Could smarter-than-human AI be created soon?

[new text for when this article is moved to the “is this real” subheader in the intro section, now that its role as a node in the old structure is obsolete]

New title emphasizing that this one, unlike the “do people seriously worry

” one, involves timelines? For example: “Might/could smarter-than-human AI be created soon?” / “Could AI surpass/match human intelligence soon?” / “Could AI become at least as smart as humans soon?”

• Summary sentence with link to the article about when do experts think etc. (mentioning that we don’t know, and that experts disagree)

• Maybe we want to very nontechnically (since it’s pretty early in the intro section) mention scaling curves, increasing investment, high current capabilities

• (maybe something like the existing sentence about it having been a goal for a long time)

• Maybe include something about many big milestones falling one after another within the past few years, that many experts thought would take decades, all of which is basically due to scaling up size of models and datasets.

<Steven 2024-05 draft>

Could AI become at least as smart as humans soon?

We don’t know when human-level AI will be created, and experts disagree a lot, but many now think it’s likely in the next decade. [and it seems possible for it to go beyond human level as well]

[human station ill defined] Current systems are not human-level, but are pretty capable at many tasks including mathematics and programming. Multiple large companies explicitly aim to create “artificial general intelligence”, and computing resources expended on training AI models are being scaled up rapidly. Performance has continued to improve as a result of this scaling. [it might be enough (link to is scaling sufficient)]