The battle for dominance in the artificial intelligence (AI) industry is on, and tech giants are rushing to ensure they are not left behind.
OpenAI’s ChatGPT, released in November last year, was an instant hit with the company securing a $10 billion investment from Microsoft.
Meta is the latest entrant in the AI industry, announcing the release of an AI language model known as LLAMA.
Meta’s powerful AI language model leaked online
Meta’s LLAMA is not open to the public like ChatGPT, with the language model an open-source package that the AI community can access.
According to the company, the goal of this model is to democratize access to AI and support research.
The company published a blog post saying:
Even with all the recent advancements in large language models, full research access to them remains limited because of the resources required to train and run such large models.
Meta is trying to ease the research process by making these large language models easily accessible.
A week after the tech giant officially released LLAMA, the language model was leaked online.
A downloadable torrent of the system was released on 4chan and has been spread to multiple AI communities – triggering a discussion on the most prudent way of sharing important research when technology is fast-evolving.
Some have said that the leak on this language model is dangerous to individuals and companies.
A tweet by one cybersecurity researcher noted that LLAMA could increase the number of personalized spam and phishing campaigns. Ladish noted that open-sourcing these language models were not a good idea.
Well Meta's 65 billion parameter language model just got leaked to the public internet, that was fast.
Get ready for loads of personalized spam and phishing attempts. Open sourcing these models was a terrible idea
— Jeffrey Ladish (@JeffLadish) March 4, 2023
However, some believe that open-sourcing is important to create safeguards for AI models, adding that similar large language models had already been made public, and no harm had been done.
Several AI researchers have confirmed that the leaked system is legitimate and matches the LLAMA model that Meta distributed.
While Meta has not confirmed whether the leaked version is legitimate, the managing director of Meta AI, Joelle Pineau, noted that some people were trying to bypass the approval process.
Is Meta’s LLAMA a dangerous AI?
There has been much discussion about the potential of AI, with some, including Twitter’s new boss Elon Musk, believing that lack of safeguards could be detrimental.
Therefore, the leak of Meta’s LLAMA model has created mixed reactions.
— Elon Musk (@elonmusk) February 28, 2023
Unlike ChatGPT, LLAMA needs technical expertise to set up and start using. The language model can only be operational by people familiar with setting up servers and dev ecosystems for complex projects.
The computational demands of this model would be the leading constraint on the effective use of LLAMA. Nevertheless, the language model is still a powerful tool.
According to Meta, the 13 billion parameters model will outperform OpenAI’s 175 billion parameters of OpenAI’s GPT-3 model.
There are also concerns that this leak could dampen the trust between tech giants and academics with whom they share the research information.
Experts believe the leak could make it more challenging for tech companies to release products.
- How to Buy Meta Stock in 2023
- Meta Slashes VR Headset Prices amid Sluggish Demand & Competition
- Meta Platforms Shares Blueprint for AR/VR Hardware with Employees
What's the Best Crypto to Buy Now?
- B2C Listed the Top Rated Cryptocurrencies for 2023
- Get Early Access to Presales & Private Sales
- KYC Verified & Audited, Public Teams
- Most Voted for Tokens on CoinSniper
- Upcoming Listings on Exchanges, NFT Drops