This poll is NOT about using AI for brainstorming, research or analysis. Its not about who came up with the ideas or the post quality. The question asks who writes the words of the post, who writes the final copy.[1] I think there are good arguments on both sides, with some takes in this thread here.
I'm aware the question has some ambiguity. I imagine the poll to be quantitative[2], so the midpoint might be half the words written by AI. This could be a heavy AI edit where a human writes first, then half the words were changed, or perhaps AI drafts first then half the words are then written by a human. Feel free to interpret the spectrum how you like.
- ^
Its also not asking how AI might be used to write, for example whether you've trained it to copy your voice and writing well. Its a quantitative question - who actually wrote the words?
- ^
This poll may not capture the most important questions about AI writing. Questions about who came up with the ideas and structure might be more important in the long run, especially after AI starts to actually write as well as very good human writers, which I believe it can't yet.
Currently, when I see something that reads as AI written that's a pretty strong signal that the nominal author doesn't fully stand behind the post. I really hate it when I engage deeply with the arguments in a post and write a carefully reply, only to learn that the author wasn't really trying to say that and didn't review the output of their AI carefully enough.
I started off at 100% but, having reflected on this for a few days, I’m now unsure. I stand by the point I alluded to elsewhere on the forum - that language has been used as a class gatekeeping tool since forever, AI levels the field, and this is good - but on the other hand, the slop is out of control.
You can’t get a read of the writer’s personality if AI is used (that might be beneficial in some cases, but I believe it's worse overall); it represents the next level of performative twoddle that I’ve decided I do not like (even if I engage in it myself sometimes), and we don’t need another way to distance ourselves from being seen by one another.
So, @NickLaing changed my mind.
Thanks @SiobhanBall I've definitely learned a bunch too from the other perspectie. Was talking to a French Canadian today and he was telling me how he feels like he can now put a whole bunch of bullet points and ideas down, then AI can draft something that he knows is correct English. After that he modifies it to make sure it is actually making his arguments (because it often adds slightly different arguments) and to add some of his own voice. Makes a lot of sent.
He used to be too nervous because of English being his second language, but AI helped him overcome some of that fear.
Currently, I think AI writing isn't good enough to be better than good human users of the Forum, but I think this will quickly change, and I want to prioritise ideas and impact over who wrote the final words. I expect it will be longer before AIs are at the frontier of doing EA research and cuase-prioritization, so I think posts with only AI ideas will be bad for a longer time to come. But posts with human ideas written up well by an AI I could well imagine being better quality than most Forum writer's posts within a year or two.
I feel differently if someone is writing something to me personally, if someone writes me a poem or a birthday card or something that has sentimental value, then AI writing reduces that. But the Forum I see as primarily content-value rather than sentimental value.
I value content and human connection about equally on the EA forum. Its a "forum" after all, I think human-to-human forum discussion has far more value than just "sentimentality" We are a group of people working together to find the best ways to make the world better and put those plans into action. That needs connection and comradery, with perhaps some sentimentality too. I think if a forum is reduced to only "content value" it will cease to be a forum at all in some ways, and will lose significance.
On the content front, If forum posts with human ideas heavily written by AI were better than or as good human posts, I think your argument would be stronger but right now they are not. I challenge you to find one excellent forum post which is over 50% written by AI. It may be there but I doubt it. From a forum norms perspective, don't think there's a lot of value speculating about the future here, we need to adjust to AI reality as it comes.
Good point re communal values of the forum, seems right.
Ah, maybe I interpreted the original question differently to what you intended. SInce you said it is not about 'post quality' I was trying to put that aside and imagine AI-written posts that are better than human-written posts, and I think in that case I would be happy to read them. But I agree that currently I am turned off by AI writing and far prefer people write themselves in most cases. I suppose I was answering the question more in principle, i.e. if an AI-written post was amazing I would be comfortable with it, but currently they are not. So for me it is more a quality issue than fundamentally and AI-written issue (except for the communal/sentimental aspects, which I agree have value).
Thanks @OscarD🔸 yes the question was ambiguous, your interpretation was entirely reasonable, mine was just more "here and now" than theoretical.
The main benefits I have in mind from allowing an AI to write ~the whole post (over just helping in other ways):
These are all context- and author-specific considerations. I can imagine preferring a post be AI-written than some possible counterfactuals: not posted at all, posted at much larger opportunity cost to the author, posted but worse to read than what an AI could have done. These point me towards permissibility and letting each author decide.
I think the author or another human should generally look over the post before the author posts it. I don't think it's necessary for them to insert their own voice.
thanks @Michael St Jules . I agree with those benefits but there's no mention here of potential costs? Maybe you don't think those are significant?
Its a completely different question but are you happy to receive 100% AI written grant application as well? And would you be happy on your grantmaker end to allow your own AI to review that application or would you insist on reading the whole thing yourself?
Just trying to prod a bit and see how far it goes...
I'd prefer human-written applications, because it can be hard to distinguish ~100% AI-written but primarily using the applicants' own ideas and reasoning from ~100% AI-generated, including writing, ideas and reasoning.[1] Grants are bets on the grantees' abilities, not just the project idea. However, I tend to also talk to applicants over calls or in person, and see their work in other ways.
I can imagine for a project for which communication by the applicant is an important part of the project's path to impact, if the application looks AI-written, I would ask them to resubmit or I would reject them, if and because the people the applicant would be communicating to dislike AI writing. This hasn't come up yet, though.
At this point, I'd insist on at least personally reading parts that are enough to be decisive one way or the other.
Of course, this leaves another possibility (and others in between the different possibilities outlined so far, including no AI use): 100% of the ideas and reasoning come from AI, but the application is 100% written by the applicant. Hopefully by writing it themself, they've taken the time to understand what they're submitting, but it would still be better if the ideas came from the applicant.
If we're assuming the post would be good quality, then I don't expect the costs (to me) to be significant, but I'm open to reasons otherwise. If the posts are sometimes low quality or repetitive, then AI could enable more of them, and that would be bad. I'd lean towards allowing 100% AI written posts and seeing what happens to the EA Forum, i.e. tracking the results and reassessing.
Maybe the voting system, minimum karma to post, and throttling based on recent net negative karma posts/comments are enough to handle this without negatively affecting engagement. Banning 100% AI-written posts is a blunt tool, and it seems worth trying other things.
At least until people accept that "oh the AI wrote that bit, I don't exactly mean that" is unacceptable. If you post it, you should stand behind every word.
Nuance, I’d be happy for an AI to write a draft, but (at this time) I will never publish something without a thorough review and strong work to put it in my own voice. I will never let a single AI-written word go unreviewed. (This is the same for ghostwritten posts made on my behalf, I don’t think AI changes much here).
Our authorial voice and trust that people put in our words is one of the few things we really have left that makes us human. When I catch it, I find reading AI-written (or ghostwritten) content gross and disturbing, because it signals to me that the author has no respect for my time, or their own humanity. I know that’s an extreme position but I find it hard to take any other one.
Yep that's similar to where I am at too, although I'm a bit less extreme. Much if it for me is about respect for their and my humanity although in still confused and haven't figured out the nuance.
This particular question is about how many words of the final product are written by a human vs. AI. I don't think you've voted though?
Well, like, I don’t care who wrote the words but I do care about who took ownership of them. If an I happened to write in my style/voice and I reviewed it and posted it, would you consider that my writing or the AI’s?
By this poll' definition the AI wrote the whole thing. This poll isn't about style or substance, it is about who actually wrote the words. I think though (disturbingly) most people seem happy to take ownership of AI written words, I don't think that's a big issue.
This question though is asked in the context of AI right now and I don't think AI can actually write nearly as good as a decent writer, even when prompted and guided to write in their voice. So I think your question might be theoretical at this point in some way?
Also you said "Our authorial voice and trust that people put in our words is one of the few things we really have left that makes us human." If we didn't write the words, is it actually our human voice? Even if if it sounds a lot like us?
I will never let AI write a single sentence! I resent reading AI-generated writing passed off as written by a human, and I would never inflict this upon my readers.
I have found that the most common explanation for why people using AI for writing is a lack of self-confidence. I keep encouraging people to write in their own words and use their own voice because all the flaws of unpolished human writing are vastly preferable to chatbot writing.
I don't really see why anyone would use AI other than maybe translate phrases if one does not know English (or the language they are writing) well.
Language models add unnecessary fluff to the text that takes space away from actual content. I'd ask the following question: is the text written by the AI longer than the prompt? If the answer is yes, then please do not use that AI text. All extra text written by the model is fluff and meaningless.
Language models are also really bad at writing. They tend to overuse stylistic devices to the point that the text is heavy and difficult to read. They split the text into unnatural chapters, tend to add more subheadings than necessary, overuse tables and lists, and in general write horribly. Some people claim that AI helps people who are bad at writing to be able to take part in discussions, and that argument really makes no sense since AI cannot write either and I'd rather read bad English than fluff. The only exception to this is using AI for translation, since that is obviously a real blocker.
Lastly, there is the question of effort. To put an effort into writing is a proof that someone cares about this enough to write it down, so I should also care about reading it. If I recognize that a text is AI written, my motivation to go through it drops immediately. In a world of constant spam and sensory – and literary – overload, my time is valuable, and effort is the currency that gets me to read your text.
I would definitely want a human reviewing and possibly iterating, but if that is happening and the AI is drafting, that's fine.
I'm comfortable with something around 30% which is aroundabout where I feel people's distinctive voice begins to fade
I do understand the arguments of the "100%ers" AI can reduce the friction and time to post, can help English second language speakers and can help people express themselves better.
But with AI in between the writer and the reader we lose part of the soul and even beauty in communication. I want to talk with you. I want to hear your voice. Are those your words? Did they come from directly from your mind? Ideas and structure is important, but the words themselves matter at least as much to me - an expression of ourselves and the people we are. What are we without our own words?
If we're happy with discourse without our own words, where does that leave us? Someone's AI bot writes a post, my AI bot replies, then maybe we read it later? This kind communication is less real, and quickly fades into irrelevance.
I'm happy to pay the time and friction costs to keep our communication pretty close to human-to-human. Its already hard enough to have good discourse on the internet without speaking face-to-face.
I agree 100% that there should be places where AI written discourse is welcomed, personally I would rather that wasn't this forum.
60%➔ 50% agree80% roughly, AI can write every word. I use voice-to-text and Claude daily and always note the methodology. Two things matter: Ownership (you clicked publish, you own every word, no "the AI wrote that" excuses) and Transparency (state how it was created, let the reader decide). If both are met, it's just an efficiency tool. The ideas and positions are yours, the words are the delivery mechanism.
(This comment was drafted via voice-to-text and Claude ^^)
Thanks for being transparent about the use ;). I've said this a few times but discussion isn't just about the ideas and position, its also about your voice and your connection with the person. Words are far more than a delivery mechanism.
I don't feel much humanity or connection after this message, I feel like its broken that connection we would have had, which makes me a little sad.
It's not just an efficiency tool. This is what an AI company might tell you. There are negative tradeoffs here, and if you're happy to lose your individual voice then well, I think your argument might hold but there is a cost to having AI write all the words.
I might be out of line here, but I don't think you necessarily actually believe the full position of your AI assisted comment. Do you really agree that its "just an efficiency tool" and that words are just "a delivery mechanism"? Has AI (even slightly) warped what you actually believe about this?
I do agree that Ownership and transparency are important, and for me should be more of an absolute basleline.
Thanks, great points I find them all convincing. The key for me would be: Is connection intended or important for this post? And if yes, then it gets lost through AI. But if it's mainly about organizing or communicating an idea then declared AI writing is fine in my intuition.
Good catch - I'd say for me it's mainly an efficiency tool, but agree that there are more aspects and indeed there are tradeoffs. I was wrong and influenced by Claude!
General about human connection: I rarely do this via text mostly via audio and in person events. Using AI there would be extremely bad and would lose a lot of connection.
Would I change my position?: Yes a bit, I think it's pretty crucial to decide: "Is it an idea post or a human connection post?" and depending on that decide whether to use AI for writing. Not sure how to update the percentage though, maybe 70 % depending on the content, I'd most posts in the forum are idea posts and only some motivational personal stories etc.
All written by me this time, Cheers ;)
One day the AIs can be much better at writing but that day is not today
That is true.
Even when they do become better than us at writing, I might be keen to keep discourse spaces separate. some spaces where humans can talk just with humans and others where it's everyone together. Obviously the AIs will be talking to each other on a scale hard to fathom.
I think if we get too used to mixing it might be difficult to separate down the line too.
This is absurdly speculative though.
AI sucks at writing and doesn’t get my style at all. If it did, I’d let it write the whole thing. (Yudkowsky predicts we have about 2 more years until AI learns to write.)
The only parts I'd let AI write are those where I'm unsure if my phrasing is really natural in English and I ask for a better one.
If I'm not the one writing the post, I don't see why anyone should bother reading it.
I have a particular writing style that I consider my "voice", and I fundamentally take pride in my writing skill and see writing as a craft and art form, so I refuse to use AI to write a single word of what I would publish to the world.
To me, using AI for writing is equivalent to having someone else write it for you.
I will take a much stronger stand and also say that I'm very sympathetic to the argument by creatives that using generative AI is inherently unethical for a wide variety of reasons, ranging from the nature of how the data is collected without permission, to the environmental and societal impacts, to the way it degrades our culture through slop.
I'll also throw in from a more doomer perspective that by using these models, we are supporting the AI industry, giving them data and (if you are subscribing) money to rush toward building ever more dangerous systems (even if LLMs don't pan out, the sheer amount of research effort now going into AI makes things likely to reach critical mass) that can already cause things like AI psychosis, are nearing the point of mass replacement of labour with capital, and could one day kill us all.
I do not subscribe to any chatbot services, and have only experimented with free versions to a extent, and have never used them for my coding or writing.
I personally, am seriously considering joining PauseAI and possibly, additionally, boycotting AI products. This from someone who used to be an ML researcher before it was cool.
I'm very much aggrieved to see my life's work used for such tremendous evil as it is now.
I think EA should take a much stronger stand against AI. The public backlash is already starting, and for once, we should pick a side. If the most recent METR trendlines are right, we don't have much time left.
I largely agree - I think its similar in many ways to having a ghostwriter
Good for brainstorming first drafts, but I'm not comfortable having it write everything, even if the content is right and coherent.
I think what matters is the content. So I am in principle comfortable with AI writing all of the words of posts. However, in practice, AI has written basically no words of my posts.
There's some ambiguity as to whether this is a personal preference question (I.e., I would never post something drafted by an AI, but don't have a problem if you do) or a normative question (I. E., I would never post something drafted by an AI and neither should you.
Yeah the question isn't perfect. It's a personal question which could apply to you or others. It's about what you are comfortable with others and yourself doing in the forum here.
The ambiguity in this regard may give the impression that there is more hostility toward people using AI to draft things than there actually is.
Seconding what @huw wrote: I’d be happy for an AI to write a draft, but (at this time) I will never publish something without a thorough review and strong work to put it in my own voice. I will never let a single AI-written word go unreviewed.
Also I usually also ask humans to review my drafts even after I have a first/final pass, and the ideas that I input into the prompt are mine, not just something I asked AI to create. Also, my AI tools know my voice quite well at this point and I'm constantly tweaking the instructions.
I'm pretty comfortable with AI taking rough ideas / bullet points and turning them into paragraphs, from which the author can then edit and refine.
I consider myself a good writer, in the sense that I can express my thoughts clearly, but I find that's different from being able write them down if I have a mental block.
Some people are proficient at dashing down their thoughts without resistance to getting started. I'm not one of those people and I'm sure there are many like me.
Some posts are meant for literary beauty and some are meant for ideas. Writing ideas well takes effort, and writing ideas with AI takes basically zero effort. If your post is mainly about ideas, and the fact that it is written by an AI doesn't make it annoying to read, I really just don't see why you wouldn't write basically entirely with AI.
As a reader, each whiff of LLM usage is a red flag. If the writer used it, it's highly probable that I should use another LLM to summarize it.
And if the final copy used it, surely previous stages used it even more, with all that implies. (I assume it's not the moment to go into that)
As a writer, it's symmetric. Why should I expect readers to spend effort that I didn't? Even if they do, how can I be sure that there wasn't a drift from my original intended meaning?