I do not understand, how are they "creating negative incentives for sharing knowledge"?
If I posted on SO before in the hope that others find it useful (and not for the karma) - and now it might help others not directly through the site, but with further steps through a llm, where is the problem? Knowledge was shared.
Part of the benefit for the answerer is the experience of interacting with the questioner, receiving upvotes and comments, having answers accepted, and having your name on an answer that's helped people. You get credit for answering questions on Stack Exchange sites. It's not much -- it's not supposed to be -- it's rarely of material consequence -- but it matters. I still get upvotes on some of my old EE.SE answers when my written work helps someone enough for them to give notice. It's a little reminder that I've done something useful in my life.
Having my work ingested into ChatGPT takes the me out of it. It turns me into, essentially, unpaid contract labor for OpenAI. They get all the credit, and I get forgotten. Why would I be okay with that?
If you want to write free code for OpenAI to improve ChatGPT, you're welcome to do so. Cut out the middlemen and send it to them directly. But please leave me and my work out of it.
"Having my work ingested into ChatGPT takes the me out of it. It turns me into, essentially, unpaid contract labor for OpenAI. They get all the credit, and I get forgotten. Why would I be okay with that?"
So you are ok with unpaid contract labor in exchange for virtual points. But if you don't get virtual points as appreciation, no one should benefit. That is ok, but then sharing knowledge is not your main, but secondary goal. Your main goal is the recognition.
But if you delete your comments, you won't get anything at all anymore.
If they remain, real humans will still benefit directly or indirectly. And why should I write exclusicly for openAI? I share my knowledge for anyone. If SO would restrict public access and favour OpenAI - that would be the moment I would want to delete everything. But at the moment LLMs just get also official access, but they had access to SO before, just in a grey legal area. So nothing really changes.
Smcin has answered your other point. Let me respond to this one:
> But if you delete your comments, you won't get anything at all anymore. If they remain, real humans will still benefit directly or indirectly. And why should I write exclusicly for openAI? I share my knowledge for anyone.
The goal -- implicitly for AI companies and explicitly for many of the commenters on this story -- is to replace sites like Stack Exchange. Stack Exchange's traffic will instead go to ChatGPT. The most likely outcome of this is that Stack Exchange will eventually shut down or severely degrade its service. If ChatGPT were a supplemental tool, one user out of many, you would be right. But it's not a complement, it's a competitor, designed to make a profit off of assimilating my work without giving me any compensation or credit.
Exactly. People join SO and other SE websites to ask questions and get answers.
With ChatGPT and similar platforms, trained on SE answers (and open Github repos,...), people will eventually skip Stack Exchange and directly go to ChatGPT.
> But if you don't get virtual points as appreciation [unpaid contract labor]... then sharing knowledge is not your main, but secondary goal. Your main goal is the recognition.
It's false dichotomy to parse out components of motivations; most SO users are motivated by a mix of altruism, sharing knowledge, some recognition, optionally linking to your profile/website/blog/resume/portfolio, getting job approaches and a dose of pride/ego/vanity. As a longtime SO user, that has historically been the bargain, when (most/)all of your submissions were directly seen by human end-users. As a plus, all of that gave you good SEO commensurate with your contributions. So, it's unreasonable to try to dichotomize into "users who mainly did it for the rep" vs ones who want to teach and share.
But the 2023 and 2024 announcements are different: the future is your submissions will be used to train AIs; however SO doesn't seem to have devoted much thought to licensees like OpenAI complying with SO's attribution requirements [0] (attribution must cite individual URL of question/answer, and SO username, which then links onwards via your SO profile page to the items mentioned above). (If the AI synthesizes an answer derived from 5 separate SO items, do they guarantee to attribute all 5 items?) So the human eyeballs are being intermediated, your incentives to participate are evaporating, and that pretty much breaks SO's historical bargain with its user community.
The next major bad development would be SO opening the floodgates on the moderation queue backlog of thousands of items of AI-generated content (which caused the 2023 moderator strike/resignations), much low-quality and arguably should be banned; if/when that feedback loop is closed, the results might well be unholy; certainly bona-fide human contributors will be marginalized and have less incentive. (and if AI were to be used for moderation, then that could be exploitable).
Inbound views/hits on your content on SO either come from a) Google + other search engines b) SO's search itself c) attribution from OpenAI's ChatGPT d) attribution from other(/future) AI licensees. If your code is scraped once but effectively viewed 1 million times from GPT, you won't see those 1 million hits show up; you can only vaguely infer they might be happening if the attribution is actually implemented, and some users click through on it (or by reverse-querying the AI). So c),d) will proportionately increase as a),b) proportionately decrease.
So everything has changed. And obviously the incentive to you to continue to provide unpaid volunteer labor ongoing without even attribution decreases.
If I posted on SO before in the hope that others find it useful (and not for the karma) - and now it might help others not directly through the site, but with further steps through a llm, where is the problem? Knowledge was shared.