2 minute read

You might be wondering why Mark Zuckerberg and Meta have recently spent hundreds of millions hiring the very best AI talent in the industry. The reason? He wants Meta to own—and power—the top large language model (LLM) so that it becomes the default tool people use, and in turn, Meta can capture uniquely rich data about human behavior.

Why does this data matter so much? Because access to every prompt—what people actually ask LLMs—offers an incredibly direct window into the way people think, work, and worry about the world. It’s deeper and more authentic than anything that’s ever been collected on a traditional social network.

Two recent studies cast light on how people use AI today and why this matters: 🟢 OpenAI’s analysis of ChatGPT usage shows that 28% of all conversations are writing-related, surging to 52% among professionals in business and management. A significant slice is about editing, critiquing, and real problem-solving, not just content generation. There’s a sharp rise in people using LLMs for decision support and as “consultants” at work, while writing-for-fun and purely factual queries are declining. 🟢 Anthropic’s research on its model, Claude, shows that software development is the single largest category of use, accounting for 15-25% of conversations. This includes web and mobile app development, AI and machine learning applications, and data analysis. Beyond professional tasks, a small but meaningful percentage of conversations are “affective,” with users turning to Claude for emotional support, advice, and companionship.

Taken together, these insights reveal that LLMs are evolving into active collaborators and knowledge partners, not just passive tools.

This brings us to the future of AI: agents. Whoever can access and understand this vast amount of prompt data could build incredibly powerful and sophisticated agents. In fact, the latest software agents have been trained on prompt data from human software engineers, and the results are truly impressive. Imagine agents that can anticipate our needs, automate complex workflows, and even help us with creative endeavors, all based on a deep understanding of human intent and problem-solving gleaned from our interactions with AI. The race for the best LLM isn’t just about answering questions; it’s about understanding humanity to build the next generation of intelligent systems.

🔗 Read the full post on LinkedIn

#AI #LLM #PromptData #Meta #OpenAI #Anthropic #SoftwareAgents

• 🔗 ES2030 Agentic AI Masterclass: https://lnkd.in/eCgy_jFX • 🔗 My new book, “Autonomous Minds”: https://lnkd.in/eJfZ3PWt • 🔗 Book1: https://bit.ly/4b31PEG • 🔗 Alignment paper: https://lnkd.in/ez8cV-jK

🔥 Like if you found this helpful ➕ Follow for more applied AI insights ♻️ Share to help your network grow

You May Also Enjoy