[ad_1]
On Reddit AMASam Altman, CEO of OpenAI, admitted that the lack of computing power is one of the main factors preventing the company from shipping products as much as it would like.
“All of these models have become too complex,” he wrote in response to a question about why OpenAI’s next AI models are taking so long. “We also face a lot of constraints and difficult decisions about allocating our computing to many great ideas.”
a lot Reports It notes that OpenAI has struggled to secure sufficient computing infrastructure to run and train its generative models. Just this week, Reuters, citing sources, said OpenAI has been working for months with Broadcom to create an AI chip to power the models, which could arrive as soon as 2026.
Partly as a result of the strained capacity, Altman said, the real-life conversation feature OpenAI is introducing for ChatGPT, Advanced Voice Mode, won’t be debuting in April anytime soon. At its press event in April, OpenAI showed off ChatGPT running on a smartphone and responding to visual cues, such as what clothes someone was wearing, in view of the phone’s camera.
Reporting Fortune later revealed that the demo had been rushed to draw attention away from the Google I/O developers conference, which was taking place that same week. Many within OpenAI didn’t think GPT-4o was ready to be unveiled, and apparently the audio-only version of Advanced Voice Mode was delayed by several months.
At the AMA, Altman noted that the next major release of OpenAI’s image generator, DALL-E, has no timeline for release. (“We don’t have a release plan yet,” he said.) Meanwhile, Sora, OpenAI’s video creation tool, has been hampered by “the need to perfect the model, and get the safety/impersonation/other stuff right,” Kevin Weil wrote. , chief product officer at OpenAI, who also participated in the AMA.
Sora has reportedly suffered technical setbacks that leave it at a disadvantage against competing systems from Luma, Runway, and others. per The original information system, unveiled in February, took more than 10 minutes of processing time to create a one-minute video.
In October, Tim Brooks, one of the co-leaders of the Sora project, left for Google.
Later in the AMA, Altman said that OpenAI was still considering allowing “NSFW” content in ChatGPT “someday” (“We absolutely believe in treating adult users like adults,” he wrote), and that the company’s top priority is improving the o1 series of models. Inference” and its successors. OpenAI Previewed A number of features are coming to o1 at the DevDay conference in London this week, including image understanding.
“We have some very good releases coming later this year,” Altman wrote. “There’s nothing we’d call GPT-5, though.”
[ad_2]