Meta, the parent company of Facebook, is entering the artificial intelligence (AI) race by releasing its latest AI model, LLaMA, for free. This move puts Meta in direct competition with ChatGPT, another tech giant, to dominate the commercial use of AI models.
LLaMA, a large-scale language model, is capable of mimicking human speech, writing, and even thought processes. It collects content from the internet and rearranges it to generate new and unique outputs. Similar models power popular chatbots like OpenAI’s GPT-4 and Microsoft’s Bing.
According to Nick Clegg, Meta’s Global President, the company believes in the wisdom of crowds. By making LLaMA available for other companies to develop their own products, Meta aims to track how people use and deploy technology by harnessing the expertise of a wide range of professionals.
While Meta is providing this AI model for free, other companies are taking steps to limit the widespread use of their technology due to concerns about misuse. Meta itself recently canceled plans to release an open-source version of Voicebox, a technology that mimics the human voice, over fears of deceptive use.
Clegg clarified that Meta’s decision to release LLaMA for free may seem counterintuitive, but it aligns with their goal of fostering innovation. Large language models have revolutionized the tech industry by enabling capabilities such as answering complex questions, writing term papers, and passing exams. However, these models also have the potential to generate highly inaccurate information, commonly referred to as hallucinations.
Experts warn about the misuse of AI, citing potential issues such as the spread of disinformation and the development of AI-driven robot soldiers on the battlefield. Nevertheless, Clegg argues that the current technology lacks the necessary agency and autonomy to pose an existential threat.
Meta sees LLaMA as both a sword and shield. Clegg claims that advancements in AI have reduced hate speech on Facebook from approximately 50% to just 0.02% over the years. The AI model may have additional undisclosed features, potentially aiding credit card companies in detecting fraud and helping doctors improve diagnoses.
By providing access to powerful search capabilities without sharing data with language modelers, Meta’s AI model empowers companies to perform operations on their own computers. This breakthrough technology offers search capabilities beyond what traditional search engines can provide.
Meta attempts to regulate the potential misuse of its AI model through a usage policy that users must agree to. However, enforcing compliance can be challenging. When it comes to potential regulations, Clegg argues for regulating the use of technology instead of the technology itself.
The future of AI models remains uncertain, but Clegg believes the industry alone should not decide whether these models should be open source or not. He suggests that the government should play a role in determining the safety and regulations surrounding such advanced technology.
In conclusion, Meta’s release of LLaMA, its latest AI model, for free puts it in direct competition with ChatGPT and other tech giants. While concerns about the misuse of AI persist, Meta believes in the power of the crowd to track and deploy technology responsibly. This move aims to foster innovation and improve existing products while minimizing the potential for fraudulent or deceptive use. As the AI industry progresses, the discussion surrounding regulations and safety becomes increasingly critical.
