OpenAI, Google and Anthropic have encountered problems creating new artificial intelligence models, which could negatively impact the development of artificial general intelligence (AGI). About this reported Bloomberg sources with knowledge.
“The bubble is bursting a little”: Bloomberg learned about OpenAI and Google’s problems with new AI models
Author: Bogdan Muzychenko
According to them, in September OpenAI completed initial training of a new neural network codenamed Orion. The company hoped that the model would surpass previous versions, but it could not achieve the desired performance, for example, Orion could not answer programming questions, agency interlocutors say.
The ChatGPT developer is “not alone in its problems” – Google is experiencing difficulties regarding AI, as the upcoming version of its Gemini model “did not live up to expectations,” as well as Anthropic, which does not have time to release the Claude 3.5 Opus neural network on time, writes Bloomberg.
One of the obstacles companies have faced is that it is becoming more difficult for them to find new untapped sources of high-quality data that are used to train AI, agency sources said. Moreover, they say, even “modest improvements” may not justify the “huge costs” of developing new models.
The failures of OpenAI, Google and Anthropic could affect the development of artificial general intelligence, which is expected to match or surpass humans in many intellectual tasks, writes Bloomberg. “The AGI bubble is bursting a little bit,” Margaret Mitchell, chief ethics officer at AI startup Hugging Face, told the agency.
Earlier, The Information sources reported a slowdown in innovation at OpenAI. They pointed out that Orion’s new neural network is barely superior to its predecessor GPT-4, including being less easy to program and may be more expensive to use in data centers.
At the end of October, The Verge’s interlocutors said that the creator of ChatGPT was preparing to release a new AI model in December, which the head of the company Sam Altman called a fake. According to Bloomberg, OpenAI is unlikely to release Orion until early 2025, as it is “still not at the level the company would like.”
Cover photo: akinbostanci /
Source: rb.ru
