In the field of artificial intelligence, the convergence of faith and technology has sparked a fierce debate surrounding the validity and future development of the "Scaling Hypothesis". This article explores the rise, divergence, and potential impact of this "bitter religion", revealing the complex relationship between faith and science. The AI community has become embroiled in a debate about its future and whether it has sufficient scale to create the doctrine of God. This article is based on a piece written by Mario Gabriele, compiled and translated by Block unicorn.
(Background: Musk's xAI completes $6 billion Series C financing, with Nvidia, BlackRock, a16z, and other industry heavyweights participating)
(Additional context: Nvidia to launch humanoid robot computing platform "Jetson Thor" next year, the physical AI ChatGPT moment is approaching?)
The Holy War of Artificial Intelligence
"I would rather live my life as if there is a God and die to find out there isn't, than live my life as if there isn't and die to find out there is." — Blaise Pascal
Religion is an interesting thing. Perhaps because it is completely unprovable in any direction, or perhaps just like my favorite quote: "You can't use facts to argue against feelings."
The hallmark of religious belief is that as faith ascends, it accelerates in an almost unbelievable way, to the point where the existence of God becomes almost impossible to doubt. When more and more people around you believe in it, how can you doubt a sacred being? When the world rearranges itself around a doctrine, where is there room for heresy? When temples and cathedrals, laws and norms are all arranged according to a new, unshakable gospel, where is there space for opposition?
When the Abrahamic religions first appeared and spread across continents, or when Buddhism spread from India across Asia, the immense momentum of belief created a self-reinforcing cycle. As more people converted and built complex theological systems and rituals around these faiths, it became increasingly difficult to question these basic premises. In an ocean of credulity, it was not easy to become a heretic. Grandiose cathedrals, intricate religious texts, and thriving monasteries all served as physical evidence of the sacred presence.
But history also tells us how easily such structures can collapse. As Christianity spread to the Scandinavian peninsula, the ancient Nordic beliefs crumbled in just a few generations. The religious system of ancient Egypt persisted for thousands of years, only to vanish when new, more enduring faiths arose and appeared in larger power structures. Even within the same religion, we have seen dramatic schisms - the Reformation tore apart Western Christianity, and the Great Schism led to the split between the Eastern and Western churches. These divisions often began with seemingly trivial doctrinal differences, only to evolve into completely different belief systems.
The Scriptures
God is a metaphor that transcends all intellectual thought levels. It's that simple. — Joseph Campbell
Believing in God is simply religion. Perhaps creating God is no different.
From the beginning, optimistic AI researchers have imagined their work as a form of creationism - God's creation. In the past few years, the explosive development of large language models (LLMs) has further cemented the believers' conviction that we are on a sacred path.
It has also validated a blog post written in 2019. Although it was only known to those outside the AI field until recently, the "Bitter Lesson" by Canadian computer scientist Richard Sutton has become an increasingly important text in the community, evolving from esoteric knowledge into a new, all-encompassing religious foundation.
In 1,113 words (every religion needs a sacred number), Sutton summarizes a technical observation: "The biggest lesson that can be learned from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin." The progress of AI models has benefited from the exponential increase in computational resources, riding the massive wave of Moore's Law.
At the same time, Sutton points out that much of the work in AI research has focused on optimizing performance through specialized techniques - increasing human knowledge or narrow tools. While these optimizations may be helpful in the short term, Sutton sees them as a waste of time and resources, akin to adjusting the fins or trying new wax on a surfboard when a massive wave is coming.
This is the foundation of what we call the "Bitter Religion". It has only one commandment, commonly referred to in the community as the "Scaling Hypothesis": Exponential growth in computation drives performance; everything else is folly.
The Bitter Religion has expanded from large language models (LLMs) to world models, and is now rapidly spreading through untransformed temples such as biology, chemistry, and embodied intelligence (robotics and self-driving vehicles).
Here is the English translation:However, as the Sutton doctrine has spread, the definition has also begun to change. This is the hallmark of all vibrant and lively religions - debate, extension, commentary. The "scaling law" no longer just means scaling compute (the Ark is not just a boat), it now refers to various methods aimed at boosting transformer and compute efficiency, with some tricks involved.
Now, the classic encompasses attempts to optimize every part of the AI stack, from techniques applied to the core models themselves (model merging, Mixture of Experts (MoE), and knowledge distillation) all the way to generating synthetic data to feed these ever-hungry gods, with a lot of experimentation in between.
The Warring Sects
Recently, a question has arisen in the AI community with a crusading spirit, and that is whether the "bitter religion" is still correct.
This week, Harvard, Stanford, and MIT published a new paper titled "The Scaling Law of Precision", which has sparked this conflict. The paper discusses the end of quantitative efficiency gains, quantification being a series of techniques to improve AI model performance and greatly benefit the open-source ecosystem. In the post below, Tim Dettmers, a research scientist at the Allen AI Institute, outlines its significance, calling it "the most important paper in a long time". It represents the continuation of a dialogue that has been heating up over the past few weeks, and reveals a noteworthy trend: the increasing entrenchment of two religions.
OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei belong to the same sect. Both are confident that we will achieve Artificial General Intelligence (AGI) in the next 2-3 years. Altman and Amodei can be said to be the two figures most reliant on the sanctity of the "bitter religion". All of their incentives tend towards over-promising, creating maximum hype, to accumulate capital in this game almost entirely dominated by economies of scale. If the scaling law is not the "Alpha and Omega", the first and the last, the beginning and the end, then what do you need $22 billion for?
Former OpenAI Chief Scientist Ilya Sutskever adheres to a different set of principles. He, along with other researchers (including many from within OpenAI, according to recent leaks) believe that scaling is approaching its limits. This group believes that maintaining progress and bringing AGI into the real world will necessarily require new science and research.
The Sutskever camp reasonably points out that the Altman camp's continued scaling doctrine is economically infeasible. As AI researcher Noam Brown asked, "After all, do we really want to train models that cost tens or hundreds of billions or trillions of dollars?" This doesn't even include the additional tens of billions in inference compute costs if we shift the scaling from training to inference.
But true believers are well-versed in their opponents' arguments. The missionary at your doorstep can easily handle your hedonistic trilemma. For Brown and Sutskever, the Sutskever camp points to the possibility of "inference-time compute" scaling. Unlike the situation so far, "inference-time compute" does not rely on larger compute to improve training, but rather uses more resources for execution.
When an AI model needs to answer your question or generate a piece of code or text, it can be given more time and compute. This is akin to shifting your attention from studying math to persuading the teacher to give you an extra hour and allow you to use a calculator. For many in the ecosystem, this is the new frontier of the "bitter religion", as teams are shifting from the orthodox pre-training to post-training/inference methods.
Pointing out the flaws in other belief systems, criticizing other doctrines without exposing one's own position, is indeed quite easy. So what is my own belief? First, I believe that the current batch of models will bring very high returns on investment over time. As people learn to work around the constraints and leverage the existing APIs, we will see truly innovative product experiences emerge and succeed.
We will transcend the reification and incremental stages of AI products. We should not view this as "Artificial General Intelligence" (AGI), as that definition has conceptual defects, but rather as "Minimum Viable Intelligence", customizable to different products and use cases.
As for achieving Artificial Superintelligence (ASI), that will require more structure. Clearer definitions and delineations will help us more effectively discuss the trade-offs between the potential economic value and economic costs it may bring. For example, AGI may provide economic value to a subset of users (just a localized belief system), while ASI may exhibit unstoppable compounding effects and transform the world, our belief systems, and our social structures. I don't believe that scaling transformers alone can achieve ASI; but alas, as some might say, this is just my atheistic belief.
The Lost Faith
The AI community cannot resolve this holy war in the short term; there are no factual grounds to be presented in this emotional struggle. Instead, we should turn our attention to what the AI community's questioning of its faith in the scaling law means. The loss of faith could trigger a chain reaction, going beyond large language models (LLMs), impacting all industries and markets.
It must be pointed out that in most areas of artificial intelligence/machine learning, we have not yet fully explored the scaling law; there will be more miracles in the future. However, if real doubts quietly emerge, it will become more difficult for investors and builders to maintain the same high confidence in the ultimate performance status of "early curve" categories such as biotechnology and robotics.
In other words, if we see large language models start to slow down and deviate from the chosen path, the belief systems of many founders and investors in adjacent fields will collapse.
Whether this is fair is another question.
There is a view that "general artificial intelligence" naturally requires greater scale, so the "quality" of specialized models should be exhibited on a smaller scale, so that they do not easily encounter bottlenecks before providing practical value. If a model in a specific field only captures a portion of the data, and therefore only requires a portion of the computing resources to achieve feasibility, shouldn't it have enough room for improvement?
This seems intuitively reasonable, but we repeatedly find that the key is often not in this: including relevant or seemingly unrelated data can often improve the performance of seemingly unrelated models. For example, including programming data seems to help improve broader reasoning capabilities.
In the long run, the debate over specialized models may be irrelevant. Anyone building ASI (superintelligence) will likely have the ultimate goal of an entity that can self-replicate, self-improve, and have infinite creativity in all domains.
Holden Karnofsky, former OpenAI board member and founder of Open Philanthropy, calls this creation "PASTA" (the process of automated scientific and technological progress). Sam Altman's original profit plan seems to rely on similar principles: "Build AGI, then ask it how to make money." This is apocalyptic AI, the ultimate fate.
The success of large AI labs like and has sparked enthusiasm in capital markets to support similar "OpenAI for X domain" labs, whose long-term goal is to build "AGI" around their specific vertical industry or field. This scale decomposition inference will lead to a normalization shift, away from OpenAI simulation, towards product-centric companies - a possibility I raised at Compound's 2023 annual meeting.
Unlike the apocalyptic model, these companies must demonstrate a series of progress. They will be companies built on scale engineering problems, not scientific organizations conducting applied research, with the ultimate goal of building products.
In the realm of science, if you know what you're doing, you shouldn't be doing it. In the realm of engineering, if you don't know what you're doing, you shouldn't be doing it. - Richard Hamming
Believers are unlikely to lose their sacred faith in the short term. As mentioned earlier, with the surge of religion, they have compiled a script of life and worship and a set of heuristic methods. They have built physical monuments and infrastructure, strengthening their power and wisdom, and demonstrating that they "know what they are doing".
In a recent interview, Sam Altman said this about AGI (the emphasis is mine):
This is the first time I feel we really know what to do. From now until building an AGI still requires a lot of work. We know there are some known unknowns, but I think we basically know what to do, and it will take some time; it will be very difficult, but it is also very exciting.
Judgment
In questioning "The Bitter Religion", the scaling law skeptics are liquidating one of the most profound discussions of the past few years. We have all engaged in such thinking in one form or another. What if we invented God? How quickly would that God appear? What if AGI (general artificial intelligence) really, irreversibly rises?
Like all unknown and complex topics, we quickly store our specific reactions in our brains: some are desperate that they will become irrelevant, most expect a mix of destruction and prosperity, and the last few expect humans to do what we do best, continue to find problems to solve and solve the problems we have created, achieving pure abundance.
Anyone with a major stake hopes to be able to predict what the world will be like for them if the scaling law holds true and AGI arrives within a few years. How will you serve this new god, and how will this new god serve you?
But what if the gospel of stagnation drives out the optimists? What if we start to think that even God may decline? In a previous article, "Robot FOMO, Scaling Laws, and Technology Forecasting", I wrote:
I sometimes wonder what would happen if the scaling law did not hold, which might be similar to the impact of revenue loss, slowing growth, and rising interest rates on many technology sectors. I also sometimes wonder if the scaling law is completely valid, which might be similar to the commercialization curve of many other pioneering fields and their value capture.
"The beauty of capitalism is that no matter what, we'll spend a lot of money to find out the answer."
For founders and investors, the question becomes: what's next? Potential candidates to become great product builders in each vertical are gradually becoming known. There will be more such people in each industry, but the story has already begun. Where will the new opportunities come from?
If scaling stagnates, I expect to see a wave of bankruptcies and mergers. The remaining companies will increasingly focus on engineering, an evolution we should anticipate by tracking talent flows. We've already seen some signs that is moving in this direction as it increasingly productizes its offerings. This transition will open up space for the next generation of startups to "overtake on the curve" by relying on innovative applied research and science, rather than engineering, in their attempts to blaze new trails and surpass existing enterprises.
The Lessons of Religion
My view on technology is that anything that appears to have obvious compounding effects usually doesn't last very long, and a common perception that people have is that any business that appears to have obvious compounding effects tends to develop at a much slower rate and scale than expected.
The early signs of religious schism often follow predictable patterns that can serve as a framework to continue tracking the evolution of "The Bitter Religion".
It usually starts with the emergence of competing interpretations, whether for capitalistic or ideological reasons. In early Christianity, different views on the divinity of Christ and the nature of the Trinity led to schisms and radically different biblical interpretations. In addition to the schisms we've already mentioned in AI, there are other emerging fissures. For example, we see some AI researchers rejecting the core orthodoxy of transformers and turning to other architectures such as State Space Models, Mamba, RWKV, Liquid Models, etc. While these are still just weak signals, they indicate the sprouting of heretical thoughts and a willingness to rethink the field from first principles.
Over time, the impatient pronouncements of prophets also sow distrust. When religious leaders' predictions fail to materialize, or divine intervention does not arrive as promised, it plants seeds of doubt.
The Millerite movement once predicted Christ's return in 1844, but when Jesus did not arrive as scheduled, the movement collapsed. In the tech world, we often quietly bury failed prophecies and allow our prophets to continue painting optimistic, long-cycle future versions, even as their deadlines repeatedly miss (hey, Elon). However, if not supported by continual improvements in the underlying model performance, the faith in Moore's Law could also face a similar collapse.
A corrupt, bloated, or unstable religion is vulnerable to apostasy. The Protestant Reformation succeeded not just because of Luther's theological views, but because it emerged during a period of Catholic decline and turmoil. When cracks appear in the mainstream institutions, long-standing "heretical" ideas suddenly find fertile ground.
In the AI field, we may focus on smaller models or alternative approaches that achieve similar results with less computation or data, such as the work done by various Chinese corporate labs and open-source teams (like Nous Research). Those who break through the limits of biological intelligence and overcome barriers long thought insurmountable may also establish a new narrative.
The most direct and timely way to observe the transformation is to track the movements of practitioners. Before any formal schism, religious scholars and clergy often privately hold heterodox views while publicly conforming. The corresponding phenomenon today may be AI researchers who outwardly adhere to Moore's Law but secretly pursue radically different methods, waiting for the right moment to challenge the consensus or leave their labs in search of theoretically broader horizons.
The tricky part about the orthodoxy of religion and technology is that they are often partially correct, just not as universally correct as their most faithful adherents believe. Just as religions have embedded basic human truths within their metaphysical frameworks, Moore's Law clearly describes the real-world dynamics of neural network learning. The question is whether this reality is as complete and immutable as the current enthusiasm implies, and whether these religious institutions (AI labs) are agile and strategic enough to lead the zealots forward. Meanwhile, establishing the printing presses (chat interfaces and APIs) that can spread their knowledge.
The Endgame
"Religion is true to the common people, false to the wise, and useful to the rulers." - Lucius Annaeus Seneca
One potentially outdated view of religious institutions is that once they reach a certain scale, they become susceptible to the survival instincts of many human-run organizations, trying to survive in competition. In this process, they neglect truth and noble motivations (which are not mutually exclusive).
I've written about how capital markets become narrative-driven information cocoons, and incentive structures often perpetuate these narratives. The consensus around Moore's Law has an ominous similarity - a deeply entrenched belief system that is mathematically elegant and extremely useful for coordinating large-scale capital deployment. Like many religious frameworks, it may be more valuable as a coordination mechanism than as a fundamental truth.
