Technology without industrialization is useless technology
In May 2020, OpenAI launched GPT-3, which had the ability to translate, answer questions, and fill in text, surpassing all previous large language models based on natural language development. However, it also had obvious flaws: it would reveal basic errors in answers and give sentences that could seriously offend people.
However, OpenAI changed its previous research-only attitude and provided open APIs for academic institutions, commercial companies, and individuals to apply. Some people developed tools for automatically designing web pages, some developed more efficient text search tools, and others developed tools for building virtual characters in stories.
Following OpenAI's usual approach, the next generation product after GPT-3 would be GPT-4, and GPT-3's mission would be completed after publishing papers. But OpenAI founder Altman keenly captured signals that the AIGC industry was about to explode in applications. After "aligning" GPT-3 internally, OpenAI released the transitional product GPT-3.5 - ChatGPT - to global users, occupying the top ecological niche of generative AI.
The industry has always been clear that GPT-3 was truly groundbreaking in technology - unprecedented parameter count, self-supervised learning - to the extent that the term "GPT-3 moment" was coined to describe this shock. But for OpenAI, the significance of ChatGPT is self-evident. Even such an imperfect product is enough to start an arms race in the AI era.
At least from the current operational path, Altman not only understands products but also understands history. OpenAI didn't wait for GPT to become flawless enough, but let the product take the first step - empowering Microsoft to launch Copilot, partnering with Apple to integrate with Siri, launching the AI search engine SearchGPT, allowing the public to use their products as much as possible.
From the perspective of contemporary technological history, the technological leaps that have been repeatedly reproduced in the West, Japan and Korea prove that effective modern scientific research cannot be separated from the drive of the mass commercial system. Without market feedback from billions of individual users and thousands of enterprise users, even the most advanced scientific and technological systems are difficult to sustain, and will face classic problems like the Soviet Union - technology capable of sending people into space but unable to produce copiers and Walkmans.
After World War II, the Soviet Union displayed enthusiasm for boldly developing black technology trees, with pioneering achievements such as radio frequency induction, tokamak fusion reactors, and ground effect vehicles, all of which were ahead of the average level of the era.
Soviet scientific research served more grand goals, and transforming results into products serving consumers was not a priority. At the same time, American companies developed automatic valve technology, first used in lawn mowers and motorcycles, while Japanese companies made rapid progress in microelectronic device technology, but first used it to make game consoles and handheld calculators.
Soviet researchers gradually discovered that their Western counterparts who developed "vulgar" commercial products often had a technological innovation speed three times their own, and technological progress serving only grand narratives became increasingly outdated. The time when local youth could listen to domestic Walkmans was 6 years later than their Japanese peers, and until 1991, copiers were still precious imported goods that had to be locked in safes in various units.
This grand history answered the strong correlation between science and commerce, technology and products. Even looking at individual destinies in the present, there would be the same answer.
If we look at today's tech giants purely from the perspective of technology supremacists, guess who is being described from the following:
"The grandson of a nuclear physicist, an undergraduate studying physics at Princeton, who couldn't outperform his Sri Lankan classmate in problem-solving, so he gave up on becoming a physicist. Found a job at a big financial firm but didn't do it well, started an online bookstore, and went bald in middle age."
"An undergraduate studying computer science at Harvard, who didn't study well, spent his days idling and causing trouble, made a small page for the whole school to rate the appearance of female classmates until he dropped out, then opened a website whose main business was selling ads."
"His family worked as waiters in restaurants to support his studies, but after completing a master's degree in electrical engineering at Stanford, he gave up research and design work at a big company at the age of 30 to start his own business making electronic game peripherals. Can only wear leather jackets at his age."
"A computer science graduate who could have pursued a career as a technical expert, but always wanted to start a business. After losing a bid, he decided to develop a chat software with his own money. At nearly 30 years old, he still had to pretend to be a female netizen to chat with users, and almost went bankrupt because the servers were too expensive."
These are the experiences of Bezos, Zuckerberg, Huang Renxun, and Ma Huateng. They chose careers to make it easier for people to buy books, have more friends, experience better graphics, and enjoy online chatting. But it's these "small things" that have completely restructured consumption patterns and interpersonal relationships since the Industrial Revolution. If they had all chosen the path of "scientists," humanity might have only gained a few more professors, and the current technological landscape would not be more outstanding.
It is because of Altman's choice of timing for ChatGPT's release, his judgment on AI industrialization, and his rebellion against the academic path that AI, an industry that was once ridiculed as having "as much intelligence as there is artificiality," has returned to the public eye.
Although the opposition surrounding Altman is growing louder, arguing that he hasn't made OpenAI further explore industrial value but instead indulges in socializing with political figures and celebrities and giving speeches sharing life experiences, we have to admit that it is this "businessman" skilled in public relations and communication, rather than Britain's Deepmind or others, who has made countless smart brains in the United States converge and once again stand at the forefront of the AI era.
Google is wading through the same muddy waters as the Soviet Union
Speaking of Deepmind, that first-generation AI hot chicken that shocked the world with Alphago, we must mention Google, which acquired Deepmind ten years ago, got an early start but arrived late, and played a royal flush hand poorly.
In terms of scientific research achievements, perhaps no laboratory in human history can compare with Bell Labs: the first fax machine, push-button telephone, cellular phone, communication satellite, high-speed wireless data system, solar cell, digital signal processor, optical fiber... As an independent research entity established by a company, Bell Labs at its peak balanced basic research, applied research, and product development, accelerating the practical application and popularization of many scientific technologies.
The legendary story of Bell Labs inspired later giants to emulate, such as Google's X Lab, but unfortunately, the latter only learned half of it.
The "X Lab" originated from the "Google X" project in 2010, where the two founders Larry Page and Sergey Brin hoped to have people within the company specifically explore technologies with a science fiction feel that would dramatically change the world one day. Their official website and recruitment pages repeatedly emphasize "changing the world tenfold, not improving the world by one-tenth."
These mottos and beliefs express: first, only do big projects and big technologies, not small money-making ventures; second, only do dramatic, transformative technologies, not ordinary technological improvements.
The ideal is plump, but reality is bony.
The result of practicing the above ideas is that many of the scientific and technological innovation projects incubated by "X Lab" were technologically ahead at the start, but were on par with or even inferior to their peers when implemented.
In 2010, Google launched its self-driving research and development project. After more than a decade, Waymo, incubated by "X Lab," has encountered setbacks in commercialization and has undergone multiple layoffs, while even traditional car manufacturers are competing in the self-driving taxi market.
In 2011, "X Lab" launched the Project Glass project and introduced Google Glass in 2013, but it was quickly abandoned by the market due to privacy issues and limited functionality. In 2022, Google announced the discontinuation of Google Glass sales.