Andrej Karpathy predicts that achieving Artificial General Intelligence (AGI) will take about a decade, citing current AI's limitations in cognitive abilities and continuous learning.

He emphasizes that no existing AI systems possess the necessary lifelong learning capabilities, which are essential for reaching AGI, and solving this will require significant time.

Karpathy remains skeptical about the current state of AI agents, stating they are not yet functional and will need around ten years to overcome key limitations.

He highlights that today's AI models like Claude and Codex lack multimodal and cognitive skills needed for complex, autonomous tasks, and that current AI agents are far from being truly intelligent.

Karpathy points out that AI's inability to perform continuous learning and generalize effectively hampers its potential for autonomous problem-solving.

He notes that improving AI reliability from 99% to 99.9% is as challenging as reaching 90% from zero, indicating significant hurdles remain in making AI dependable.

Critics like Anil Dash warn against overhyping large language models, cautioning that inflated expectations could lead to overinvestment and misguided optimism in AI's capabilities.

Despite skepticism, Karpathy considers his timeline for achieving AGI to be optimistic, viewing ten years as a realistic estimate given current progress.

He advocates shifting the focus from chasing AGI to developing augmented intelligence that enhances human capabilities and scientific discovery over the next decade.

Meanwhile, industry developments include WhatsApp planning to ban general-purpose AI chatbots starting January 2026, restricting AI to customer service applications.

Other tech updates involve Meta securing $27 billion for a new data center, NVIDIA and TSMC producing the first U.S.-made AI wafer, and a decline in Wikipedia traffic due to AI summaries and social media trends.

Overall, Karpathy's outlook emphasizes that true AI autonomy and general intelligence are still a decade away, urging patience and realistic expectations in the industry.