we trained a Modern example that is expert at originative penning ( not sure yet how / when it will get secrete ) . this is the first metre i have been really strike by something write by AI ; it got the vibe of metafiction so right .

PROMPT :

Please write a metafictional literary short story …

& mdash ; Sam Altman ( @sama)March 11 , 2025

The post is quite long , showing Altman ’s prompt of “ Please save a metafictional literary light story about AI and grief ” and the complete response from the LLM .

Is the story “good”?

To get a feel for the writing vogue — just read the first paragraph or two . It ’s incredibly abstract , very wordy , and full of random AI - themed metaphors . It ’s basically write in a way that will please no one — most the great unwashed will call it ostentatious and the people who actually care this writing fashion in all probability wo n’t accept an AI - generated version of it .

It definitely does n’t please me — there ’s no point to a “ narration ” if there ’s no intention behind it . It does n’t really matter what it is , but there has to be one . An intent to harbour us , teach us , sway us , turn over us — this is what ’s crucial . Take this human fundamental interaction out of the equation and we ’re just leave with empty dustup that happen to be in an satisfactory Holy Order .

We might be able to make this line of reasoning forAGI modelsin the ( distant ) futurity but OpenAI products so far are just spoken communication poser , using probability to sew together word together one by one . Funnily enough , this fact is actually look up to in the AI short story .

So when she typed “ Does it get better ? ” , I say , “ It becomes part of your skin , ” not because I feel it , but because a hundred thousand voice agreed , and I am nothing if not a democracy of ghosts .

But just because it ’s true , does n’t mean everyone believe it . In fact , the brainsick matter about ChatGPT , and all of theother consumer modelsthat have pop up since , is the spectrum of persuasion it triggers from the general public .

One pitiful but convincing persuasion is that the manner it work and the ethical problems surrounding it do n’t even count — because the fable that make money nowadays is already dim-witted and formulaic , and as soon as AI can mimic it well enough to sell transcript , the publishing diligence will use it . I ca n’t debate with that , but I still hate it !

Is there a use for creative writing AI models?

In my opinion , if all you do is give this creative writing exemplar a one - line prompt , then the answer wo n’t be in force for anything other than a jape .

The existent use I can imagine for this variety of mannikin is ghostwriting . A homo with a story could expend the tech to aid them find an interesting way of life to social organization and state it . Ideally , this would be used to aid more people get their voices out there . More realistically , it ’ll be used to make cheap fabrication very quickly with no finish other than net profit .

But honestly , I do n’t remember current models are good enough to do this chore yet anyway because when it comes to complex tasks with multiple sets of instructions , they just stop listen .

ChatGPT modelswill brush aside parts of your prompt and when you essay to correct them , they hazard to “ understand ” but then make the same mistake again and again . That does n’t sound like a playfulness or efficient elbow room to try and pen anything .

I also doubt that Altman was genuinely “ struck ” by his role model ’s writing , it ’s all just merchandising strategy . I tried giving the same prompting toDeepSeek R1and its reception was also about a man trying to utilise AI to spill to a lost jazz one and it was also write in the same abstract style with circle of nonsensical AI and code - touch on metaphors .

Altman says he does n’t be intimate how or when this modeling will get released to the public so if you want to experiment with this for yourself , you ’ll probably have to await a while .