Author: Techub News Compilation

Introduction

Recently, OpenAI co-founder and CEO Sam Altman attended the “Express Adda” dialogue event hosted by The Indian Express in India. In an intensive Q&A session lasting nearly an hour, Altman candidly and deeply shared his insights on the technological leaps in artificial intelligence, its global impact, geopolitical issues, and OpenAI's strategic thinking. This dialogue comes at a time when AI technology is rapidly penetrating various global industries, and geopolitical competition is intensifying. As a leader in the field, Altman's insights are of significant reference value for understanding the future direction of AI.

Summary

• The capabilities of AI have achieved an "astonishing" leap in the past year: evolving from solving high school math problems to making breakthroughs in unknown mathematical research questions, completely transforming programming work.
• Computing power is the key infrastructure for the future: the world needs to build computing power on an unprecedented scale, which will be the most expensive and complex collective infrastructure project in human history.
• Technology must be democratized: Altman strongly advocates for widely handing AI technology to people to avoid excessive concentration of power, arguing that this is central to addressing geopolitical competition and ensuring a positive future.
• AI will reshape employment rather than eliminate jobs: history shows that technological revolutions create new positions; the key will be rapid adaptation and the cultivation of core skills such as using AI tools, adaptability, and collaboration.
• AGI is "quite close," and superintelligence (ASI) is not far off: based on the accelerated capabilities of current models, Altman has revised his previous expectations about the speed of technological development, believing that the "takeoff" will come sooner.

The "Astonishing" Leap of AI Capabilities and Changes in the Indian Ecosystem

Sam Altman opened the dialogue by discussing the exponential progress of AI capabilities over the past year. He recalled that just a year ago in India, the fact that AI could solve high school math problems was still astonishing. Now, AI not only performs reasonably well in the highest-level math competitions but can also solve entirely new research problems that even top mathematicians have not cracked—OpenAI's latest model answered 7 out of 10 unknown math problems correctly. Similarly, in fields like physics, AI is also pushing the boundaries of human knowledge. Altman summarized, “AI has developed from performing tasks that we expected a very intelligent high school student could do to pushing the edges of human knowledge.”

Another disruptive change has occurred in the programming field. Altman pointed out that the nature of programming work has changed more dramatically in the past year than in any year since he became an adult. From code autocompletion tools to the current ability to generate complete applications with just an idea input (he specifically mentioned OpenAI's code generation tool Codex), the barriers and paradigms of programming have been completely transformed. He revealed that India is the fastest-growing market for Codex globally, which led to his observations about changes in the Indian AI ecosystem.

Altman felt that India has shifted from being an "AI consumer" a year ago to a vibrant "AI builder." During his visit to the Indian Institute of Technology Delhi, he was amazed by the energy of the new generation of startups and looks forward to India bringing upward surprises to the world in this field. Regarding earlier controversial remarks about India struggling to create frontier models with $10 million, he clarified that this referred to the high costs associated with building cutting-edge (frontier) models and was not a denial of India's exceptional capabilities in specific domain models. He stressed that with sufficient funding, Indian companies are fully capable of creating frontier models, and the smaller, more specialized models being developed in India are already “incredible.”

Computing Power Competition, Infrastructure, and the Role of Government

When asked whether the world has enough computing power to support India becoming an "AI-first" society, Altman straightforwardly replied, “Not yet.” However, he believes the world must urgently collaborate to address this issue. He proposed a thought experiment: how many GPUs would each person want working continuously for them? Answers range from at least one to thousands. If this number is multiplied by the global population, the scale of computation required is currently unattainable. Altman believes this indicates the ambition of the infrastructure we need—building sufficient computing capacity will be “the most expensive and complex infrastructure project ever undertaken collectively by humanity,” which concerns the well-being of all.

The good news is that AI and robots will help humans complete this construction. Regarding the currently debated concept of "space data centers," Altman believes it is unrealistic in the short term. Considering launch costs, energy acquisition, and the difficulty of maintaining hardware in space, orbital data centers will not scale in this decade.

Such massive infrastructure construction will inevitably tighten the relationship between organizations like OpenAI and governments. Altman acknowledged that given the deep impact of AI on society and the need to truly democratize this technology, government involvement is crucial, and tech companies must collaborate with governments. He observed that the tech industry initially held strong libertarian views, but as companies have grown and become more central to the economy and the way the world operates, this relationship has significantly changed. The AI era will make such collaboration more important than ever.

Regarding intergovernmental relations, Altman predicts that AI will become one of the most important political issues in the world, a higher-order factor in geopolitical tension and cooperation. However, he believes the landscape will not be fixed, and political alliances will evolve rapidly with the advancement of technology. When asked about China’s role in the AI race, he noted that China is clearly leading in certain areas (like manufacturing, physical robotics, and energy construction), but is not necessarily leading in others, and this distribution of advantages is normal. He emphasized that competition and cooperation coexist, but the important thing is that “democratic systems should lead AI development.”

Altman further explained that he does not specifically mean that no single country should lead, but rather that there should not be any single superintelligence in the world, regardless of whether it is held by individuals, companies, or countries. He believes that the world will be in its best state when power is widely distributed, there are diverse thoughts, and power is balanced enough to provide mutual checks.

Concentration of Power and Democratization: Key Choices in the AI Era

Altman believes that how AI will affect the power structure is one of the most important questions today. The world may swing to one extreme: AI may concentrate power to a great extent, with a single company or country accumulating tremendous global power and wealth through AI. Alternatively, it could swing to another extreme: everyone on Earth has a rule-free superintelligence, leading to chaos. He personally leans towards a more democratized version.

He believes that some regulation and guardrails are needed, but putting technology in the hands of the majority is possible, which will bring about the decentralization of economic power. The clearest example is in the past six months, especially the past three months, where the energy that one or two or three-person companies can wield was impossible a few years ago. In this new world composed of Codex and general knowledge working intelligence agents, we must quickly adapt to the fact that startup companies with only a few people can achieve great success, which he finds “fantastic.”

Regarding industry competition, Altman acknowledged that it is very intense, but he also pointed out that almost all teams with advanced models are jointly driving the development of the field, and are very committed to ensuring safety and alignment, and are willing to collaborate on this front.

Employment Impacts, Educational Challenges, and Future Skills

When addressing the potential employment impacts of AI on India's IT and services industry (which accounts for about 8% of GDP), Altman did not shy away. He believes this will have significant consequences; pretending otherwise is unhelpful. However, he expects, as has occurred with every advancement in the programming field, that people will work at a higher level of abstraction and be able to accomplish more tasks. The broad demand for programming still exists, and the volume of products and code output will increase dramatically, as will expectations. As long as countries and companies can quickly adapt to this new world, there will be plenty of new opportunities.

Altman describes himself as not a "job apocalypse theorist." Historically, every technological revolution has sparked panic over job losses, but each has also created new jobs in other respects. The promise of "permanent vacation" has never been fulfilled, and he believes it will not happen this time either. Human desires—to be useful to one another, to crave more and better things, to express creativity—seem to have no natural limits. He suggests that regardless of the future, universal skills will include: proficiency in using AI tools, resilience, adaptability, insight into the needs of others, and collaboration.

Regarding education, he acknowledged that some children misuse AI to cheat, which is bad, and it is essential to continue teaching children to learn, think, and create, while utilizing these tools. However, more children are demonstrating the ability to use new tools to complete amazing projects. He recalled how the education system adapted during the emergence of Google, believing that history will repeat itself: with increased potential and rising expectations, we will have to teach people to think harder and create more. He believes that children born today will be able to accomplish tasks that are currently unachievable by anyone.

Rapid Q&A: From Competitors to Personal Reflection

During the quick Q&A session of the dialogue, Altman shared a series of sharp perspectives:

• Appreciations for Google: First, its team invested with strong belief in the early days of modern AI; second, its relentless focus, execution, and model scaling capabilities after being clearly behind were impressive.
• Criticism and Defense of AI: In response to the “excess concentration of power,” he argued that unless we actively promote democratization; regarding “data centers consuming vast amounts of water,” he called current claims about how many gallons of water ChatGPT uses per query “completely untrue”; concerning “energy consumption,” he admitted the total is real and needs to quickly shift to nuclear, wind, and solar energy, but in terms of “per inference energy consumption,” AI might already be more efficient than humans.
• Company Positioning: OpenAI is a research-first company, as most of creating good products hinges on solid research.
• Personal Reflection: One of his biggest regrets is not holding any equity in the early days of OpenAI, which sparked unfounded conspiracy theories. One of his greatest disagreements with co-founder Greg Brockman is whether the company should focus or pursue multiple lines concurrently; he acknowledged that his past attempts to do too many things were wrong, learning from Greg the importance of extreme focus.
• AGI and ASI Timeline: Altman believes AGI (artificial general intelligence) is “quite close,” while superintelligence (ASI) “is not far off,” and has revised expectations indicating that the speed of technological “takeoff” will be faster than initially imagined.
• Vision for New Hardware: Collaboration with Jony Ive aims to design an entirely new family of products around AI, integrating technology into life, understanding context, maximizing assistance while minimizing disruption, with more information expected later this year.
• Unified Advice to Global Leaders: Whoever they are, they must democratize this technology and put it in the hands of people.

Conclusion: Creators, Professions, and Responsibility

During the audience Q&A session, Altman further elaborated on his views on specific topics. Regarding creators' concerns that humans will become passive, he noted observations showing that creators using AI tools can iterate “idea-feedback” cycles faster, resulting in better content without widespread automation.

On the most threatened professions, he used his own experience as an example, pointing out that traditional software engineers (who manually write C++ code) have “ended,” and many professions will disappear in the old sense, but people will adapt and find new things to do. As for the least vulnerable professions, he cited examples of art and nursing: AI-generated art costs zero, but art created by humans is increasing in value; when people are sick, it is still the caring nurses looking after them, not robots.

Finally, regarding the balance between responsible AI and the speed of technological advancement, Altman stated that this is a significant focus for the company. Their principle is “start conservatively, then expand access,” recognizing the tension between democratizing AI and ensuring sufficient safety, but this principle has worked well so far and will continue to be upheld.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。