Inside AI Dev 25 X NYC

DeepLearning.AI welcomed 1200 attendees (and a total audience of more than 1400) to its developer conference in Manhattan. We packed Convene Brookfield Place to capacity for speakers and representatives from Google, Anthropic, Amazon, Vercel, Groq, ARM, and more. They delivered talks on topics as diverse as agents, context engineering, multimodal models, financial technology, and responsible AI development. We captured it all in real-time on video, in photographs, and via social media. Still, here’s a quick recap highlighting the day’s first events for those who prefer their updates the old-fashioned way.

Andrew Ng kicked off the event with an opening keynote on the current state of AI development and AI’s reception by the broader public. AI is not slowing down, but accelerating, he observed, doubling the length (in human terms) of tasks it can accomplish every 7 months — and for coding tasks, that number is closer to 70 days.

AI acceleration lets developers build production software up to 50 percent faster and prototypes up to ten times faster. Instead of creating a single application, teams gather data on twenty prototypes and see which ones work. Most of these will never go into production, but they make the ones that do stronger and better informed by user feedback. As ARM’s Laurence Moroney said later at the conference, replace “move fast and break things” with “move fast and make things.”

Still, outside Silicon Valley and other high-tech hubs, artificial intelligence has yet to win over hearts and minds, particularly in the United States. Americans want AI to help automate more than 46 percent of their work tasks, especially those they find tedious or repetitive. Yet 52 percent of them find themselves more worried than excited about AI. Andrew asked the attendees to remember these 52 percent and to work hard to win them over.

Andrew then joined Equal AI CEO Miriam Vogel and The Atlantic CEO Nicholas Thompson for a panel on AI governance and literacy. Vogel argued that people are afraid of AI because they don’t understand what they’re using. If the public knew what experts and developers know, she said, they’d better understand the technology’s strengths and limitations. She also noted that while safety observers focus on the large AI and cloud companies, they weren’t focused enough on the applications built on top of them. Amazon scholar Michael Kearns later called this “the AI supply chain.”

Andrew was more sanguine. There is a lot of truth to statements about safety problems in AI, he said, but just as with successes, the hype goes too far. Some companies, he added, use the specter of AI dangers to try to improve their own position in the ecosystem. These organizations hype up red-team insights, like the Anthropic AI agent who was induced to blackmail, until they become media sensations. When Thompson asked if more regulations and safety disclosures could actually lead to more fear, Andrew said he was more worried in the short term about this hype cycle.

As for governments contributing to fear, Andrew said he was most disappointed by anti-immigrant rhetoric. “One of the worst things we’ve done is to reduce the number of H1-B immigrant visas for education and work,” he said. “Us turning away these wonderful immigrants who want to come here is a tragedy. We’re so blessed that so many people want to study and work in America.”

“Every company is now an AI company,” Vogel said. “We will fail if more people don’t turn around their attitudes on AI because they don’t trust the systems. We need to give them a reason to.” The goal of the conference was for developers and the AI giants to make those connections, and to encourage them all to move fast and build responsibly with AI.

Thank you to everyone in attendance, to our hosts, partners, and sponsors, and everyone who contributed to AI Dev 25 X NYC. For more highlights from panels and workshops, see DeepLearning.AI’s streams on YouTube (full videos coming soon), X, Instagram, or LinkedIn. And we hope to see many more of you next year in San Francisco for AI Dev 26!