ChatGPT entrepreneurship, Wang Xiaochuan, Wang Huiwen to learn from Hoffman

Standing at the crossroads of science and technology, ChatGPT pearl jade in front, how to open up a way of entrepreneurship beyond its brilliance? The AGI entrepreneurship of the old gun Reed Hoffman may be more relevant for Chinese entrepreneurs.

Written by|Blue Hole Business Zhao Weiwei

A middle-aged Internet veteran, as the founder of the company, has a proud record in the Internet wave, and after experiencing ChatGPT, he was excited, and he jumped into the battle to start an AI big model entrepreneurship, full of optimism and positive attitude towards AI, aiming to become a trillion-dollar company in the future.

Is the protagonist who got the script of this story familiar?

In China, it may be Wang Xiaochuan or Wang Huiwen who is light-years away, or Kai-Fu Lee of Project AI 2.0 or Yan Junjie of MiniMax, in short, in climbing the golden mountain of Artificial General Intelligence (AGI), the brothers work hard and wait for the peak to meet.

ChatGPT has become a staple for Chinese entrepreneurs to catch up, but in the United States, there is a hidden entrepreneurial story of ChatGPT and Sam Altman, which demonstrates another possibility of middle-aged Internet veterans in the AGI wave of entrepreneurship: Reid Hoffman.

As one of the founders, Reed Hoffman sold PayPal and Linkedin to eBay and Microsoft, and then flexibly roamed the American Internet world as an investor.

Along with Musk, Hoffman was one of the early investors in OpenAI, ChatGPT's parent company, and after experiencing ChatGPT, he excitedly advocated for the change brought about by AGI and co-published a book with ChatGPT, Impromptu: Amplifying Our Humanity Through AI.

As a board member of OpenAI, he brokered Microsoft's investment in OpenAI, but this year he stepped down from OpenAI's board to start his own venture — raising at least $225 million to start a startup Inflection, and just released the Pi chatbot in May, which aims to build personal AI for everyone to solve real-world problems.

Standing at the crossroads of science and technology, ChatGPT pearl jade in front, how to open up a way of entrepreneurship outside the light of ChatGPT and big factories? People should be high-profile, things should be low-key, and the old cannon Reed Hoffman's AGI entrepreneurship may be more relevant for Chinese entrepreneurs.

Before you get off the field, actively beat the side drum

"How many Californians do you need to change a light bulb?"

"Five, one person twists the light bulb, four people provide experience." It's an old-fashioned English joke.

In July 2022, Reed Hoffman experienced GPT-4 as a board member of Open AI, and he took the question in this joke to GPT-4, and GPT-4 gave a serious version and a humorous version of the answer.

The serious version is that changing the bulb requires an inspector to follow the correct procedures and use the appropriate tools and equipment, and if the bulb is in a hazardous area, an extra person to help and supervise.

And the joke version is that it takes four inspectors to change a light bulb, one person to escalate a ladder, one person to unscrew the old bulb, one to screw the new one, and another person to write a ticket because the wrong wattage was used.

Reed Hoffman asked GPT-4 to give a different answer to the bulb replacement problem, in the style of Wittgenstein and American comedian Jerry Seinfeld.

Hoffman was thrilled by GPT-4's smooth and complete answer, his first intimate encounter with ChatGPT, and since then, he has been on the road to advocating the change and impact of ChatGPT.

The most visible example of beating the public was in March, when Hoffman published a book with GPT-4, Impromptu: Amplifying Our Humanity Through AI. He likens the book to a travelogue that takes his interaction with GPT-4 and explores the use of general artificial intelligence.

The central point in the book is that the AGI represented by ChatGPT is an amplifier of human capabilities.

Hoffman showed a positive and optimistic side, he believes that just as clothes, cars, glasses, smartphones and other products constitute part of human life, human beings have become what they are today, a large part of which is composed of technology, so general artificial intelligence will also become part of these technical components, and the future human beings may be as inseparable from smartphones as they are now, inseparable from general artificial intelligence.

In fact, much earlier than actively beating the drum for ChatGPT, Hoffman was involved in the creation of OpenAI as an investor.

"I jumped at the chance when I had the opportunity to be one of the original funders of OpenAI in 2015. The vision of artificial intelligence it plans to pursue feels like a natural extension of the goal that inspired me to co-found LinkedIn in 2002." Hoffman writes in the book.

In 2015, at an Italian restaurant in California, Hoffman met with Musk and Altman to discuss founding OpenAI. Hoffman's venture capital firm, Greylock Partners, has since invested in at least 37 AI companies, his preferred field, so it's only natural that he is willing to bet on OpenAI to ensure that the most powerful AI can "benefit all mankind."

Hoffman's finding Microsoft, the gold owner for Open AI, is Hoffman's biggest achievement for ChatGPT.

Having experience with Linkedin's acquisition by Microsoft, Hoffman encouraged Ultraman to meet with Microsoft to find corporate partners. At the time, Ultraman was worried that Microsoft's responsibility for shareholders might affect Open AI's mission and profit requirements, but Hoffman, as a board member of Open AI and Microsoft, gave Ultraman some advice, he modeled Microsoft and inspired Ultraman to think about what the two sides care about and are good at, and what Microsoft is not good at.

Finally, in 2019, OpenAI and Microsoft reached a $1 billion investment agreement, and Microsoft gave money and infrastructure technology. Although Hoffman did not participate in the negotiations to avoid a conflict of interest, the story since then is more widely known, and with Microsoft's support, Open AI has developed into an industry leader.

"The ratio of big companies and startups winning is five to five"

In March, just as ChatGPT was on fire around the world, Hoffman avoided a potential conflict of interest, resigned from OpenAI's nonprofit board, and went down to do AGI himself.

The general artificial intelligence chat product Pi, launched in May, is Hoffman's entrepreneurial work, which comes from the startup Inflection AI, co-founders and former Google DeepMind leaders Mustafa Suleiman and Karen Simonian, and raised at least $225 million in the first financing.

Inflection AI's official website has begun offering access to the Pi through natural language, and it describes itself as "an AI studio dedicated to creating personal AI for everyone," notably emphasizing that "safety is at the heart of our mission and culture."

How to make people remember in the big model entrepreneurship wave? The first is to find a celebrity endorsement.

Hoffman, who is well-connected, is clearly well versed in this, such as Bill Gates, who is bullish on AGI, named Hoffman's company, saying at an event in San Francisco, "I'm impressed with several startups, including Inflection." Bill Gates also believes that AGI will subvert the current Internet, and the proportion of winners in this artificial intelligence technology war, large companies and startups, is five or five.

And Hoffman's entry itself demonstrates the possibility: the opportunity for startups is no smaller than for big companies. Looking back years later at investing in generative AI companies like Open AI, Hoffman knew at the time: There was gold on these mountains.

The second is to find a way to differentiate from ChatGPT, IQ can't fight, then fight EQ.

The memo on Pi's official website mentions that the current Pi is only the first version, there is still a long way to go, it is a compassionate artificial intelligence that will help us dream, create, learn, solve, plan and complete things, be it a coach, confidant, creative partner or soundboard.

In the podcast, Hoffman also mentioned that general artificial intelligence will not replace human communication, and its key ability is to help people interact, such as "can help me solve your friend to make you angry" is what Pi is doing, "A lot of artificial intelligence is trained by highly intelligent people, but what about the high emotional intelligence you want?" That's part of what Inflection and Pi are doing."

Essentially, Inflection is still in its infancy, and its core is still figuring out how to help humans improve their own capabilities through AGI and become a basic tool for doing everything.

That doesn't prevent it from finding a way to differentiate itself from ChatGPT, and one of the co-founders, Suleiman, has said that the vision of founding Inflection is to make a chatbot that can deal with real-life problems.

Even though Pi is nowhere near as popular as ChatGPT in the U.S. market, Hoffman and the Inflection behind it embarked on an exploratory attempt to be essentially a public welfare enterprise (PBC) that requires companies to operate like traditional companies, but with higher standards of purpose, accountability and transparency.

Hoffman argues that this entrepreneurial nature, while not solving all the limitations of traditional companies, is a small step in the right direction.

Expanding model products in the form of a public enterprise, Hoffman's goal is to operate our AI studio in a way that promotes specific public welfare purposes, "developing products and technologies that utilize artificial intelligence to enhance human well-being and productivity, while respecting individual freedoms, working in the public interest, and ensuring that our products reach present and future generations widely."

That's why Pi first emphasizes that safety is at the core of our mission and culture, and the company culture and values behind it are: a culture that always seeks feedback and critical reflection, focusing on continuous adaptation and steady improvement.

On the road to large-model entrepreneurship, avoiding the overturning accident of Google Bard, preventing unknown risks, and lowering your posture is a very important survival rule, which is also an important lesson for Chinese AGI startups.

High-profile personality and low profile

The chubby Hoffman is an AGI techno-optimist, and with his experience in the Internet industry, he predicted that general artificial intelligence will revolutionize the industry within two years, and that trillion-dollar companies will be born in the field of big language models in the future.

Hoffman has led three podcasts, and recently been frequently interviewed, expressing three thinking perspectives on the future of general artificial intelligence in the public opinion field:

One is the industry assistant, whether in the fields of law or medicine, every worker will have a personal artificial intelligence assistant in the future, which is equivalent to graphic designers having Photoshop;

The second is the research assistant, the answer generated by the immediacy of general artificial intelligence, even if it occasionally makes mistakes, but immediacy is very important;

Third, changes in products and services will bring significant improvements in efficiency in the B-end market, such as customer service making it easier to connect and help customers.

He also believes that the risks of GMI are exaggerated at the moment, and he does not think that GMI will cause potential problems such as unemployment and economic chaos, because the obvious solution is to provide more technology and larger models, and "the solution exists in the future, not in the past."

There is no shortage of war of words in China's AI big model market, and the United States is no exception. Even Elon Musk, a former partner who co-founded OpenAI, Hoffman is not hesitant to criticize.

Musk called for a global moratorium on artificial intelligence more advanced than GPT-4 in March this year for six months, signed by thousands of experts and scholars. Hoffman believes these public calls are well-intentioned because it's important to prevent the negative effects of technology, but Musk's starting point may be self-interested.

"Everyone else slows down, so I can speed up," Hoffman speculated about Musk's motives. In Hoffman's view, Musk always has a "I have to build it myself" idea, "you can see what he has done with SpaceX and Tesla, there is a "only I do it will be great" method."

Criticism to criticism, interests to interests, in the face of reality, the safest strategy is to adapt to change, and the high-profile Hoffman may also have to bow his head.

Just on May 30, a signed statement that shocked the industry appeared. More than 300 industry experts and scholars, including Sam Altman, CEO of Open AI, jointly issued an "AI Risk Statement", which was short and only one sentence: "Mitigating the risk of AI extinction should be a global priority along with other society-scale risks such as epidemics and nuclear war."

The statement does not set specific goals, but rather hopes to start discussions to prepare for the risks that GMI technologies may pose, which also means that a global risk consensus on GMI technologies is already coalescing.

In the statement, Chinese experts and researchers are mainly concentrated in academia, including Zhang Yaqin, dean of the Intelligent Industry Research Institute of Tsinghua University, Zhan Xianyuan, assistant researcher of the Tsinghua Intelligent Industry Research Institute (AIR), and Zeng Yi, a professor at the Institute of Automation of the Chinese Academy of Sciences.

Reed Hoffman also does not appear in this list, but Mustafa Suleiman, also co-founder of Inflection AI, is one of the signatories, which in a way represents the company's recognition of the statement.

Hoffman and Inflection are one of the representatives of the current observation of the field of general artificial intelligence and the wave of large-model entrepreneurship, Pi does not support Chinese for the time being, to catch up with ChatGPT, obviously there is still a long way to go, and Hoffman's high profile, and Inflection's deliberately lowered posture, just shows that benefits and risks are the biggest variables to climb the golden mountain of ChatGPT.