Condé Nast Looks to Amazon's 'Out-Of-The-Box' Capabilities

Key Concepts

• On-premise vs. Cloud Computing: The shift from running infrastructure locally to utilizing cloud services.
• Edge Computing: Processing data closer to its source, often for real-time applications.
• Foundation Models: Large, pre-trained AI models that can be adapted for various tasks.
• Amazon Bedrock: A service that provides access to foundation models from various AI companies.
• Generative AI: AI that can create new content, such as text, images, or code.
• Personalization: Tailoring content and experiences to individual users.
• AI Co-worker: The concept of AI augmenting human capabilities in the workforce.
• Natural Language Search: Interacting with systems using everyday language.
• User-Generated Content (UGC): Content created by users of a platform.
• AI-based Moderation: Using AI to review and manage user-generated content.
• Data Infrastructure: The underlying systems and technologies for managing data.
• Databricks: A cloud-based platform for data engineering and machine learning.

Condé Nast's AI Strategy: Leveraging Cloud and In-house Models

This discussion focuses on Condé Nast's evolving approach to Artificial Intelligence (AI), particularly their strategic decision to leverage cloud infrastructure and services, specifically Amazon Web Services (AWS), while continuing to develop proprietary personalization models. The conversation highlights a departure from the trend of building extensive on-premise AI capabilities, emphasizing the benefits of scale and specialized services offered by cloud providers.

Shifting from On-Premise to Cloud Reliance

Condé Nast is moving away from a heavy reliance on on-premise infrastructure for AI, opting instead to utilize AWS. The rationale behind this shift is rooted in their operational needs. As users of AI rather than creators of foundational models at a massive scale, investing in and building out data centers for their operations is deemed economically unfeasible. They acknowledge that while they have trained models in the past, it was at a much smaller scale.

Core Focus: Personalized Content Delivery

The primary use case for Condé Nast's AI efforts is delivering personalized content. While they acknowledge the existence of foundation models from providers like AWS, their current strategy involves leveraging the scale and data platforms of cloud providers.

In-house Personalization Models

A key point of distinction is that Condé Nast builds and trains its own personalization models in-house. This has been an ongoing effort for the past three to four years, predating the widespread adoption of foundation models. They maintain a dedicated data science team responsible for developing these "homegrown" models for most of their personalization and recommendation work.

Evolving Relationship with Amazon's AI Offerings

The conversation explores how Condé Nast might increase its reliance on Amazon's AI technologies. For specific generative AI use cases, they are already utilizing Amazon Bedrock. Examples include:

• Contracts Management and Rights Clearance System: A recently launched system that incorporates Bedrock capabilities.
• AI-based Moderation of User-Generated Content: Leveraging AWS for moderating content generated by users.

This indicates a growing inclination towards using "out-of-the-box" capabilities provided by Amazon, reducing the perceived need to build and train every model from scratch.

Partnership with OpenAI

Condé Nast has a progressing relationship with OpenAI, being one of the first to establish a deal in the early days.

• Internal Enterprise Use: OpenAI's technology is being used "quite widely" internally within the enterprise.
• External Use Case: Bon Appétit Recipe Search: An AI-based recipe search has been launched on the Bon Appétit website, with plans to integrate it into their app. This feature allows users to perform natural language searches and modify recipes according to their preferences. Condé Nast is exploring other external use cases with OpenAI.

AI and Workforce Transformation

The broader theme discussed at AWS is the transition from using AI for internal assistance to an "AI co-worker" or hybrid workforce model. Regarding Condé Nast's past layoffs, it's clarified that while AI tools are used extensively for daily work, leading to increased productivity and potentially eliminating the need for some roles ("we can do more things with fewer people"), the layoffs were not primarily driven by AI impacts across the entire organization.

Alexa Integration

Condé Nast has already integrated with Amazon Alexa. This integration allows users to access content from some of their publications directly through Alexa.

Condé Nast's Position in the Entertainment Space

Condé Nast views itself as operating within the entertainment space, rather than a daily news outlet. Their brands primarily cover leisure, fashion, and lifestyle. This positions them in competition with other entertainment outlets like streaming services (Netflix, Hulu, Amazon Prime) and social media platforms (TikTok, Instagram). To succeed, they recognize the need to excel in personalization and their "great journalism."

Future Technology Goals: Content Recognition and Accessibility

Looking ahead, the primary technological goal for Condé Nast is centered around AI. They have a robust data infrastructure built on Amazon with Databricks. The next critical task is to "recognize all of our content and make it readily available in real time to elements." This is identified as their number one challenge for the year ahead.