GuShiio.com Article source: AI Technology Base Camp
Image source: Generated by AI
Early this morning, OpenAI co-founder and CEO Sam Altman posted a blockbuster article “Three Observations” on his personal blog, once again announcing that the era of General Artificial Intelligence (AGI) is approaching and “Three major observationsAt the core, it puts forward “three major observations” about AI economics: the intelligence level of AI models has a logarithmic relationship with training resources; the cost of using a specific level of AI drops by about 10 times every 12 months; and linearly growing intelligence will create super-exponential social and economic value.
Sam Altman emphasized that the Scaling Law of Artificial Intelligence will continue to be effective, and “There is no reason to stop exponentially growing investment for the foreseeable future“, which has sent a strong positive signal to investors, which can be said to inject a” reassurance “into the AI investment field.
In addition, the article once again makes a more detailed prediction of AI Agents,”Eventually, such agents will be able to do most of the work that software engineers with years of experience in top companies can do, processing tasks that take no more than a few days.”
The following is the full text of the “Three Major Observations”:
Our mission is to ensure that General Artificial Intelligence (AGI) benefits all mankind.
Heralding the emergence of systems in the AGI* era, we believe it is crucial to understand the critical moment we are in. Although AGI is a term that is not yet clearly defined, we usually refer to a system that can solve increasingly complex problems in many fields at the human level.
(* The word AGI is used here for clear communication. We use the word “AGI” here to make it clear and avoid interpretation out of context, not to modify or reinterpret the definitions and processes of our existing partnership with Microsoft. We fully look forward to working with Microsoft for the long term.)
Human beings are natural tool creators, and the intrinsic driving force drives us to constantly understand and create, thereby making the world a better place. Each generation builds on the discoveries of its predecessors and builds more powerful tools-electricity, transistors, computers, the Internet, and the upcoming AGI.
Throughout history, although mankind’s innovation and progress have experienced twists and turns, they have always moved forward steadily, bringing unprecedented prosperity and improvement to almost all aspects of people’s lives.
In a sense, AGI is just another cornerstone that we are building together on the ladder of human progress. But in another sense, it heralds the beginning of a new era, and we have to admit that “this time is really different.” The prospects for economic growth before us are amazing, and we can even dream of a world where we can cure all diseases, have more time with our families, and realize our full creative potential.
Perhaps within ten years, everyone on earth will be able to achieve more than the most influential people today.
We continue to witness the rapid progress of artificial intelligence development. Here are three observations we make about the economics of artificial intelligence:
1. The intelligence level of an artificial intelligence model is roughly proportional to the amount of resources consumed (after taking the logarithm) to train and run it.These resources mainly include computing power, data and reasoning computing power required for training. It seems that as long as you continue to invest money, you can achieve sustained and predictable intelligence improvements; the law of scale for predicting this trend has been verified on multiple orders of magnitude.
2. The cost of using the same level of artificial intelligence drops approximately 10 times every 12 months.Lower prices have significantly contributed to the increase in usage. We can see this from the change in the token cost of GPT-4 in early 2023 to the token cost of GPT-4o in mid-2024-the price per token dropped by about 150 times during the period. Moore’s Law once changed the world at a rate that doubled every 18 months; the cost of artificial intelligence is falling even faster than that.
3. With the linear improvement of intelligence level, the growth of its social and economic value is essentially super-exponential. Extrapolating from this, we believe there is no reason to stop exponentially growing investment in artificial intelligence for the foreseeable future.
If these three observations continue to hold true, then the impact of artificial intelligence on society will be far-reaching.
We are starting to roll out AI Agents that will eventually work with us like virtual colleagues.
Taking software engineering agents as an example, we expect that such agents will be particularly important.ImagineEventually, such agents will be able to do most of the work that software engineers with years of experience in top companies can do, processing tasks that take no more than a few days.It may lack the most innovative ideas, require a lot of human supervision and guidance, and may perform well in some respects, but be surprisingly inferior in others.
Still, think of him as a real but inexperienced virtual colleague. Now imagine having 1000 such colleagues, or even 1 million. Imagine further that every knowledge work area is equipped with such an Agent.
In some ways, artificial intelligence may eventually be like a transistor in the economic field-a great scientific discovery that is not only easy to be applied on a scale, but alsopenetrate into almost every corner of the economy。We don’t pay much attention to transistors or transistor companies nowadays, but the benefits they bring have become widespread. We naturally expect computers, televisions, cars, toys and other equipment to continue to create miracles.
The world will not change dramatically overnight. Historical experience tells us that change is always gradual. In the short term, lifestyle will not change much, and people in 2025 will still live the same as in 2024. We still fall in love, start a family, argue online, go hiking in nature, etc.
However, the future is rushing towards us in a way that cannot be ignored, and the long-term impact of artificial intelligence on society and the economy will be huge. We will find new work directions, new ways to help each other, and new models of competition, but these may be very different from today’s work.
Initiative, willpower and determination can become extremely valuable. In a rapidly changing world, correctly judging direction and mastering coping methods will be of great value; resilience and adaptability will also be important skills worth cultivating. AGI will become a powerful lever to amplify human willpower as never before, giving individuals greater influence than ever before, rather than weakening it.
We expect the impact of AGI to be uneven.Although some industries may have changed little, scientific progress may be far faster than it is today. AGI’s influence in the field of scientific progress may surpass all other fields.
The prices of many goods will eventually fall sharply (currently, the costs of intelligence and energy constrain development in many areas), while the prices of luxury goods and some inherently scarce resources, such as land, may rise more significantly.
From a technical perspective, our path forward seems relatively clear. However, public policy and collective awareness of how to integrate AGI into society are crucial. One of the reasons why we release products early and frequently is precisely to allow time for co-evolution of society and technology.
Many of us believe that it is necessary to give people more control over technology than ever before, including a greater degree ofopen source, and recognize the trade-off between security and personal empowerment.
While we never want to be reckless and may need to make some important and potentially unpopular decisions and restrictions on AGI security, overall, as we get closer to achieving AGI, we believe that personal empowerment is crucial.
Ensuring that AGI’s well-being reaches the public is crucial. Historical experience of technological progress shows that most indicators we care about (such as health levels, economic prosperity, etc.) will improve overall and in the long run, but the improvement of social equity does not seem to be the inevitable result of technological progress. To achieve this goal, new ideas may be needed.
Of particular concern is that the balance of power between capital and labor can easily be disrupted, which may require early intervention. We are open to ideas that sound a little forward, such as providing everyone with a certain “computing budget” so that people around the world can make full use of artificial intelligence. Of course, we have also seen that simply continuing to reduce smart costs can achieve the expected results to a large extent.
Anyone in 2035 should be able to mobilize the intellectual resources of all mankind in 2025。Everyone should have the right to access unlimited “genius” power and be able to use it as he pleases. Today, there are a large number of talented groups in the world who lack adequate development resources. If we can change this situation, the creativity unleashed will bring great benefits to all of us.
