Author: Sam Altman
Compiled by: TechFlow
It has been just over a month since ChatGPT's second birthday, and we are now entering the era of models capable of complex reasoning. The new year always prompts reflection, and I would like to take this opportunity to share some personal insights from this journey, as well as some of the gains along the way.
As we get closer to AGI, this is an important moment to reflect on the progress our company has made. There are still many unknowns to explore, and many questions yet to be solved, as everything is still in the early stages. But compared to when we first started, we have learned a lot.
We founded OpenAI about nine years ago because we believed that AGI was possible and had the potential to be the most impactful technology in human history. We wanted to find a way to build AGI and ensure it could benefit humanity widely; at the same time, we also yearned to write our own page in history. Our goals were immense, and we were also convinced that this work could give back to society in an equally grand way.
At that time, almost no one cared about AGI. If anyone was interested, most thought we were unlikely to succeed.
In 2022, OpenAI was still a low-profile research lab, developing a project temporarily called "Chat With GPT-3.5." (We were indeed far better at research than naming.) We noticed that developers really enjoyed the experience of conversing with the model through the API's playground feature. So we thought that building a demo around this conversational experience would not only showcase the future's possibilities but also help us improve the model's performance and safety.
Ultimately, we decided to call it ChatGPT and launched it on November 30, 2022.
We had always vaguely known that the AI revolution would come one day, but we didn't know what that moment would look like. To our surprise, this release became that turning point.
The launch of ChatGPT sparked an unprecedented wave of growth - in our company, across the industry, and globally. We finally saw the immense potential of AI and could foresee more breakthroughs on the horizon.
This path has not been smooth, and making the right choices has never been easy.
Over the past two years, we have built an entire company from scratch around this new technology. Facing such challenges, the only way to learn is through hands-on experience. When the technology itself is entirely new, no one can tell you exactly what to do.
Establishing a company at such a rapid pace, with almost no existing experience to draw from, was inevitably chaotic. We often took two steps forward and one step back (sometimes even one step forward and two steps back). While mistakes were constantly being corrected, there was almost no manual or clear direction when venturing into uncharted territory. Rapidly advancing into the unknown is an exhilarating experience, but it is also a huge source of pressure for all involved. Conflicts and misunderstandings were inevitable.
These years have been the most rewarding, interesting, and fulfilling of my life, but also the most exhausting and stressful. Especially the past two years, which could be considered the most difficult period, but my deepest feeling is one of gratitude. I know that one day, when I'm retired and sitting on a farm watching plants grow, feeling a bit bored, I'll look back on the opportunity to realize a childhood dream, and how fortunate that was. I try to remind myself of this every Friday, even when by 1 PM there are already seven things that have gone wrong.
About a year ago, on one such Friday, the thing that caught me off guard was being fired in a video call, with no warning whatsoever. Immediately after we hung up, the board published a blog post announcing the news. I was in a hotel room in Las Vegas at the time, and the feeling was almost indescribable, like a dream suddenly turning into a nightmare.
Being publicly fired without any warning led to hours of chaos and days of uncertainty. The strangest part was the "information fog." None of us could get clear answers about what had happened or the reasons behind it.
In my view, this entire event was a governance failure, even though everyone involved, including myself, acted in good faith. Looking back, I certainly wish I had handled things differently. I like to believe that I am more mature and thoughtful today than I was a year ago.
This incident made me realize that a board needs diverse perspectives and deep experience to navigate complex challenges. Good governance requires a foundation of deep trust and credibility. I am grateful that many have worked together to build a stronger governance system for OpenAI, allowing us to continue pursuing our mission of AGI for the benefit of all humanity.
The biggest realization I've had this past year is that I have so much to be thankful for, and so many people to thank: the colleagues at OpenAI who chose to dedicate their time and energy to realizing this dream; the friends who supported us in times of crisis; the customers and partners who trusted us, worked with us, and depended on our technology to succeed; and the people in my life who expressed care and concern.
After that, we re-engaged with our work in a more united and positive way. I am immensely proud of the focus we have shown since then. We have completed some of the most outstanding research in OpenAI's history. Our weekly active users have grown from around 100 million to over 300 million. Most importantly, we have continued to introduce technologies that people genuinely love and that solve real-world problems.
Nine years ago, we had no idea what we would ultimately become. Even today, we only have a vague sense of the future. AI development is full of ups and downs, and we expect there will be more challenges and opportunities ahead.
Some of these changes are exciting, while others are full of challenges. The steady stream of research breakthroughs is exhilarating, and many former skeptics have become steadfast supporters. Of course, we have also witnessed some colleagues choose to leave and become our competitors. With the team's constant growth, turnover is inevitable, and OpenAI's expansion has been particularly rapid. Startups typically experience personnel changes during each scaling phase, but at OpenAI, this change happens almost on a monthly basis. The changes of the past two years are equivalent to a decade's worth of development for a typical company. As a leading industry player growing at such a fast pace, it's inevitable that team members' goals and priorities will diverge. Additionally, as the industry leader, we have become the target of various criticisms and attacks, especially from competitive pressures.
Our vision has never changed, but our strategies have continuously adjusted and evolved over time. For example, in the beginning, we never imagined we would need to build a product company; we thought we could just focus on doing outstanding research. Similarly, we did not foresee the need for such massive funding to support our work. Now, we need to construct many things that were unimaginable just a few years ago. And in the future, there will undoubtedly be more unforeseen challenges awaiting us.
We are proud of the achievements we have made in research and technology application, and we are committed to continuing our deep exploration of safety and alignment. We have always believed that the best way to make AI systems safe is to introduce them to the world gradually and rhythmically, allowing society time to adapt and co-evolve with the technology. Learning from real-world applications and continuously improving the safety of the technology. We know that being a global leader in safety and alignment research is crucial, and this research needs to be guided by real-world feedback.
Today, we are confident that we have mastered the methods to build AGI according to the traditional definition. We believe that by 2025, we may see the first batch of AI agents "join the labor market" and significantly boost the productivity of enterprises. This will be a historic moment. We still firmly believe that the best way to achieve widespread and far-reaching impact is to gradually put excellent tools in the hands of people.
We have already begun to turn our sights to even more ambitious goals, which is the pursuit of true superintelligence. Although we are full of love for our existing products, our mission is to strive for that wonderful future. With superintelligence, we will be able to achieve all possible goals. Superintelligent tools will greatly accelerate scientific discoveries and technological innovations, helping us solve complex problems that cannot be completed by human capabilities alone, and significantly enhance the prosperity and affluence of society.
Now, all this may sound like science fiction, and even discussing it may seem a bit crazy. But that's okay - we've faced similar situations before, and we're excited to take on this challenge again. We are very confident that in the coming years, everyone will see the prospects we see, and recognize how important it is to maintain a high degree of caution while maximizing universal benefits and empowerment. Given the potential and possibilities of our work, OpenAI is destined to be more than an ordinary company.
Being able to participate in this endeavor makes me feel both fortunate and deeply honored.
(Thanks to the inspiration from Josh Tyrangiel. I really wish we could have more time to exchange.)
During those days, many people made tremendous efforts for OpenAI and myself, but two people stood out in particular.
Ron Conway and Brian Chesky's help went far beyond the call of duty, to the point that I don't even know how to express my gratitude. I had long heard of Ron's capabilities and indefatigable spirit, and in the past few years, I have also received countless help and advice from Brian.
But it is only in times of real crisis that one can truly see their extraordinary abilities. I can say without exaggeration that without their help, OpenAI might have collapsed. They worked around the clock for several days until the problem was completely resolved.
Despite their extremely intense workload, they remained calm, displayed clear strategic thinking, and provided extremely valuable advice. They prevented me from making several mistakes, while they themselves made no mistakes. They utilized their extensive network of contacts to meet all the needs and successfully handled many complex situations. I am sure they did many things that I was not aware of.
However, what touched me the most was their care, empathy, and support.
I thought I knew how to support a founder and a company, and to some extent, I did have some understanding. But I had never seen, or even heard of, a support like theirs. Now I finally understand why they have such a legendary reputation. They are truly exceptional and fully deserve such an evaluation. They possess extraordinary abilities to provide help in critical moments, overcoming all obstacles, and demonstrate unwavering commitment. Our tech industry has become better because of them.
Of course, there are others like them. This is a particularly special aspect of the tech industry, more important than people imagine, and it allows the entire industry to operate smoothly. I look forward to passing on this spirit.
On a more personal note, I am especially grateful to Ollie for his support over that weekend and always. He is unparalleled in every way, and no one can find a better partner than him.
