Most predictions about transformative technology tend to get the direction of change right, while badly misjudging the speed at which it actually arrives.
Historically, new technologies are usually overestimated in the short term and underestimated in the long term. Early hype cycles set expectations that don't materialize on schedule, which produces a wave of disappointment — followed by changes that eventually happen more thoroughly than the original hype suggested, just later.
Why timeline errors happen so consistently
Predicting a new capability existing is far easier than predicting how long integration into existing systems, regulations, and human behavior will take. The technology curve and the adoption curve are not the same curve, and most predictions conflate them.
A more useful question than "when"
Rather than trying to predict an exact timeline — which even experts routinely get wrong — it's more useful to ask which current capabilities are already reliable enough to use today, and build familiarity with those, rather than waiting for a fully mature version that may take longer than expected to arrive.
This week's exercise
Identify one AI capability you've been "waiting to get better" before trying. Test the current version this week on a low-stakes task, rather than continuing to wait for a hypothetical future timeline.
Note: this article discusses general projections about technology adoption, not guaranteed outcomes.