Meta’s Own AI Bot: Zuckerberg announces Meta to take on ChatGPT with their LLaMA AI model

Meta’s Own AI Bot: Zuckerberg announces Meta to take on ChatGPT with their LLaMA AI model

Feb 28, 2023 - 09:30
 0  30
Meta’s Own AI Bot: Zuckerberg announces Meta to take on ChatGPT with their LLaMA AI model

OpenAI’s ChatGPT, Google’s Bard, and Microsoft’s Prometheus and Bing Chat has a new large language model-powered artificial intelligence to deal with. Meta, formerly known as Facebook, introduced its own AI bot today, called LLaMA or the Large Language Model Meta AI.

In a Facebook article, CEO Mark Zuckerberg detailed his company’s addition to the buzzy AI technology sphere.

Meta throws their hat into the ring
“Today, we’re launching LLaMA, a new state-of-the-art AI big language model intended to assist academics progress their work,” Zuckerberg wrote on Twitter. “LLMs have shown a lot of potential in producing writing, having discussions, summarising written material, and more difficult jobs like solving math equations or predicting protein structures,” he added.

However, the Meta executive did not specify which of those duties LLaMA could presently perform.

Also read: Facebook & Instagram owner Meta to launch monthly subscription service priced at $11.99

To cut through all that was said, Zuckerberg’s only comment on the big language model in Friday’s statement was that Meta is “committed to this open model of research, and we’ll make our new model accessible to the AI research community.”

Meta provided considerably more information in a business blog entry released Friday afternoon that included a link to a complete research report on the AI and its Github model card. “LLaMA, like other large language models, works by taking a series of words as input and predicting a next word to iteratively create text,” according to the firm.

How different is LLaMA from ChatGPT, Bard, and Bing’s Chat?
LLaMA was trained on content from 20 distinct languages, according to Meta. According to the model card, training data included freely accessible text from CCNet, C4, Wikipedia, ArXiv, and Stack Exchange.

Also read: Google’s AI goof’s up: BARD AI’s mistake during demo sends Google’s stocks tumbling by $100 bn

Meta’s AI is characterised as a “smaller base model” that “requires far less processing power and resources” than other big language models, and LLaMA will be offered in a variety of configurations. The business also stressed its dedication to transparency and responsible AI development and repeated that access will be granted only to AI researchers.

“To keep the integrity and avoid misuse, we are distributing our model under a noncommercial licence centred on study use cases. Academic researchers, those associated with groups in government, civic society, academia, and business research laboratories around the globe will be given access to the model on a case-by-case basis, according to the company.

LLaMA has not yet been incorporated into any of Meta’s goods or networks, including Instagram and Facebook, revealed a source close to the matter. The source also stated that the business has no information about a public trial or extended public access at this time.

Is Meta’s LlaMa better than the other text generators?
At the moment, it’s difficult to predict how LLaMA will fare against the AI efforts of other tech behemoths in a quickly crowded sector. However, nearly every recent big language model debut has had its own set of hiccups.

Also read: AI goes bonkers: Bing’s ChatGPT manipulates, lies and abuses people when it is not ‘happy’

Google and Microsoft both incorporated AI-text generators into their search platforms and unintentionally advertised incorrect information. Then there are the unsolved issues of copyright, privacy, whether AI can develop “feelings,” and how to handle the methods that many people on the internet have discovered to circumvent limits and produce objectionable content or even viruses.

In its blog, Meta recognised some of these pitfalls. “More study is needed to handle the dangers of bias, toxic remarks, and hallucinations in big language models. “LLaMA, like other models, faces these challenges,” the firm stated. Meta, on the other hand, appears to think that its open, research-oriented strategy is a move forward in resolving AI’s numerous, ongoing problems.

Read all the Latest NewsTrending NewsCricket NewsBollywood News,
India News and Entertainment News here. Follow us on FacebookTwitter and Instagram.

What's Your Reaction?

like

dislike

love

funny

angry

sad

wow