What you need to know
- OpenAI CEO Sam Altman says superintelligence might only be “a few thousand days” away.
- Altman claims that AI will get better with scale and will help make meaningful improvements to the lives of people around the world, including fixing the environment, but a lot still needs to be done.
- A former OpenAI researcher warned that the firm wouldn’t know how to handle AGi and all it entails.
With the rapid prevalence of AI and the broad adoption of its tools, there has been a growing concern about job security, the possible extinction of humanity, and other critical issues. Close watchers of the technology like Elon Musk claim the world is on the verge of the biggest technological breakthrough with AI, However, there might not be enough electricity to power its advances.
We recently learned that AI tools like Copilot and ChatGPT consume large amounts of water for cooling — up to 3 water bottles to generate a mere 100 words. Despite the highlighted issues, OpenAI CEO Sam Altman has expressed his aspirations for achieving artificial general intelligence (AGI). However, a former researcher for the ChatGPT maker warned that the AI firm wouldn’t know how to handle all that it entails.
In a new blog dubbed The Intelligence Age, Sam Altman indicates:
“It is possible that we will have superintelligence in a few thousand days (!); it may take longer, but I’m confident we’ll get there.”
The CEO used the rapid advancement of deep learning and its power in general as the basis for his claims. “That’s really it; humanity discovered an algorithm that could really, truly learn any distribution of data (or really, the underlying “rules” that produce any distribution of data),” added Altman. “To a shocking degree of precision, the more compute and data available, the better it gets at helping people solve hard problems.”
In the past few months, top officials at the firm have made strong claims about the progression of their AI models and products. While at the 27th annual Milken Institute Global Conference, OpenAI COO Brad Lightcap indicated:
“In the next couple of 12 months, I think the systems that we use today will be laughably bad. We think we’re going to move toward a world where they’re much more capable.”
On another account, Sam Altman indicated:
GPT-4 is the dumbest model any of you will ever have to use again by a lot. It’s important to ship early and often and we believe in iterative deployment.
He even promised with a high scientific degree that OpenAI’s GPT-5 model would be smarter than GPT-4, which he admitted: “kind of sucks.”
Superintelligence is no walk in the park
While OpenAI CEO Sam Altman is focused on achieving superintelligence, he admits it’s going to be an uphill task:
“There are a lot of details we still have to figure out, but it’s a mistake to get distracted by any particular challenge. Deep learning works, and we will solve the remaining problems.”
And despite the challenges, Altman categorically indicates that AI will get better with scale. Consequently, it will “lead to meaningful improvements to the lives of people around the world,” including fixing the climate, the discovery of all of physics, and more.
This news comes after a handful of OpenAI’s super alignment team recently departed from the AI firm, including co-founder and Chief Scientist Ilya Sutskever, who announced he was departing from the firm to focus on a project that was “personally meaningful” — Superintelligence Inc.
OpenAI has been accused of prioritizing shiny products over safety processes, with former employees even referring to it as the “Titanic of AI.” Moreover, the ChatGPT maker reportedly sent invitations to GPT-4o’s launch before testing took place, pressuring the safety team to scheme through the entire process to meet the assigned deadline in under one week.
Elsewhere, OpenAI was recently dragged back to court by Elon Musk over a stark betrayal of its founding mission and alleged racketeering activities. It remains unseen how OpenAI plans to handle the wide array of issues as it edges closer to superintelligence.