‘OpenAI has achieved more than I dared to dream’, says Sam Altman on ChatGPT maker’s ‘crazy’ 10 years of success
In a heartfelt blog celebrating ChatGPT maker OpenAI’s 10th anniversary, CEO Sam Altman talked about “a decade of breakthroughs, learnings, and the path to AGI that benefits all humanity.”
Sam Altman’s post reminded us of the turbulent beginning of the artificial intelligence platform in the last 10 years and its journey to success. “OpenAI has achieved more than I ever dared to dream; we set out to do something crazy, improbable and unprecedented. From an extremely uncertain start and against all reasonable odds, with continued hard work it now looks like we have a chance to succeed in our mission,” he said.
He noted that the company officially started operating at the beginning of January 2016, and that they announced themselves to the world 10 years ago on December 11, 2015.
‘It was a wildly fun time: despite a small chance of success, we had a belief that we held deeply.’
Altman added that a lot has changed in 10 years, but daily life is not much different. “Ten years is a very long time in one sense, but in terms of how long it usually takes for the curve in society to bend, it’s not long at all. Although daily life doesn’t feel all that different from what it was ten years ago, the field of possibilities before all of us today feels very different from when we were 15 nerds sitting around trying to figure out how to make progress,” he said.
He added, “Looking back at photos from the early days, I was first struck by how young everyone looked. But then I was struck by how irrationally optimistic and how happy everyone looked. It was a wildly fun time: although we were deeply misunderstood, we had a deep belief, a sense that this was so important that it was worth working hard, even with a small chance of success, with very talented people and a sharp focus.”
I’ve had a few wins (and a lot of losses)
He added that the team had spent time building and understanding and had “a few wins (and a lot of losses)”, but “they always had a great spirit of finding the next obstacle in front of us”.
“In those days, it was difficult to figure out specifically what to work on, but we built an incredible culture to enable discovery. Deep learning was obviously a great technology, but it didn’t seem right to develop it without having experience running it in the real world,” he shared.
He summarized: “In 2017, we had a few key results: our Dota 1v1 results, where we took reinforcement learning to new levels of scale. The unsupervised emotion neuron, where we saw a language model undeniably learn semantics rather than syntax. And we had the reinforcement learning result from human preferences, which showed a rudimentary way to align an AI with human values. At this point, the innovation was not yet complete, but we knew we had to scale up. Each of these results has tremendous computational power.”
ChatGPT launch: ‘AGI is no longer a crazy thing to consider’
Sam Altman also shared the launch of ChatGPT announced in 2022, stating that “suddenly AGI is no longer a crazy thing to consider.”
“The last three years have been extremely busy, full of stress and heavy responsibilities, as this technology has integrated into the world at a scale and speed that no technology has ever had before. This required an extremely demanding implementation in which we had to develop new muscle immediately. Going from scratch to a large company during this time was not easy and required us to make hundreds of decisions a week. I am proud of how many of them the team got right, and the ones we got wrong were mostly my fault,” he added.
He added that the company is making “new kinds of decisions,” such as how AI will “provide maximum benefit to the world,” developing an iterative deployment process (where newer updates are released after the product is released). Altman explained this as a way for intuition, society, and technology to evolve together.
“This was pretty controversial at the time, but I think it turned out to be one of our best decisions ever and has become the industry standard,” he added.
‘Artificial intelligence can do extraordinary things, we are optimistic about our business’
Altman added that after 10 years at OpenAI, he believes the company has “an AI that can outperform many of our smartest people in our toughest intellectual competitions.”
“The world has been able to use this technology to do extraordinary things, and we expect even more extraordinary things next year. The world has also done a good job of mitigating potential downsides so far, and we need to work to continue doing that. I’ve never felt more optimistic about our research, our product roadmaps, and our overall outlook on our mission,” he added.
The CEO added that he is confident that OpenAI will be able to create superintelligence within the next 10 years. “I expect the future to be strange; in a sense, daily life and the things we care most about will change very little, and I’m sure we will continue to focus on what other people do rather than what machines do. In another sense, the humans of 2035 will be capable of doing things that I don’t think we can easily imagine right now,” he said.
He added that his mission is to ensure that YGZ benefits all humanity. “We still have a lot of work ahead of us, but I’m really proud of the trajectory the team is taking us on. We’re seeing tremendous benefits in what people are doing with technology today, and we know there’s much more to come in the next few years,” he said.
