Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• AI game-changer for non-coders

• White House makes AI core to efficiency agenda

• Wikipedia says traffic is falling due to AI

• AGI is still a decade away

Let’s get right into it!

Sign Up | Advertise | Tools

STARTUPS

AI game-changer for non-coders

Anthropic just made coding with AI dramatically easier by launching Claude Code for web, a browser-based version of its popular AI coding assistant. Previously, using Claude Code required comfort with command-line tools, but now anyone — even non-technical users — can create and manage AI coding agents directly from a simple web interface or mobile app. This makes it possible to build scripts, automate workflows, and even prototype apps without ever opening a terminal.

For non-coders, it’s a game-changer: instead of writing syntax or debugging manually, you can describe what you want in plain English and let Claude handle the logic. The tool’s agents can work autonomously, turning users into project “managers” rather than programmers. In short, Anthropic is making advanced software creation accessible to anyone who can type a clear idea — effectively turning natural language into working code.

GOVERNMENT

White House makes AI core to efficiency agenda

The Trump administration is making artificial intelligence a central pillar of its government efficiency and innovation strategy. Beyond improving agency operations and reducing workforce burdens, the White House’s FY2027 science and technology priorities highlight AI as a driver of national competitiveness and scientific progress. The memo emphasizes federal investment in advanced AI architectures, data-efficient and high-performance systems, and research into interpretability, controllability, and security.

Applied AI efforts will focus on areas like energy production through nuclear fusion, quantum information science, space analytics, autonomous systems, and embodied AI—including robotics, drones, and self-driving vehicles. The administration also calls for stronger AI evaluation methods and structured scientific datasets to ensure reliability, accuracy, and continued U.S. leadership in AI research and application.

LEGACY TECH

Wikipedia says traffic is falling due to AI

Wikipedia, long regarded as one of the last trustworthy corners of the internet, is seeing human pageviews drop by 8% year-over-year, according to the Wikimedia Foundation’s Marshall Miller. The decline became clear after improved bot detection revealed that much of the earlier traffic spike came from bots evading filters. Miller attributes the drop to changing information habits—people now get answers directly from AI-powered search results and social media platforms like TikTok rather than visiting source sites.

While Wikipedia’s content still underpins much of the knowledge surfaced by generative AI, this shift poses risks: fewer visitors mean fewer volunteer editors and donations. Miller urged AI, search, and social companies to drive users back to Wikipedia and announced new initiatives to improve attribution and reach new readers. He also encouraged the public to support “trusted, human-curated knowledge” by seeking out original sources and recognizing the real people behind the information that fuels today’s AI systems.

THOUGHT LEADERS

AGI is still a decade away

Andrej Karpathy, co-founder of OpenAI and former Tesla AI lead, believes artificial general intelligence (AGI) is still at least a decade away, pushing back against industry hype that suggests it’s imminent. Speaking on Dwarkesh Patel’s podcast, he noted that today’s AI systems lack key human-like abilities such as continual learning and true multimodality, making them incapable of performing complex tasks autonomously. Karpathy admitted he, too, once bought into the “year of AI agents” narrative but now sees the 2020s as the “decade of agents,” requiring steady progress rather than quick breakthroughs. 

While he praised models like Claude and Codex as “extremely impressive,” he said they still fall short of replacing human employees. Despite his caution, Karpathy insists his ten-year outlook is actually optimistic compared to current overhyped expectations. A Stanford-trained researcher, he previously led Tesla’s Autopilot vision team and helped develop GPT-4 before leaving OpenAI in 2024 to launch an educational AI initiative.