The Limits of Generative AI Have Emerged: Exploring the Future and Strategic Development of AI Investment
-
Gates recently stated that many within OpenAI, including Altman, believe GPT-5 will significantly outperform GPT-4. However, he argues there are many reasons to believe that generative AI has currently reached its limits.
So what will the next breakthrough be? Gates believes it will be explainable AI, but he estimates it won't be achieved until the next decade (2030-2039).
Gates also thinks Nvidia doesn't hold an absolute advantage in AI chips, as Microsoft, Google, Amazon, and even OpenAI itself are developing or considering developing their own AI chips.
Data shows that compared to the arms race in large models, Bill Gates is more optimistic about AI Agents. He once said that in the near future, any internet user could own an AI Agent.
Gates also admits that while he holds reservations about the future development of generative AI, he acknowledges that his assessment might be incorrect.
Prospects and Development Strategies for AI Investment
ChatGPT's popularity remains high. On March 21, Alibaba's technical vice president Jia Yangqing announced on social media his resignation, sparking rumors that his startup would focus on AI applications, including large models. On March 19, Innovation Works' chairman and CEO Kai-Fu Lee announced plans to "not just create a Chinese version of ChatGPT." Meanwhile, Baidu's Wenxin Yiyan, released on March 16, has become a hot topic of discussion.
Counting from February 4, when ChatGPT first trended on Weibo, the "fire" of ChatGPT has been burning in China for nearly two months. From tech companies with relevant technical reserves announcing their entry into the field to industry leaders like Meituan's "Two Kings," former Sogou CEO Wang Xiaochuan, and former JD Group senior vice president Zhou Bowen issuing "heroic invitations," the ChatGPT market in China has been constantly evolving.
It seems to have become a trend for tech "big shots" in China to continuously bet on ChatGPT. On the morning of March 21, "framework guru" Jia Yangqing posted on social media, announcing his resignation from positions including Alibaba's technical vice president. Rumors suggest that Jia Yangqing's entrepreneurial direction is focused on AI architecture, particularly AI applications including large models, and that his new company has already secured initial funding intentions.
Coincidentally, on March 19, Kai-Fu Lee, Chairman and CEO of Sinovation Ventures, officially announced on his social media that he is personally organizing Project AI 2.0. Lee stated that Project AI 2.0 is not just about creating a Chinese version of ChatGPT. "I believe AI 2.0 is not merely a high-powered chat tool or an AIGC generator for text and images. Co-pilot and the applications we see today are just the beginning of AI 2.0's capabilities."
It's worth noting that on March 15, Zhou Hongyi, founder of 360 Group, revealed the company's AI development strategy during a live broadcast. "360 will 'fly with two wings.' On one hand, we will continue to fully develop our own generative large language model technology, building our own 'engine.' On the other hand, we will focus on creating products for specific scenarios and launch related services as soon as possible," Zhou said.
However, when it comes to the most significant ChatGPT-related event in China recently, it has to be Baidu's release of Ernie Bot, the Chinese version of ChatGPT. Following its launch, many companies quickly announced partnerships with Ernie Bot, including giants in their respective fields. For example, on March 17, ZTE announced that its servers would support Ernie Bot, providing strong computing power for AI product applications.
Google's AI Bard Officially Launches Public Testing
Faced with the relentless advances of OpenAI and Microsoft, Google has finally decided not to sit idly by.
Today, Google announced the official public testing of Bard, its ChatGPT competitor, initially rolling out to users in the U.S. and the U.K. Currently, users in these two countries can join the waiting list at bard.google.com. As testing progresses, Bard will gradually become available in other regions.
However, compared to the confident stance of OpenAI and Microsoft, Google, which previously stumbled with Bard, is clearly more cautious. Unlike New Bing's large-scale open strategy, Bard's testing slots will be released gradually, and the initial version will only respond to text.
This time, Google also showcased more details about Bard's interface and features. Overall, Google's Bard is quite similar to New Bing in terms of usage patterns and functionality. Bard also provides users with a standalone chat interface where they can ask questions and search, such as asking Bard to help draft a plan, list a recipe, or search for obscure facts.
In terms of positioning, Google emphasizes that Bard is not a replacement for its search engine but rather a "complement to search." Bard describes itself as "your creative and helpful collaborator," while New Bing refers to its AI assistant as a "copilot."
After asking a question on Bard, you'll find four buttons at the bottom of the answer: like, dislike, regenerate, and Google it. This means you can seamlessly connect to Google's main search interface within the chat, with both complementing each other.
NVIDIA Launches AI Supercomputing Cloud Rental Service
According to reports, NVIDIA CEO Jensen Huang announced a plan on Tuesday to open access to supercomputers used for developing AI technologies like ChatGPT to commercial clients. Nearly any business can now use these powerful but expensive devices through cloud rental services.
NVIDIA's rental prices are not cheap, with a configuration of 8 flagship A100 or H100 chips costing $37,000 per month. However, making these supercomputers available to a broader range of commercial clients will accelerate AI development. The recent surge in AI technology has already driven NVIDIA's stock price up by 77% this year, giving it a market capitalization roughly five times that of its long-time competitor Intel.
NVIDIA currently dominates the AI chip industry and has helped partners like Microsoft build massive systems for OpenAI, the developer of ChatGPT, enabling capabilities such as answering questions in human language or generating images from prompts.
At NVIDIA's annual developer conference, Huang stated that the company is collaborating with partners like Oracle to provide cloud access to its DGX supercomputers. These supercomputers integrate up to 32,000 NVIDIA chips. With this service, anyone with a web browser can log in to the supercomputer.
"The iPhone moment of AI has begun," Huang said during his online keynote. He also mentioned that NVIDIA is working with Microsoft and Alphabet to offer its supercomputers as cloud services for developing new AI products.