Facebook’s owner has announced plans to make a version of the artificial intelligence technology behind ChatGPT freely available to researchers and companies.
The latest version Metas The large language model, called LLaMa 2, will compete with OpenAI’s GPT-4, which powers chatbot applications.
The technology came to the fore after it was released last November Chat GPT Built a user base of 100 million in just two months.
In a press release, Meta said it decided to open up access to LLaMa 2 so that businesses and researchers can access more AI tools and experiment with them.
LLaMa 2 is the second generation of Meta large-scale language models, first released in February.
Compared to LLaMa 1, the second version of the model was trained on 40 percent more data, according to Meta.
AI has been around for decades and is used in satellite navigation systems and virtual assistants like Siri, but large language models usher in a new era of generative AI.
The term describes algorithms that can be used to create new content, including text, audio, images, and video.
AI can now speak human language fluently
Globally, regulatory developments to ensure companies are transparent about how they train their AI and prevent misinformation and bias are still in their early stages.
After launching ChatGPT last year, Google responded with its AI-powered search engine Bard, and Microsoft launched its own version called BingChat.
Unlike models developed by OpenAI and Google, Meta is making LLaMa 2 open source, meaning the code is free to download and modify.
read more:
Mark Zuckerberg trains with UFC champion ahead of possible fight with Elon Musk
Can artificial intelligence help dating apps succeed?
OpenAI launched ChatGPT last year, and tech giants followed suit
“We believe an open approach is the right approach for AI model development today, especially in the rapidly evolving field of generative technology,” Meta said in a statement Tuesday evening.
“Open access to today’s AI models means a generation of developers and researchers can stress test them as a community, identifying and fixing problems quickly.”
The model will be available through Azure, Microsoft’s cloud computing platform, as well as Hugging Face and Amazon Web Services.
Meta CEO Mark Zuckerberg said in a Facebook post: “Open source drives innovation because it empowers more developers to build with new technologies… I believe that if The more open the ecosystem, the more advancements it will bring, which is why we open-sourced LLaMa 2.”
Oli Buckley, professor of cybersecurity at the University of East Anglia, said it was important to improve our understanding of artificial intelligence before the code was made public.
“Every major technological innovation of the past 100 years has had some potential for abuse, and there is no shortage of people who are ready and willing to actually abuse it,” he explained.
“The difference between a nuclear weapon and a nuclear weapon [large language model] Yes, we can at least identify the people who are procuring the parts needed to build nuclear weapons, but it is much harder to identify who is using artificial intelligence to do bad things. “
Mhairi Aitken, an ethics researcher at the Alan Turing Institute, the UK’s National Institute for Data Science and Artificial Intelligence, said that the openness of Meta does not mean the transparency of the model training content.
“The worry here is that instead of seeing the democratization of AI, as models become more accessible and used in more widespread ways, we’ll see more and more marginalized or disadvantaged communities. suffer the worst of its effects, while developers find new ways to profit from its use,” she added.