OpenAI reportedly failed to keep promise to control potentially dangerous AI
The team assembled to combat dangerous AI tech within the company has been disbanded.
As OpenAI continues to navigate the backlash to its use of a Scarlett Johansson soundalike for its GPT-4 language model, the company is now being accused of failing to properly combat dangerous AI. A new report claims that OpenAI neglected to provide previously promised resources to control potentially dangerous AI, and has now disbanded the team dedicated to doing so.
A report from Fortune brought new accusations against ChatGPT creator OpenAI. According to sources at the company, OpenAI failed to provide adequate resources to its Superalignment team, which was formed last summer to combat potentially dangerous artificially intelligent technology. OpenAI had committed to using 20 percent of its computing power to accomplish this goal. Now, the Superalignment team at OpenAI has been disbanded altogether.
Source: OpenAI
The report also alleges that OpenAI leadership routinely turned down requests from the Superalignment team for computer processors and specialized chips necessary to complete the group’s task. Sources at the company say OpenAI never allotted the promised 20 percent of computing power to controlling dangerous AI.
This of course comes as OpenAI is embroiled in controversy over its use of a voice eerily similar to Scarlett Johansson’s for its GPT-4 language model, something that the actress condemned in a statement. Be sure to visit our OpenAI topic page for more stories about the company leading the charge on the development of artificially intelligent technology.
-
Shacknews
replyDonovan Erskine posted a new article, OpenAI reportedly failed to keep promise to control potentially dangerous AI
-
I just posted this elsewhere but it’s worth reposting here: https://www.wheresyoured.at/sam-altman-is-full-of-shit
-
