Sam Altman's New Year Confession and Outlook: Putting Powerful Tools in People's Hands, First Wave of AI Agents Expected to Join Workforce This Year
TechFlow Selected TechFlow Selected
Sam Altman's New Year Confession and Outlook: Putting Powerful Tools in People's Hands, First Wave of AI Agents Expected to Join Workforce This Year
Putting excellent tools into people's hands step by step is the best way to achieve broad and profound impact.
Navigating Web3 tides with focused insightsAuthor: Sam Altman
Translation: TechFlow
Just over a month has passed since ChatGPT’s second birthday, and we have now entered an era of new models capable of complex reasoning. The new year always invites reflection, so I’d like to take this opportunity to share some personal thoughts about this journey and lessons learned along the way.
As we draw closer to AGI, it’s an important moment to reflect on how far our company has come. There is still much we don’t know and many problems left unsolved—everything is still just beginning. But compared to where we started, we’ve learned a great deal.
We founded OpenAI about nine years ago because we believed AGI was possible and could become the most impactful technology in human history. We wanted to figure out how to build it and ensure its benefits would be broadly shared across humanity—while also hoping to carve out our own place in history. Our ambitions were enormous, and we firmly believed this work could give back to society in equally monumental ways.
At the time, almost no one cared about AGI. And for those who did, most assumed we had no chance of success.
In 2022, OpenAI was still a quiet research lab working on a project temporarily called “Chat With GPT-3.5.” (We’ve always been better at research than naming things.) We noticed developers particularly enjoyed conversing with the model through the API playground. So we thought building a demo around that conversational experience could not only show people what might be possible but also help us improve the model’s performance and safety.
Eventually, we decided to name it ChatGPT and officially launched it on November 30, 2022.
We always had a vague sense that the AI revolution would arrive someday—but we didn’t know what that moment would look like. To our surprise, this release became that turning point.
The launch of ChatGPT triggered an unprecedented wave of growth—within our company, across the industry, and globally. We finally saw the vast potential of AI and could begin envisioning even greater breakthroughs ahead.
This path hasn’t been smooth, and making the right choices has never been easy.
Over the past two years, we’ve built an entire company from nearly nothing around this new technology. Faced with such a challenge, the only way to learn has been by doing. When the technology itself is entirely new, no one can tell you exactly what to do.
Building a company at such speed, with almost no existing playbook to follow, has inevitably been chaotic. We often took two steps forward and one step back (sometimes even one step forward and two steps back). While mistakes get corrected over time, there are hardly any manuals or clear directions when pioneering uncharted territory. Moving quickly into the unknown is exhilarating—but also intensely stressful for everyone involved. Conflict and misunderstandings were unavoidable.
These years have been the most fulfilling, interesting, and rewarding of my life—and also the most exhausting and stressful. The last two years, in particular, have likely been the hardest period I’ve ever experienced. Yet deep down, what I feel most is gratitude. I know that one day, when I’m retired on a farm watching plants grow and feeling a bit bored, I’ll look back on having had the chance to fulfill a childhood dream—and realize how incredibly lucky that was. I try to remind myself of this every Friday, even when seven things have already gone wrong by 1 PM.
About a year ago, on one such Friday, the unexpected happened: I was abruptly fired during a video call, with absolutely no warning. Right after we hung up, the board published a blog post announcing the decision. I was sitting in a hotel room in Las Vegas—the feeling was indescribable, as if a wonderful dream had suddenly turned into a nightmare.
Being publicly dismissed without warning triggered hours of chaos, followed by days filled with uncertainty. The strangest part was the state of “information chaos.” None of us could get clear answers about what had happened or why.
In my view, the entire episode was a failure of governance—even though everyone involved, including myself, acted in good faith. Looking back, I certainly wish I had handled things differently at the time. I’d like to believe I’m wiser and more thoughtful today than I was a year ago.
The event taught me that a board needs diverse perspectives and broad experience to navigate complex challenges. Good governance must be built on deep trust and credibility. I’m grateful that many people worked hard together to establish a stronger governance structure for OpenAI, allowing us to continue pursuing our mission of ensuring AGI benefits all of humanity.
Over the past year, my biggest realization has been how much I have to be grateful for, and how many people deserve thanks: my colleagues at OpenAI who chose to dedicate their time and energy to this dream; friends who supported us during the crisis; customers and partners who trusted us, collaborated with us, and relied on our technology to succeed; and everyone in my life who showed care and concern.
Afterward, we re-engaged with renewed unity and determination. I’m incredibly proud of the focus we’ve demonstrated since then. We’ve completed some of the best research work OpenAI has ever done. Our weekly active users have grown from roughly 100 million to over 300 million. Most importantly, we’ve continued releasing technologies that people truly love and use to solve real-world problems.
Nine years ago, we had no idea what we’d eventually become. Even today, we only have a fuzzy picture of the future. The development of AI is full of ups and downs, and we expect more challenges and opportunities ahead.
Some of these changes are joyful; others are difficult. The steady stream of research breakthroughs is exciting, and many former skeptics have now become strong supporters. Of course, we’ve also seen colleagues leave and become competitors. As teams grow larger, turnover is inevitable—and OpenAI has expanded at an extraordinary pace. Startups typically go through cycles of personnel change with each stage of scaling, but at OpenAI, this happens almost every few months. The past two years have felt like a decade of evolution for a typical company. When a leading company grows this fast, differences in goals and priorities among team members are bound to emerge. Meanwhile, as industry frontrunners, we inevitably face criticism and attacks, especially pressure from competitors.
Our vision has never changed, but our strategy continues to evolve over time. For example, we never imagined at the start that we’d need to become a product company—we thought focusing on groundbreaking research would be enough. Likewise, we didn’t anticipate needing massive amounts of capital to support our work. Now, we’re building things that were unimaginable just a few years ago. And surely, many more unforeseen challenges lie ahead.
We’re proud of what we’ve achieved so far in research and technological application, and we remain committed to deep exploration in safety and benefit-sharing. We firmly believe the best way to make AI systems safe is through gradual, measured deployment—giving society enough time to adapt and evolve alongside the technology. Learning continuously from real-world use and improving safety accordingly. We recognize the critical importance of being global leaders in safety and alignment research—work that must be guided by real-world feedback.
Today, we are confident we understand, by traditional definitions, how to build AGI. We believe that by 2025, we may see the first AI agents “enter the workforce,” significantly boosting enterprise productivity. This will be a historic moment. We remain convinced that putting powerful tools gradually into people’s hands is the best path to widespread and profound impact.
We’re already looking toward even more ambitious goals—the pursuit of true superintelligence. While we deeply value our current products, our mission is focused on creating that brighter future. With superintelligence, we could achieve virtually any goal. Superintelligent tools would dramatically accelerate scientific discovery and technological innovation, helping us solve complex problems beyond human capability alone, thereby greatly enhancing societal prosperity and abundance.
Right now, this may sound like science fiction—even talking about it might seem crazy. But that’s okay. We’ve faced similar skepticism before, and once again, embracing this challenge fills us with excitement. We’re very confident that in the coming years, everyone will come to see the future we see, and recognize how crucial it is to proceed with extreme caution while maximizing broad benefits and empowerment. Given the potential and possibilities of our work, OpenAI was never destined to be an ordinary company.
I feel both fortunate and deeply honored to be part of this endeavor.
(Thanks to Josh Tyrangiel for inspiration. I wish we could have spent more time talking.)
In those days, many people worked tirelessly for OpenAI and for me personally—but two individuals stood out above all.
Ron Conway and Brian Chesky went so far beyond the call of duty that I don’t even know how to adequately express my gratitude. I had long heard of Ron’s capabilities and relentless dedication, and over the past few years, I’ve received countless help and advice from Brian.
But only when fighting side-by-side in true emergencies can you truly witness someone’s extraordinary abilities. I can say without exaggeration that without their help, OpenAI might have collapsed. They worked around the clock for days until the situation was fully resolved.
Despite the immense pressure, they remained calm, demonstrated clear strategic thinking, and offered invaluable guidance. They prevented me from making several serious mistakes—and made none themselves. They leveraged their extensive networks to meet every need and successfully navigated numerous complex situations. I’m certain they did many things I’ll never even know about.
Yet what moved me most was their care, empathy, and support.
I thought I knew how to support a founder and a company—up to a point, I did. But I had never seen, nor even heard of, support delivered in the way they provided. Now I finally understand why they have such legendary reputations. They are truly exceptional and fully deserving of that recognition. Both possess rare abilities to deliver critical help against all odds and demonstrate unwavering commitment when it matters most. Our tech industry is better because they exist.
Of course, there are others like them. This is one of the special aspects of the tech industry—more important than people realize, and essential to keeping the whole ecosystem functioning. I look forward to passing on this spirit.
On a more personal note, I especially want to thank Ollie for his support during that weekend and throughout everything. He is unmatched in every way—no one could ask for a better partner.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@sama
