[ad_1]
Yuan release The company released its flagship plain text model Llama 3.1 on Tuesday (July 23), which is the best-performing open source artificial intelligence (AI) model to date. This is likely to be the last model the company will make available to users in the European Union.
Meta, the parent company of Facebook and Instagram, is under pressure from EU regulators over its Privacy Policy around Artificial Intelligenceand its “Pay or confirm” model. In response, the company decided not to bring its latest artificial intelligence product to the EU.
Last week, the company said it would not release its latest multimodal AI model in the European Union, citing the “unpredictability of the European regulatory environment,” a Meta spokesperson said. Tell EuractivPreviously, the company abandoned plans to launch Meta AI, which would have used users’ public posts to train artificial intelligence.
In addition to text, multimodal models can also handle content formats such as video, images, and sound.
A Meta insider with knowledge of the matter told Euractiv that under the current circumstances Meta will not release any future multimodal models in the EU, and it is unclear whether Meta will apply the same strategy to text-based models.
Meta argues that the company must exclude the EU from its model due to regulatory unpredictability, which would ultimately harm the interests of European businesses and consumers.
This has led to a technological gap between Europe and the rest of the world, Meta deputy privacy officer and vice president of policy Rob Sherman told reporters. Financial Times. Sherman also hinted that the EU might not introduce multimodal or text-based transport again.
Developers can download open source models (usually for free) and deploy them into their own applications, which also enables them to change and audit the model. Closed source models like OpenAI’s ChatGPT are available through subscription or similar interfaces.
Many companies adopt these open source models to reduce costs and gain greater control over their own software and data.
These models are released under an open source license, which allows downloading and using them. Certain restrictions.
To prevent EU users from using Meta’s multimodal model, Meta could put some geo-restrictions on the license or block access from European IP addresses.
Not open enough
How open Meta’s AI development is is a matter of debate in the industry. While Meta shared more technical details about Llama 3.1 with closed-source developers, the company did not reveal specific information about what data the model was trained on.
“Keeping the training data is not in the spirit of open source development.” Zuzanna Warso, research director at the nonprofit Open Futures, told MIT Technology Review in March. Wossow is participating in the Open Source Institute, the de facto arbiter of open source matters.
The idea behind open source software is to make technology available to a large number of developers outside of the original creating entity, while also enabling developers to improve the software.
Meta’s motivation for opening its model to developers isn’t entirely altruistic.
“I hope our model can be further improved through open source contributions,” said Meta founder and CEO Mark Zuckerberg. told investors in April.
Zuckerberg said that as Meta’s AI models get better, the company may eventually “let people pay for access to bigger AI models and more computing power.”
(Editing by Zoran Radosavljevic and Eliza Gkritsi)
Read more by Euractiv
[ad_2]
Source link
