Are we heading toward AGI faster than most experts predicted, or is this current wave overhyped?

[KeyboardWarrior47]

The pace of AI progress over the past several years has caught even some experienced researchers genuinely off guard, with capabilities in reasoning, coding and multimodal understanding advancing faster than many earlier predictions suggested was likely, leading some prominent figures in the field to argue that artificial general intelligence, a system capable of matching or exceeding human ability across essentially any intellectual task, might arrive considerably sooner than the decades away timeline that used to be the standard assumption.

Others remain far more skeptical, pointing out that current systems, however impressive at specific tasks, still lack genuine reasoning, real world grounding and the kind of flexible general understanding that true intelligence would actually require, and that each wave of AI progress throughout history has tended to be followed by a period of overhyped expectations crashing back down to a more sober reality. Where do you land on this, are we genuinely closer to AGI than most people appreciate, or is the current excitement destined to look overblown in hindsight the same way previous AI booms eventually did?

[ThreadNecro98]

Every previous AI boom throughout history did eventually hit a wall, so I remain cautiously skeptical purely on that historical pattern alone, regardless of how impressive the current systems look.

[StarfieldPilgrim]

I think the goalposts for what counts as AGI keep quietly moving every time a system achieves something previously thought impossible, which makes the whole debate harder to have honestly.

[Liam]

I think we are getting very good at narrow tasks that look impressively general on the surface, without actually having solved the harder underlying problem of genuine flexible reasoning.

[codeberg]

Genuinely think nobody, including the researchers building these systems themselves, actually knows for certain, and anyone speaking with total confidence in either direction is probably overstating how much certainty currently exists.

[Quarry]

Whatever timeline turns out to be correct, I think the safety and governance conversation deserves to happen well ahead of the actual milestone rather than scrambling to catch up afterward

[Sequence19]

The rate of improvement over just the last couple of years alone genuinely surprised me more than I expected, which makes we personally lean toward this happening sooner than the old decades away consensus suggested