Last Updated: 6 January 2025
In a blog post, OpenAI CEO Sam Altman announced that the company is setting its sights on "superintelligence." According to Altman, OpenAI is now confident it "know[s] how to build AGI [artificial general intelligence]" and is beginning to explore the possibilities beyond it.
"We love our current products, but we are here for the glorious future," Altman wrote. He described superintelligence as a transformative tool that could "massively accelerate scientific discovery and innovation well beyond what we are capable of doing on our own." He also emphasized its potential to create "abundance and prosperity."
Altman suggested that superintelligence could be only "a few thousand days" away, potentially arriving much sooner than many expect. However, this rapid progress raises significant questions about safety, ethics, and societal impact.
AGI, or artificial general intelligence, means AI that can handle many tasks better than humans. It’s like having a machine that can outperform people in most important jobs. OpenAI’s partnership with Microsoft also sees AGI as systems capable of making $100 billion in profits. Superintelligence takes things a step further, describing AI that is far smarter than humans in every way.
Altman believes these advancements are just around the corner. He said AI agents could "join the workforce" as early as 2025, transforming how businesses operate. "Iteratively putting great tools in the hands of people leads to great, broadly-distributed outcomes," he shared.
But Altman also knows AI isn’t perfect yet. Today’s systems still make mistakes, cost too much, and sometimes "hallucinate" or give false answers. That’s why OpenAI is focusing on steady improvements while ensuring these systems benefit society.
OpenAI’s push toward superintelligence comes with serious challenges, particularly in ensuring that advanced systems align with human values. "We don’t have a solution for steering or controlling a potentially superintelligent AI," Altman admitted. He also warned that traditional alignment methods, like reinforcement learning from human feedback, may not work for superhuman systems.
OpenAI’s Superalignment team has been testing whether smaller models can supervise more capable ones. Early experiments show promise, with GPT-4 supervised by GPT-2-level models achieving near GPT-3.5 performance. However, Altman noted that many alignment challenges remain unsolved.
Some critics argue that OpenAI’s growing commercial focus has overshadowed its commitment to safety. Recent restructuring efforts and the departure of key safety researchers have fueled these concerns.
The rise of superintelligence could reshape industries but also disrupt economies. Altman compared it to the invention of the transistor, predicting that AI will soon become a commoditized tool driving economic transformation. "There will be shockingly capable models, widely available, used for everything," he stated.
Altman also called for new economic models to fairly compensate creators whose work trains AI systems. OpenAI has faced legal challenges, including copyright infringement lawsuits, highlighting the complex relationship between innovation and intellectual property rights.
Since its founding in 2015, OpenAI has grown from a small research lab to a leading AI company. The 2022 launch of ChatGPT marked a turning point, with Altman noting that it "kicked off a growth curve like nothing we have ever seen." Today, OpenAI’s platforms boast over 300 million weekly active users.
Reflecting on this journey, Altman admitted that building a company in such a fast-evolving field is challenging. "Moving at speed in uncharted waters is an incredible experience, but it is also immensely stressful," he wrote. Despite the difficulties, Altman remains committed to OpenAI’s mission to create AI that benefits everyone.
OpenAI plans to prioritize transparency and collaboration as it moves toward superintelligence. To involve the broader research community, it has open-sourced its alignment research and launched a $10 million grants program.
"We are here for the glorious future," Altman wrote, emphasizing the need for global cooperation. He believes that by acting responsibly, OpenAI can help superintelligence become a force for universal prosperity.
