Meta’s Paras Sharma on the role of AI in creation of content , moderation and personalisation
In the domain of AI and content creation, there has been a rise in concern with regards to the imperative of responsible navigation. Beyond the allure of technology, the crux lies in its consequential impact. This understanding resonates with Meta, which claims to transcend mere rhetoric into tangible action. As per the company, it actively engages in collaborative initiatives, forging partnerships within and beyond its organisational sphere to ensure the ethical comportment of its AI innovations. The company claims to have incorporated safety measures into its AI models.
Continue reading this story with Financial Express premium subscriptionSubscribe NowAlready a subscriber? Sign in
Meta claims to actively solicit feedback, treat mistakes as learning opportunities, and demonstrate a resolute commitment to refinement. In a conversation with BrandWagon Online, Paras Sharma, director of content and community partnerships, Meta talks about AI in content creation, responsible AI,besides the future of more personalised feed, among others. (Edited Excerpts)
Meta has utilised AI across various domains for many years. The foundation of content feeds on platforms such as Facebook and Instagram has been driven by AI, primarily through recommendation engines. Now, with a focus on generative AI, the aim is to enhance content creation and user experience by augmenting both aspects.
There are four pillars through which Meta approaches this. First is the developmental framework, which includes open-source technologies such as Llama and Llama 2, large language-based models that collaborate within a global ecosystem. The second pillar involves PyTorch, a deep learning framework that accelerates the prototyping to production development pace.
From a user experience perspective, Meta leverages AI to make content feeds more intuitive. The recommendation engine, powered by AI, ensures users see content tailored to their preferences, resulting in a seven percent growth in time spent within the ecosystem.
Meta has introduced AI conversational bots as well, available across a family of apps. The launch of AI stickers allows users to create stickers using AI technology. Additionally, AI personalities have been crafted, representing global figures in various fields, enabling users to engage in discussions with these virtual personalities.
For creators, Meta offers features like using Emu, a foundational layer of technology, to generate background images and enhance content. Creators can also leverage AI personalities to engage with diverse audiences on specific topics.
In advertising and marketing, Meta has introduced AI tools that enable brands to create multiple campaigns efficiently. These tools generate creative content adaptable to various forms and surfaces, empowering small and medium-sized businesses globally to build effective campaigns.
What are the steps Meta has taken to ensure diversity and inclusion in AI generated content?
The company is acutely aware that, in the process of developing new technology and advancing its role in extending the reach of this technology to more people, it collaborates with all relevant bodies in a responsible manner. The approach involves opening up technologies while ensuring the absence of biases or any undesired elements, with multiple measures in place to address these concerns. This responsibility extends not only to the company but to the entire ecosystem and, notably, its family of apps, catering to approximately 3.8 billion monthly active users. Facebook alone has three billion monthly active users.Managing this massive user base underscores the significant responsibility to ensure that every new element we construct and introduce incorporates the same level of detail as we have consistently applied to all our existing platforms.
What are the upcoming tools and features for ad creatives?
We have recently introduced generative AI for ad creative in Ads Manager. This new tool is going to roll out next year, aiming to enhance ad campaigns for advertisers.
How do you moderate content across your products? What is the role AI is expected to play?
In handling content moderation across our family of apps, we rely on machine learning as humans alone can’t manage the sheer volume of content created daily—hundreds of millions of pieces. Our AI presence, rooted in various manifestations, has been long standing. The foundational layers, including a large language model and its PyTorch framework, are open source, inviting developers to contribute and build upon them. Even as users leverage generative AI for background changes, transparency remains key. We’re committed to indicating when AI is at play, ensuring responsible user experience across our platform.
Follow us onTwitter,Instagram,LinkedIn,Facebook
