- OpenAI’s Sam Altman and Chief Scientist Jakub Pachocki list future goals for the AI giant
- The world economy is now beginning to shape itself around artificial intelligence and is committed to providing tools that people would use
- The memo also reaffirmed OpenAI’s commitment to AGI, with one caveat: ensuring it benefits all of humanity
With modern AI solutions moving far beyond simple chatbots to agents and expected to evolve into operators, one could assume that the automation of everything is an ultimate goal.
However, this has been rejected by OpenAI CEO Sam Altman and Chief Scientist Jakub Pachocki, who said the goal of AI research and implementation is not to automate everything, but to empower people to make better decisions as AI improves their lives.
In a note titled ‘Built for the benefit of all’, marking a break from OpenAI’s AI modeling capabilities, two of the most important figures in the AI ecosystem wrote an unusual value-forward document outlining their future plans for AI.
AI for everyone equally?
The note highlighted three main focuses for OpenAI:
– Building an automated AI researcher
– Speed up the economy
– To give everyone on Earth a personal AGI
OpenAI estimates that by March 2028, a significant portion of its research will be performed by AI systems, in addition to its own researchers. This will help them traverse a ‘post-AGI world’.
This, combined with the focus on giving everyone an AGI, is an interesting perspective because it assumes everyone agrees on what AGI will look like. The definition is not set in stone and can vary from person to person and also at organizational level.
OpenAI’s statement also hints at what an AGI would be like, with an “automated AI researcher” providing both a path to AGI and an important cog.
OpenAI’s narrative that artificial intelligence benefits everyone globally is not new, but its focus on equality is interesting, especially given the timing: OpenAI’s memo appeared exactly the same day it filed confidential papers for its IPO, making it perhaps read more like PR than it might otherwise be perceived.
OpenAI’s latest models are state-of-the-art, but many feel that Anthropic’s now-banned Fable pushes the limits of models even further than what GPT currently offers in several segments. Training new models is increasingly capital intensive, even as new features are introduced, tested and refined over time.
OpenAI also has something of an image problem after it stepped in to replace Anthropic’s Claude and Mythos-class solutions for the US military earlier this year, a move the latter company maintains was necessary because the limitations it insisted on the use of its AI were important.
When OpenAI stepped in to replace Anthropic on classified networks, it was widely seen as willing to look past these restrictions to some extent, although Sam Altman insists that the same two principles (no domestic mass surveillance and use of force only allowed by humans) would apply, with many critics pointing to a ‘softer’ approach to the matter that OpenAI will bring to fill the contract with the military in the future.
The note therefore reads like a checklist for the future, but also paints OpenAI as a more magnanimous organization before its IPO, and that may be the primary intention here, but it does not weigh in on the growing power consumption, although one could also consider it a response or acknowledgment of a similar note from Anthropic on recursively self-improving AI solution, where it already works as an effective AI solution.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews and opinions in your feeds.
