TL;DR
The head of Google DeepMind, one of the few people actually building the thing, now reportedly says a machine that thinks at a human level in every field is two to three years away. In January he said five to ten. If he is right, the late 2020s are when people start sharing the world with a second kind of mind that can do science, not just answer questions about it. If he is wrong, the reason will be the three abilities he says today's systems still lack: learning continuously after training, planning over the long term, and coming up with genuinely new ideas. His forecast reaches the public secondhand, the definitions of "AGI" vary wildly from lab to lab, and his parent company is spending around $200 billion this year on the bet.
Below the line, for paid subscribers: The full story: where it stands, how we got here and what's dated next; our Odds board on 4 milestones, starting with "A leading AI lab declares it has built AGI by 2027"; our labeled speculation on what it adds up to; and every source.