Senior members The new Superinteigence Laboratory in Meta We were discussed away from the strong open AI AI model, the giant, and instead developing a closed model, New York Times reports.
Sources told the Times that Mita had completed the giant training, but delayed its issuance due to the overwhelming internal performance. When the new SuperIntellign declared laboratory was launched, the test stopped on the form.
Discussions are just discussions. META, CEO of Meta, will still need to record any changes, and a company spokesman Techcrunch that Meta at Open Source AI “has not changed.”
“We are planning to continue to launch the leading open source models,” the spokesman said. “We have not exported everything we have developed historically and expect to continue training a mixture of open and closed models to move forward.”
The spokesperson has not commented on the possible transformation of Meta away from the giant. If this happens until Meta can give priority to closed source models, this company’s main philosophical change will be shattered.
While META publishes more advanced source models internally, such as those who run Meta AI, I have made Zuckerberg. Open source An essential part of the company’s external strategy of the company – a way to maintain the development of artificial intelligence moves faster. Loud Lama family Openness as distinctive from competitors like Openai, which Zuckerberg I criticized publicly To be closed after partnership with Microsoft. But Meta is under pressure to income from ads as billions are poured into artificial intelligence.
This includes paying huge signing bonuses, nine numbers for the highest steadfast researchers, building new data centers, and covering the tremendous costs of developing artificial public intelligence (AGI), or “Superinteigence”.
Although there is one of the best artificial intelligence research laboratories in the world, Meta still falls behind its competitors such as Openai, Anthropic, Google DeepMind and Xai when it comes to marketing artificial intelligence.
If Meta gives priority to closed models, this may indicate that openness was a strategic, not ideological play. Previous comments from Zuckerberg hinted to a contradiction towards the commitment to open Meta models. On a Podcast He said last summer:
We are very open, but I haven’t had to release everything we do. I mainly tend to believe that open sources will be useful for society and also good for us because we will benefit from innovations. However, if there is some qualitative change at some point in what the thing could be capable of, and we feel that it is not responsible for opening its source, we will not do that. It is very difficult to predict.
Closed models will give Meta more control and more ways to achieve income-especially if he believes that the talent you have gained can provide competitive and better performance in its class.
Such a shift can reshape the artificial intelligence landscape. Open source momentum, largely driven by dead and models like Llama, can slow down The open model remains. Energy can be swinging towards the main players who have closed environmental systems, while open source open development may remain a product of popular efforts. The effects of ripples through the ecosystem will continue to start operating, especially for smaller companies that focus on the exact and safety transition and the typical alignment that depend on access to open foundation models.
On the global theater, Meta’s retreat from the open source to China, which embraced the open source artificial intelligence-such as Deepseek and Moonshot ai As a way to build local power and global influence.
https://techcrunch.com/wp-content/uploads/2025/02/GettyImages-2195497483.jpg?resize=1200,800
Source link