Is AI a "Future Miracle" or a "Century Bubble"?

Recently, this discussion has been incessant, especially after SoftBank sold off its shares in Nvidia yesterday, making it a hot topic of conversation.

I recently came across a rare roundtable discussion featuring #AI leaders: Nvidia CEO Jensen Huang, former Meta Chief AI Scientist and Turing Award winner Yann LeCun, AI pioneer Fei-Fei Li, AI father Geoffrey Hinton, Turing Award winner Yoshua Bengio, and Nvidia Chief Scientist Bill Dally. These six world-class #AI experts engaged in dialogue for the first time, and the intensity was comparable to a celestial battle.

When the host asked, "Is there an AI bubble?", this group of the most knowledgeable people in #AI surprisingly gave completely different answers.

The first camp, represented by Jensen Huang, is the non-bubble faction. Huang provided an example from the internet bubble era, where most of the laid fiber optics were "dark fibers" that were not lit up, indicating that the industry's supply far exceeded actual demand. However, now, almost every available GPU is running at full capacity, which shows that the demand is real. He also believes that #AI is an emerging industry that requires factories to produce intelligence, and we are just at the beginning of this intelligent construction.

Nvidia Chief Scientist Bill Dally agrees with this viewpoint, stating that we have only just begun to scratch the surface of applications, estimating that we have only reached about 1% of the ultimate demand, which is still far from satisfying the market's needs.

In contrast to Huang's optimistic view of #AI, former Meta Chief AI Scientist Yann LeCun provided a nuanced answer, stating that there is a bubble in the short term, but not in the long term.

Why does he say there is no bubble in the long term? He believes that based on the current large language model technology, there are still many applications waiting to be developed, and these applications are sufficient to justify all current investments in software and infrastructure. From this perspective, AI investment is not a bubble.

As for why there is a bubble in the short term, he personally does not believe that, under the current large language model paradigm, AI can evolve to reach human intelligence levels. Otherwise, why have we not yet created a robot as smart as a cat? The progress of AI cannot rely solely on infrastructure investment and data accumulation; it requires some fundamental breakthroughs.

Fei-Fei Li also offered her unique perspective, stating that #AI as a field is still very young. Physics has developed for over 400 years, while AI is not yet 70 years old, and there are many new areas waiting to be explored. However, from a market perspective, there will always be its own dynamics and adjustments, which subtly suggests that AI will experience short-term fluctuations.

Similarly, Yoshua Bengio believes that past technological validations have been very successful, and he is optimistic about the long-term development of AI. However, unmet short-term expectations can bring financial risks, so caution is still needed in the short term.

Jensen Huang focuses on building infrastructure, while Yann LeCun calls for finding new paths. Whose viewpoint do you think is closer to the future of #AI? Feel free to discuss in the comments. 🧐

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。