An idea can arrive before its tools

Another idea may be promising while the tools around it are still limited.

Researchers explored machines that learned from examples during the same decades as rule-based AI. Early versions could learn only simple patterns. Computers were slower, useful examples were harder to collect, and important training methods were still being worked out.

Later researchers had better methods, faster computers, and far more examples to work with. That changed what they could test.

History does not sort every idea into "right" or "wrong." A better question is what failed, why it failed, and whether the missing piece could change.