Author: OpenAI CEO Sam Altman
Recently, OpenAI CEO Sam Altman published an article on his personal blog titled "The Gentle Singularity." In this article, he presents a more realistic and thought-provoking determination: the "singularity moment" of AI has arrived, but it is happening quietly in a gentle and gradual manner.
Do you feel that the world is undergoing some profound yet subtle changes? It seems that we have quietly entered a brand new era. The "technological singularity" depicted in science fiction—that moment when machine intelligence surpasses humanity and the world order is completely transformed—has not arrived with a violent explosion. Instead, it is like a gentle tide, slowly and steadily washing over our ankles.
This great takeoff has already begun.
Yesterday's magic is today's routine.
Not long ago, the idea of robots walking the streets and AI becoming our all-day conversation partners was just a distant fantasy. Today, while these scenarios have not yet fully realized, we have already created "digital brains" that are far smarter than humans in many fields. For example, GPT-4, which we are familiar with, contains scientific breakthroughs that are a result of the wisdom of countless scientists, and this is just the beginning.
The most peculiar aspect of this transformation lies precisely in its "non-peculiarity."
We have adapted too quickly. We have swiftly changed from being amazed that AI can write a beautiful poem to expecting it to create a great novel; we have transitioned from admiring AI's ability to assist doctors in making life-saving diagnoses to desiring it to directly develop cures for diseases.
Miracles are rapidly transforming into everyday tools that we take for granted. This is the essence of the 'gentle singularity': it does not rely on disruptive shocks, but instead continuously integrates future miracles into our present, subtly reshaping the world.
Looking ahead to the next few years, this trend will become clearer:
By 2025, AI 'agents' that can independently perform complex cognitive tasks (such as coding) will reshape the software Industry.
By 2026, systems that can discover entirely new scientific insights may emerge.
By 2027, robots capable of performing tasks in the real world may arrive by our side.
Accelerating Engines: A Self-Evolving Future
You might ask, why is progress so rapid? Because we have ignited several powerful 'self-accelerating engines.'
The core engine is using AI to create stronger AI. Today, scientists around the world are utilizing AI tools to increase their research efficiency two to three times. This means that the speed at which we discover new algorithms, explore new materials, and construct new theories is growing exponentially. It's like a primitive 'recursive self-improvement'—we use today's intellectual achievements as a lever to unlock tomorrow's greater wisdom. Research that used to take ten years to complete may only take one year, or even one month, in the future.
In addition, there are two huge flywheels spinning:
The economic flywheel: The amazing economic value created by AI is attracting an unprecedented influx of capital for the construction of larger data centers and computing power facilities. More money means a stronger "brain."
The physical flywheel: The era of robots manufacturing robots is not far away. Imagine that when we produce the first batch of one million humanoid robots, they can take over the entire supply chain, from mining minerals, driving trucks to operating factories, to build more robots, more chip factories, and data centers. What a magnificent picture of exponential growth that will be!
Ultimately, when the production of data centers is fully automated, the cost of "intelligence" will approach the cost of electricity. It will become a cheap and ubiquitous social foundational resource, just like today's electricity.
Life in the 2030s: What is changing, what is eternal?
What will our world look like in the 2030s?
On one hand, at the most fundamental human level, life will remain the same. We will still love our families, enjoy gatherings with friends, express our creativity, swim in lakes, and feel the sunshine. The human pursuit of love, beauty, and connection is eternal.
On the other hand, at equally important social levels, the world will be unprecedentedly different. The two long-standing shackles of human civilization, intelligence and energy, will be completely broken. By then, the things an ordinary person can accomplish will far exceed what we can imagine today.
So, what about work? Many people worry about their jobs being replaced. This is normal, but history tells us that human adaptability and creativity are limitless.
Let's do a thought experiment: if a farmer from a thousand years ago were to observe our work today, how would he feel? Seeing us sitting in front of computers typing away, he would likely think we are playing some "fake game," as we are already free from worry about food and enjoy a luxurious life that he cannot imagine.
I hope that in a thousand years, humanity will look back at the "work" we take pride in today and also consider it very "fake." Their new "work," which we cannot imagine today, will surely be filled with unparalleled meaning and satisfaction for them. Because humans have an advantage that AI can never match: we are inherently concerned with each other's thoughts and creations; we live for "people," not for "machines."
Our choice: two paths to a bright future.
This journey to superintelligence comes with immense opportunities, but it also poses severe challenges. To ensure we reach that beautiful future instead of falling into chaos, we must take the right two key paths:
Step one: solve the "alignment" problem and equip AI with a "compass."
"Alignment" means what? Simply put, it ensures that the ultimate goals and behaviors of AI are aligned with our overall long-term well-being as humans. The algorithms of social media serve as a perfect counterexample: they are very "smart" and know how to exploit your brain's weaknesses to keep you refreshing, but this clearly contradicts your long-term desires for efficient living and deep thinking. We must ensure that future superintelligence helps us become better versions of ourselves, rather than exploiting our weaknesses. We must equip this powerful "brain" with a reliable "moral compass."
Step two: promote inclusivity, decentralizing superintelligence.
After addressing the safety issues, we must do our utmost to make super intelligence affordable, accessible, and not monopolized by any minority individuals, companies, or nations. This power is too great; it must become the wealth of all humanity. We must believe in the resilience and creativity of society itself. As long as tools are placed in everyone's hands and a common baseline is set for everyone to adhere to, the collective wisdom of humanity can harness it, leveraging its strengths while avoiding its weaknesses.
Conclusion: An era belonging to the 'visionaries'.
We (including OpenAI and the entire industry) are building a 'global brain' for this world. It will be deeply personalized, allowing everyone to use it with ease. By then, what limits us will no longer be the technological barriers, but whether we have good enough ideas.
In the past, those who had brilliant ideas but struggled to bring them to life were often jokingly referred to as 'idea kings.' Now, they are about to welcome their moment in the spotlight.
We are moving towards an era where 'intelligence is as cheap as air.' This may sound a bit crazy, but think back to the predictions we made in 2020 about today; weren't they crazier than our predictions today about 2030?
The road ahead is mostly illuminated, and darkness is quickly retreating.
May we sail smoothly, steadily, and without disturbance towards that bright future of super intelligence.
Editor/melody
