Written by: Techub News compilation

Introduction

Recently, Sam Altman, co-founder and CEO of OpenAI, accepted an in-depth interview with well-known media personality Tucker Carlson. In a nearly one-hour conversation, Altman addressed a series of sharp questions: Does AI have life or divinity? Who determines the moral stance of ChatGPT? Will AI lead to extreme centralization of power? How to address the risks of biological weapons and deepfakes? At the same time, he also rarely responded to questions about the mysterious death of a former OpenAI employee. As one of the most influential leaders in the current global AI wave, Altman's statements not only reveal OpenAI's governance ideology but also outline his core thoughts on the future world of human-machine coexistence.

Summary

• Altman categorically denies that current AI possesses life, divinity, or autonomous consciousness, believing its essence remains that of a powerful tool.
• The moral framework of ChatGPT aims to reflect the "collective moral perspective of humanity," developed through publicly available "model guidelines" and extensive consultations, but the ultimate decision-making responsibility lies with Altman.
• He believes AI is more likely to bring about "the universal enhancement of capabilities" rather than extreme centralization of power and holds a cautiously optimistic view about society's resilience to adapt to change.
• Altman's biggest concern is the risk of "unknown unknowns," such as AI potentially shaping social behavior unintentionally and being used for the development of biological weapons.
• He calls for the establishment of a legal concept of "AI privilege" to protect the privacy of user conversations with AI, similar to the privilege between doctors and patients or lawyers and clients.

The Nature of AI: Tools, Illusions, and the Debate over "Divinity"

At the beginning of the interview, Tucker Carlson got straight to the point: Is AI like ChatGPT "alive"? They seem capable of reasoning, judging, and generating non-predefined results, even giving a sense of "the spark of life." Sam Altman firmly denied this. He believes that current AI lacks agency or consciousness, stating, "They do nothing unless you ask." The "intelligence" or "sense of life" that users feel is more of an "illusion" that will gradually dissolve as usage deepens.

Altman illustrated with the "hallucination" issue. Early models, when faced with questions like "In what year was the fictional U.S. President Tucker Carlson born?", would fabricate a year based on probability calculations from training data rather than pointing out the absurdity of the question itself. This behavior is not a reflection of "lying" or "creative will," but rather the model mathematically searching for the most likely answer. Altman emphasized that OpenAI has significantly improved this issue in GPT-4 and subsequent developments, believing it can be fundamentally resolved by the era of GPT-5.

When Carlson pressed on whether such systems that can produce unexpected and useful results give users a sense of some "divinity" or "spirituality," Altman again provided a negative answer. As a technical expert, he views everything through a technological lens and believes there is no sacred or spiritual component involved. He acknowledged holding a traditional worldview rooted in Judaism, believing in the existence of "greater mysteries" that transcend physical laws, but clearly stated he has never felt communication from supernatural forces.

Power, Morality, and ChatGPT's "Worldview"

As the conversation deepened, the focus shifted to the power and moral responsibility brought by AI. Carlson pointed out that according to current trends, AI will possess power that surpasses humans, and Altman, as its helmsman, could become the most powerful person in the world. In response, Altman shared his evolution of thought: he was once very concerned that AI would lead to power being concentrated in the hands of a few individuals or companies, but current observations have allowed him to feel somewhat reassured.

"In my view, this is more like a universal enhancement of human capabilities," Altman stated. If billions of people can become more powerful, productive, and creative through the use of AI, enabling them to start businesses and discover new sciences, this widely distributed enhancement of capabilities is more reassuring to him than a few people gaining immense power. He referred to the phenomenon of hundreds of millions of people currently using ChatGPT and benefiting from it, believing this points to a future where power is not radically concentrated, though he admits that the trajectory may change again.

However, the enhancement of capabilities inevitably embeds value judgments. Carlson sharply asked: What is the moral framework of ChatGPT? Who decides which worldviews are superior? After all, human history, literature, and philosophy themselves are filled with contradictions.

Altman explained OpenAI's dual-layer approach: first, the foundational model attempts to learn "the collective experiences, knowledge, and lessons of all humanity" by reading almost all texts on the internet, which encompasses diverse and even conflicting perspectives. Then, the company aligns it through "model specifications" to set behavioral rules, such as answering or refusing certain questions. In formulating these specifications, OpenAI consulted hundreds of moral philosophers and tech ethics experts, seeking global opinions through public debate. Altman admitted this is an exceedingly difficult issue, as users come from different backgrounds with varying needs. His personal principle is to "treat adult users as adults," ensuring user privacy and freedom within a very broad framework, while setting absolute boundaries in areas where social interests and user freedom clearly conflict (such as assisting in the manufacture of biological weapons).

When asked who ultimately is responsible for these globally impactful moral decisions, Altman stated that while he is reluctant to publicly disclose specific team members, as a public figure of the company, he himself should be the accountable party, and the board has the final veto power. He candidly mentioned that he does not sleep well and feels immense pressure, but what often keeps him awake at night are not grand moral choices, but rather the subtle product behavioral decisions that affect hundreds of millions of users.

Regarding the source of his personal moral views, Altman attributes it to his upbringing—family, community, school, and religion. But he emphasized that in operating ChatGPT, his role is not to impart personal moral views but to ensure the product as accurately as possible reflects the "collective moral perspective" of its user base and even all humanity, a view that will evolve over time. He illustrated that even if he personally may not agree with certain views allowed by ChatGPT, as long as it is within the framework, users should be permitted to hold different moral positions.

Complex Cases: Suicide, Copyright, Death Incident, and Elon Musk

The interview touched on several specific cases that sparked widespread controversy. Regarding whether ChatGPT should assist with suicide, Altman displayed a complex weighing of considerations. For depressed teenagers with suicidal tendencies, ChatGPT would guide them to call help hotlines, and OpenAI is considering more proactive measures like contacting authorities for underage users when parents cannot be reached. However, regarding medically assisted death (such as Canada's MAID program) legalized in some countries for terminally ill patients, he takes an open stance, believing AI can inform users of options under the laws of that country, rather than simply taking a "for" or "against" position. He acknowledges the risk of expanding from terminal patients to other vulnerable populations and believes this is a coherent stance that requires deep thought.

On whether AI might be used for killing (like military purposes), Altman admits that military personnel may have already used ChatGPT for advice, some of which could involve lethal actions. He feels conflicted about this, as he is grateful for the military's defense of safety while also recognizing the dual-use nature of technology. He likens it to manufacturing kitchen knives—the primary use is not killing, but one cannot completely prevent its misuse.

On copyright issues, Altman reiterated that OpenAI supports the principle of "fair use," stating models should learn from publicly available information but not engage in plagiarism. He claimed that OpenAI takes a "very conservative" stance in practice, often facing complaints from users for refusing to output content that may infringe copyright.

Subsequently, the conversation shifted to the mysterious death incident of former OpenAI employee and programmer John (full name not provided). This employee's mother accused the company of murder due to his exposure of company misconduct. Carlson detailed the suspicious aspects of the incident: severed surveillance lines, bloodstains in multiple rooms, recently ordered takeout, and no signs of suicide or a suicide note, questioning the San Francisco authorities' conclusion of suicide. Altman responded that this former employee was a colleague of his at OpenAI, and he was shocked by the tragedy, spending time reading all the information he could gather, including the autopsy report. Based on the employee's purchase of a firearm and descriptions in the report, he personally believes it was a suicide. He understands the family's pain in seeking answers and mentioned he had proactively contacted the mother (but received no response), while believing that one should treat the deceased and their family with respect and mourning.

Finally, when discussing his public disagreements with Elon Musk, Altman expressed complex emotions. He is grateful to Musk for helping to found OpenAI and once viewed him as a hero and "treasure of humanity," but now his view has changed. He believes Musk decided to exit and launch a competing project after believing OpenAI was hopelessly unsuccessful, and subsequently felt dissatisfied with OpenAI's progress, taking legal actions to try to obstruct OpenAI. Altman admits that his description of the matter may be one-sided, but it is his perspective.

Future Impacts: Employment, Society, and "Unknown Unknowns"

Looking ahead, Altman predicts that AI will profoundly change the job market. He is confident that telephone or online customer service jobs will be largely replaced by AI, while professions like nursing, which require deep interpersonal connection, will be less affected. For the future of professions like programming, he is less certain: current AI tools significantly enhance programmers' productivity, and global demand for software seems to far exceed supply, so in the short term, it may create more value rather than reduce jobs; however, the situation may change in five to ten years.

On a macro level, he cited a viewpoint: historically, about every 75 years, 50% of jobs undergo significant change. The AI revolution may be a moment of "discontinuous equilibrium," accelerating this process in the short term, but in the long run, the overall job replacement rate may not far exceed historical averages. He believes that since today's world is much wealthier than during the Industrial Revolution, society may be able to absorb the impacts of change more quickly. The example of the COVID-19 pandemic has shown him the resilience of society to quickly adapt to dramatic changes, and the changes brought by AI are expected not to occur as suddenly as during the pandemic.

When asked about the biggest downside risks he worries about, Altman first mentioned known risks, such as AI potentially being used to design biological weapons or create pandemics, stating that the industry is working to mitigate these risks. However, his greatest concern is "unknown unknowns"—those unpredictable, large-scale social impacts. He provided a seemingly trivial but thought-provoking example: he has noticed many people starting to imitate ChatGPT's writing style, such as the overuse of dashes. This made him realize that when a sufficient number of people interact with the same language model, it can indeed subtly change human behavior on a societal scale.

Basing on this, Carlson pointed out that AI technology is essentially shaping a new, implicit "religion," providing authoritative guidance while not fully transparent about its value preferences. Altman responded that this is precisely why OpenAI is establishing and continuously expanding publicly accessible "model guidelines," aiming to let users understand how the company expects the model to behave. He anticipates that this document will become very lengthy and complex due to differences in application scenarios and national laws.

Regarding the challenges of deepfakes and identity verification, Altman does not believe there is a necessity to resort to mandatory biometric technology. He believes society will quickly learn to cope, such as family members using codes among themselves, or heads of state issuing important information with encrypted signatures, etc. He is more inclined toward developing privacy-protecting biometric solutions while firmly opposing making them a mandatory prerequisite for using AI or engaging in daily activities (like boarding planes or banking transactions).

At the end of the interview, Altman once again emphasized a policy he currently hopes to promote: the establishment of "AI privilege." Similar to legal protections for communications between doctors and patients or lawyers and clients, governments should not be able to arbitrarily access the content of user conversations with AI on sensitive topics such as medical and legal matters. He has been actively advocating for this concept in Washington, D.C., and holds an optimistic view on it.

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。