Predictions for when artificial general intelligence will arrive range from "within 5 years" to "never," and the honest answer is that nobody knows. Here's what leading experts and organizations actually say, and why the question is harder than it seems.

Current predictions from major figures:

  • Sam Altman (OpenAI CEO): Has suggested AGI could arrive by 2027-2028, describing it as closer than most people think.
  • Dario Amodei (Anthropic CEO): Has discussed "powerful AI" arriving within 2-3 years that may not be full AGI but would be transformative.
  • Demis Hassabis (DeepMind CEO): Has estimated AGI could arrive within a decade, with significant progress likely by 2030.
  • Yann LeCun (Meta Chief AI Scientist): Argues current approaches (large language models) won't achieve AGI. Believes fundamental new architectures are needed, potentially pushing AGI to 2040 or beyond.
  • Gary Marcus (NYU professor): Skeptical of near-term AGI, arguing LLMs lack the reasoning and world understanding necessary for general intelligence.

Survey data: When hundreds of AI researchers were surveyed, the median estimate for "50% chance of human-level AI" was around 2060, though this has shifted earlier in recent surveys as LLM capabilities have surprised even experts.

Why predictions are so uncertain:

We don't have a definition: There's no agreed-upon test for AGI. Is it passing all human cognitive benchmarks? Is it being able to do any intellectual task a human can do? Is it achieving consciousness? Different definitions lead to vastly different timelines. Some argue narrow but highly capable AI systems already approach AGI for practical purposes.

Scaling laws may or may not hold: Current AI progress has been driven largely by scaling — bigger models trained on more data with more compute. If scaling continues to produce emergent capabilities, AGI could arrive relatively soon. If we hit diminishing returns (which some researchers argue is already happening), fundamental breakthroughs may be needed.

Unknown unknowns: Every major AI advancement came as a surprise to most researchers. Transformers weren't predicted before the 2017 paper. ChatGPT-level capabilities weren't widely expected before they appeared. The next breakthrough could be similarly unpredictable — in either direction.

What we can say with more confidence:

  • AI systems will become significantly more capable over the next 5 years regardless of whether AGI is achieved
  • The economic impact of AI will be transformative well before AGI arrives
  • Each generation of models narrows the gap between narrow and general intelligence
  • The safety and alignment challenges need to be solved before or as AGI arrives, not after

What this means for you:

Don't make decisions based on AGI predictions. Instead, plan for increasingly capable AI systems that can handle progressively more complex tasks. The trajectory matters more than the destination date. Whether AGI arrives in 2030 or 2050, AI will be significantly more capable in 3-5 years than it is today, and that's what matters for business strategy, career planning, and policy decisions.

The most productive stance is to prepare for rapid AI capability growth while acknowledging genuine uncertainty about the upper bounds and timeline.