who has witnessed rapid technological advancements, notably AI, where resource utilization is exponentially higher than the benefits gained. His acknowledgment of the situation in Canada suggests he views the challenges as manageable, but the evidence indicates otherwise. Chess, while impressive, represents just a fraction of the complexity involved in achieving Artificial General Intelligence (AGI). As previously mentioned, developing a solution for chess is still far beyond the reach of current computer technologies, both in terms of calculation power and energy consumption.
Relying on AI to solve complex problems resembles a cat chasing its tail—endless but unproductive. While human ingenuity may temporarily prolong our advancements, the inevitable truth remains: we face significant long-term challenges. Recently, Nvidia highlighted these issues by announcing delays in their next-generation LLM chips. Even the staunchest proponents of AI must acknowledge that substantial progress is still needed, with numerous generations of chips ahead of us. The pressing concern is not whether we can keep innovating but whether we can manage the escalating complexity of future chips or secure enough energy to support their production and functioning.
Can Greater Sophistication Change The Game?
Many still perceive AI as merely a tool for enhancing productivity in routine tasks. However, the true potential of Artificial General Intelligence (AGI) goes far beyond this; it is poised to make groundbreaking contributions to the sciences and innovate entirely new mathematics branches. This capability could be instrumental in tackling challenges ranging from the Riemann hypothesis to intricate combinatorial problems. Renowned mathematician Paul Erdős and others have emphasized the urgent need for advanced mathematical frameworks to address issues involving group dynamics in networks, among numerous other areas. We face a critical juncture as we contemplate the escalating power and computational demands of increasingly sophisticated AI. AI is undoubtedly on the brink of becoming more ubiquitous, yet we must recognize that we are already in the later stages of developing this enhanced sophistication.
Why do we persist in believing what amounts to magic?
One key issue is that we—the collective West—have deviated sharply from the path the Declaration of Independence laid out through the early 1970s. Today, the West is one of the most materialistic societies in history, with America at the forefront. Science has largely eclipsed spirituality, leaving a significant void. Research, including my Ph.D. thesis, has consistently demonstrated a troubling link between blind acceptance of narratives and a propensity to embrace authoritarian leadership. Furthermore, this overly pragmatic worldview stifles creativity and innovation.
It’s striking how we often condemn nations like China and Russia for their restrictions on individual liberties. While it’s indisputable that these countries fall short of embracing essential democratic principles, we must also recognize that they stifle freedom of thought—the foundation that Thomas Jefferson championed as vital for scientific progress and societal health. Alarmingly, both China and Russia appear to be outpacing us in these areas.
Numerous credible sources indicate that China has significantly outpaced the United States in STEM fields. According to the Netherlands’ bibliometric evaluations of global universities, China occupies the top ten positions in STEM. In contrast, the highest-ranked U.S. university, MIT, barely scores above 40, which reflects an average across the two categories constituting STEM. Remarkably, Harvard doesn’t even rank in the top 100, and no other U.S. institution has cracked the top 50. Additionally, findings from Australia highlight that China excels in 90% of scientific disciplines. The only area where the U.S. maintains an edge over China is quantum computing—an outcome driven by China’s recent hesitance to publicly share its research findings. A recent IEEE report emphasizes this situation:
“Now, scientists in China have tested two quantum computers on what they say are more challenging tasks than Sycamore faced, and they have shown faster results. They note their work points to “an unambiguous quantum computational advantage.”
In one study, the researchers experimented with Zuchongzi, which used 56 superconducting qubits on a task whose solutions are random instances, or samples, from a given spread of probabilities. They found that Zuchongzi had completed the sampling task in 1.2 hours, which they estimated would take at least 8.2 years. They also noted that this sampling task was tens to hundreds of times more computationally demanding than what Google used to establish a quantum advantage with Sycamore. In another study, the scientists tested Jiuzhang 2.0, a photonic quantum computer, using Gaussian boson sampling, a task where the machine analyzes random data patches. Using 113 detected photons, they estimated Jiuzhang 2.0 could solve the problem roughly 1024 faster than classical supercomputers.
Although the sampling task used in experiments with Zuchongzi has no known practical value, the Gaussian boson sampling problem on which Jiuzhang 2.0 was tested potentially has many valuable applications, such as identifying which pairs of molecules are the best fits for each other. As such, this work may have quantum chemistry applications in simulating vital molecules and chemical reactions, says physicist Chao-Yang Lu at the University of Science and Technology of China in Hefei, a co-author on both studies.”
China’s advancements in quantum computing have outpaced those of the U.S., both in practical applications and theoretical research. In a future blog post, I will explore how nearly two centuries of Jeffersonian principles—equality, spirituality, and democracy—have fostered an incredibly creative society in American history. Notably, the absence of spirituality from our discussions is a fundamental barrier to unlocking creativity. When was the last time a U.S. president genuinely invoked God during a significant address? The answer takes us back to John Kennedy. Numerous quotes are available online to explore this further. Clearly, America has much to rediscover, and AI is not the right guide for this journey.
Final Thoughts
In any AI application, the primary objective is to convert an initial dataset and its parameters into more sophisticated datasets, establishing connections between the original data and elevated parameters. Success hinges on the AI’s accuracy in delivering results across various scenarios. This ultimately depends on the computational effort needed to achieve those results. Human involvement is important at nearly every stage, from defining the database to outlining what constitutes success. Philosophically, while setting the groundwork for a language game may be relatively simple, unraveling the paradoxes tied to that game’s regulations poses a significantly greater challenge, particularly in how those paradoxes intersect with conscious experience—an area where AI is unlikely to excel. In chess, achieving victory in a match represents a basic accomplishment, whereas accurately solving any given position presents a far more formidable challenge that remains beyond our current capabilities.
