Meta Prepares to Launch Commercial AI Model to Catch Rivals
Meta is gearing up to release a commercial version of its artificial intelligence (AI) model, allowing startups and businesses to leverage its technology for building custom software. By doing so, Meta aims to compete with Microsoft-backed OpenAI and Google, who have made significant strides in the development of generative AI. Powered by large language models (LLMs) trained on extensive datasets and requiring substantial computing power, Meta’s software has the capability to generate text, images, and code.
Widely Available and Customizable: Meta’s New AI Model
While Meta had previously released its language model, LLaMA, to researchers and academics earlier this year, the upcoming version will be more widely accessible and customizable by companies. According to sources familiar with the plans, the release is imminent. Meta intends to make its LLMs “open-source,” meaning that details of the new model will be publicly available. This differs from the approach of competitors like OpenAI, whose latest model, GPT-4, is considered a “black box” where third parties cannot access the data and code used to build the model.
Meta’s Open Source Approach and Catching Up with Rivals
Speaking at a conference in Aix-en-Provence, Meta’s Vice President and Chief AI Scientist, Yann LeCun, stated that the AI competitive landscape will soon change with the introduction of open-source platforms that match the capabilities of proprietary models. Meta’s move towards openness aligns with its global affairs chief, Nick Clegg’s, belief that openness is the best way to address concerns surrounding AI. Furthermore, an open model allows companies of all sizes to enhance the technology and develop applications, aiding Meta in catching up with its competitors.
Meta’s Ambitions and Increased Commercial AI Investment
Meta, previously known as Facebook, has been investing heavily in AI research and development, as it seeks to establish itself as a dominant player in the AI field. CEO Mark Zuckerberg’s announcement of building a metaverse, a digital world filled with avatars, has resulted in significant expenditure of over $10 billion annually. However, this costly ambition has faced investor pushback, leading Meta to ramp up its AI investment. To achieve this, Meta established a generative AI unit led by Chief Product Officer Chris Cox, focusing on both research and product development to create new businesses.
Open Source Benefits and Potential Monetization Plans
The adoption of open-source models provides several advantages, such as greater user uptake, leading to more data for AI processing and improved capabilities. Additionally, open source allows researchers and developers to identify and address bugs, enhancing technology and security simultaneously—a critical consideration for companies like Meta that have faced scrutiny over privacy and misinformation scandals. Although Meta’s technology is currently free and open source, the company has reportedly explored the possibility of charging enterprise customers for fine-tuning the model using proprietary data. However, no plans to charge customers are expected in the upcoming release.
Considerations and Risks of Open Source AI
While open-source AI models offer numerous benefits, there are also potential risks and challenges. The misuse of open source AI by bad actors, such as the generation of child sexual abuse imagery, remains a significant concern. Additionally, previous Meta AI models, like BlenderBot 2, have been found spreading misinformation, although efforts have been made to enhance the content generation process in BlenderBot 3. Legal and regulatory risks related to intellectual property and copyright also exist. Comedian and actor Sarah Silverman has recently filed a lawsuit against both Meta and OpenAI, claiming her work was used to train models without consent.