{"id":1564,"date":"2026-06-19T02:06:48","date_gmt":"2026-06-19T00:06:48","guid":{"rendered":"https:\/\/pletzenauer.com\/2026\/06\/19\/karpathy-ki-agenten-jahrzehnt\/"},"modified":"2026-06-19T02:12:14","modified_gmt":"2026-06-19T00:12:14","slug":"karpathy-ai-agents-decade-not-a-year","status":"publish","type":"post","link":"https:\/\/pletzenauer.com\/en\/2026\/06\/19\/karpathy-ai-agents-decade-not-a-year\/","title":{"rendered":"Karpathy on AI Agents: Why It Takes a Decade, Not a Year"},"content":{"rendered":"

Two very different narratives are currently circulating in the AI industry. One promises that autonomous AI agents will replace entire professions within the current year. The other comes from one of the field’s most experienced practitioners and sounds considerably more sober. Andrej Karpathy – co-founder of OpenAI, former head of AI at Tesla – openly pushes back against the hype in his conversation with Dwarkesh Patel: this will not be the year of agents, but the decade of agents<\/strong>.<\/p>\n

For decision-makers in mid-sized companies, this perspective is valuable because it neither downplays nor overstates. Karpathy uses tools like Claude and Codex every day and finds them impressive – while also pinpointing precisely why they often fall short for serious work today. We summarise the key points and put into context what they mean for operational decisions.<\/p>\n

\nKey takeaways<\/strong><\/p>\n
    \n
  • Karpathy expects capable AI agents to take roughly a decade to reach maturity – not a year.<\/li>\n
  • Today’s models lack continuous learning, reliable multimodality and computer use; they are „cognitively patchy“.<\/li>\n
  • For programming, AI tools already work well – but barely at all for genuinely novel code. His sweet spot remains autocomplete, not „vibe coding“.<\/li>\n
  • The realistic deployment model is an „autonomy slider“: AI takes on a growing share, while humans supervise and deliver the decisive rest.<\/li>\n
  • Karpathy sees AI as a continuation of automation – spreading gradually, not through a sudden upheaval.<\/li>\n<\/ul>\n<\/div>\n
    \"Two-column
    Karpathy draws a clear distinction between AI’s current strengths and capabilities that still need years.<\/figcaption><\/figure>\n

    Why a decade and not a year?<\/h2>\n

    Karpathy’s thesis is a deliberate response to the widespread claim that this is „the year of agents“. He considers that a serious overestimate. His benchmark: when would you deploy an AI agent like an employee or an intern? Not today – because the systems simply do not work reliably enough.<\/p>\n

    His reasoning rests on roughly 15 years of experience in the field. The problems are solvable, but stubborn. Concretely, the models fall short in several places:<\/p>\n