The meta unveiled a more cautious approach to the Chat GPT frenzy.

- Advertisement -

Facebook owner Metta on Friday unveiled its own version of the artificial intelligence behind apps like ChatGPT, saying it would give researchers access to address the technology’s potential vulnerabilities.

- Advertisement -

Meta describes its own AI, called LLaMA, as a “smaller, higher-performance” model designed to “help researchers advance their work,” including This can be seen as a veiled criticism of Microsoft’s decision to widely release the technology while keeping the programming code intact. secret

- Advertisement -

Microsoft-backed ChatGPT has taken the world by storm with its ability to generate finely crafted texts such as essays or poems in seconds using a technology known as Large Language Models (or LLM). .

LLM is part of a field called generative AI, which includes the ability to execute images, designs or programming code almost instantly on a simple application.

- Advertisement -

Normally a more established player in big tech, Microsoft has deepened its partnership with OpenAI, the creator of ChatGPT, and earlier this month announced that the technology would be integrated into its Bing search engine as well as the Edge browser. will be added.

Google, seeing a sudden threat to its search engine dominance, soon announced that it would release its own language AI, known as Bard.

But reports of troubling exchanges with Microsoft’s Bing chatbot — including issuing threats and talking about wanting to steal nuclear code or seduce a user from his wife — have gone viral, raising alarm bells that the technology is developing. do not have.

These problems, sometimes called hallucinations, could be better addressed if researchers improved access to expensive technology, Meta said.

The company said the full research was “limited by the resources required to train and run such large models.”

This was “impeding efforts to improve their robustness and reduce known problems, such as bias, toxicity, and the potential to generate misinformation,” Metta said.

OpenAI and Microsoft strictly limit access to the technology behind their chatbots, criticizing that they are choosing potential profits over improving the technology more quickly for society.

“By sharing the code for LLaMA, other researchers can easily test new methods to limit or eliminate these problems,” Meta said.




CR Bard

- Advertisement -

Hot Topics

Related Articles