This reminds me of low-background steel. Certain applications of steel requires very low amounts of radiation. After nuclear testing started releasing radioactive isotopes into the atmosphere, newly forged steel incorporated those radioactive isotopes. The only way to get low-background steel is by finding a pre-atomic age source of scrap steel like a ship wreck.
If the authors dont disclose to publishers they cant know too. For authors they get pressure to write faster because everyone is atleast AI assisted.
TBH i dont care if its AI written but the result of AI writing is sloppy. Lots of fluff and weird wording. Also sloppy research is likely as AI spits out facts and the authos dont have to look into the source anymore. Peer reviewing must become stronger and will be a quality sign for publishers in future.
Publishers are using AI to edit and approve books. But that seems to fall into their acceptance of using AI as tools, just not the actual writing process. So it’s basically a placeholder rule until they figure it out and change it, and they say as much.
I admire your take, but it would amaze me if there are many fortune 500, or even fortune 1000, companies not using generative AI. Especially those focused on white collar work.
I think the real question is how much outsourcing of thought each company, and each department, is pushing on its employees. AI as a proof reader? AI as a ghost writer? Those are very different things.
Using is one thing, as a passive productivity tool. I use spell check and google too. Its the promise of grand tranformation that failed. Agents fail, non deterministic logic is fickle, and prices are insane.
I research this topic as my day job, run an AI shop, and teach ai adoption for fortune 500 and governments btw. Which is why I can speak confidently about the lack of ROI. Im not anti-AI. Im just a realist which makes me sound cynical compared to most prosteletyzing AI hype.
Though, even the ones with strict rules may sometimes have difficulty determining if something was written with AI. Especially if the “author” does some basic proofreading and cleans it up manually before submitting it to the publisher.
What is terrible is it will be hard to tell sometimes, publishers are awash with slop.
Its to the point if you can prove it was written before LLMs existed then the value of your work is suddenly much higher.
This reminds me of low-background steel. Certain applications of steel requires very low amounts of radiation. After nuclear testing started releasing radioactive isotopes into the atmosphere, newly forged steel incorporated those radioactive isotopes. The only way to get low-background steel is by finding a pre-atomic age source of scrap steel like a ship wreck.
That’s so neat and yeah very similar example!
If the authors dont disclose to publishers they cant know too. For authors they get pressure to write faster because everyone is atleast AI assisted.
TBH i dont care if its AI written but the result of AI writing is sloppy. Lots of fluff and weird wording. Also sloppy research is likely as AI spits out facts and the authos dont have to look into the source anymore. Peer reviewing must become stronger and will be a quality sign for publishers in future.
Not everyone is AI assisted for every single project.
Publishers are using AI to edit and approve books. But that seems to fall into their acceptance of using AI as tools, just not the actual writing process. So it’s basically a placeholder rule until they figure it out and change it, and they say as much.
I know several publishers with strict rules about LLM use. Some folks here write as if LLMs are used absolutely everywhere but there are not.
I admire your take, but it would amaze me if there are many fortune 500, or even fortune 1000, companies not using generative AI. Especially those focused on white collar work.
I think the real question is how much outsourcing of thought each company, and each department, is pushing on its employees. AI as a proof reader? AI as a ghost writer? Those are very different things.
Using is one thing, as a passive productivity tool. I use spell check and google too. Its the promise of grand tranformation that failed. Agents fail, non deterministic logic is fickle, and prices are insane.
I research this topic as my day job, run an AI shop, and teach ai adoption for fortune 500 and governments btw. Which is why I can speak confidently about the lack of ROI. Im not anti-AI. Im just a realist which makes me sound cynical compared to most prosteletyzing AI hype.
Though, even the ones with strict rules may sometimes have difficulty determining if something was written with AI. Especially if the “author” does some basic proofreading and cleans it up manually before submitting it to the publisher.
True I ironically see momentary spelling mistakes as a sign of humanity, informally.