I find what's being unsaid here disturbing: the reason removing em-dashes from ChatGPT responses is a "win" that is worthy of an announcement is that em-dashes are primarily used to sniff out if a published work was written by ChatGPT.
What purpose does this serve except to more covertly pass off AI writing as human writing?
If people didn't care that an article etc was AI generated, why would this sort of thing be worthy of a dedicated announcement?
An argument I've seen is that writers who frequently use em-dashes in their work are wrongly being accused of using ChatGPT, and this is a step toward mitigating that.
But this is a smokescreen that functions on the assumption that people are only upset when they can see the cracks and are otherwise indifferent to AI writing if they can't tell the difference.
The truth of this sneakily insidious announcement is that it is a declaration not only that ChatGPT's designers are aware that people do not want to read things written by AI, but also that ChatGPT functions better as a product when it deceptively hides its fingerprints.
Were AI designers fine with laws or industry standards requiring disclosure around the use of AI, this seemingly innocuous fix wouldn't have even crossed Mr. Altman's mind to announce. In fact, if disclosure was a standard, writers would not be accused of using AI in the first place; the simple omission of "written using AI" from their work would be enough.
Small-but-happy win:
If you tell ChatGPT not to use em-dashes in your custom instructions, it finally does what it's supposed to do!