Here is the AGI question that no one asks, but which we must first address.
The phrase of the week has been “AGI has been achieved internally.”
When “Jimmy Apples” tweeted it on September 18th, it was mysterious and derided as crap. When Sam Altman echoed it in a Reddit comment a week later, it was shocking—for good or evil. And when he edited it to show he was only “memeing,” it ended in a great joke.
The most accurate interpretation of the sentence is that it is false, among other things. Internally, AGI has not been accomplished.
But has it?
We truly don’t know, though. Partially due to the lack of a universally accepted definition for the word “AGI” (while true and significant, this justification is dull). In part because Altman’s business has grown so closely that “OpenAI” has become a meme in and of itself. Additionally, Schrodinger’s cat has been replaced and we don’t know how to open the box in part because we lack the tools necessary to determine whether something is or is not an AGI.
However, regardless of its epistemic standing, that statement — repeated for emphasis: “AGI has been achieved internally” — is a potent one.
coming from the right mouth, at least. If I speak it, nobody will listen. Reasonably. Jimmy Apples made the claim in that anonymous account, and while some people gave it no credit because it had previously anticipated the release date of GPT-4 and named Gobi months before The Information reported on it, the majority of people simply didn’t believe him (?) and the word never spread.
But if Altman says it—even in a Reddit comment that was obviously intended to be a joke—the responses come flooding in, ranging from annoyance that the CEO of the top AI lab would tease us with something so serious to worry over the possibility that it might actually happen to the ecstasy of seeing what would ultimately turn out to be the most significant event in human history.
Imagine the ramifications if Altman opted to say that same remark in a more serious setting, without any signals of retraction, like a blog post from the… Altman quickly corrected his comment, perhaps astonished by how quickly it traveled around AI Twitter (and the media).
