User:Dronebogus

User language

en-N	This user has a native understanding of English.

This user is neurodivergent.

This user supports the Wiki Workers United union of WMF staff.

This user is lazy.

Sign the Wiki Workers United solidarity statement!

Enwiki is Mean Girls meets Franz Kafka

Reasons and arguments why generative AI sucks I am sick of repeating over and over

AI content known for bring full of errors and “hallucinations”. Even if accurate it simply doesn’t inspire trust among anyone with critical thinking skills who have seen the utter nonsense it has generated in the past. This damages the reputation of Wikimedia and threatens to put it on the level of the mass-produced, unverified AI slop flooding the Internet.

There is no offense to either the pro-AI nor anti-AI camps in using exclusively human-made content in the vast majority of contexts. However, if AI content is used, the anti-AI camp will be offended not only by the use of AI but the pro-AI camp’s subsequent justification of it; both parties come out unsatisfied and hostile toward each other with no real benefit to show for it. Therefore human content is a win for both parties and AI is a lose for both parties.

Whether AI violates copyright is a serious concern. The more Wikimedia becomes entrenched in using AI, the higher the risk to the project if popular generative AI software is found to be systematically violating copyright.

Even if AI doesn’t violate copyright, using other people’s work for training purposes without permission is by definition plagiarism.

AI has a serious environmental impact due to requiring vast amounts of electricity (mostly from fossil fuels), rare elements, and fresh water and prodicing large quantities of e-waste. While these are common environmental impacts of computing in general, it can be argued that the poor cost-benefit ratio of AI makes the environmental footprint unjustifiable.

AI primarily exists to cheaply replace human labor. This goes against the philosophy of Wikimedia as a non-profit project to which people, not robots, can freely contribute their knowledge and skills.

If I wanted ChatGPT’s advice on a subject, I would ask ChatGPT for advice on a subject. That’s what people miss when they say “generative AI is just a tool like any other”— no, it’s not just a tool if it does 97% of the work. It’s like saying you cooked a meal because you ordered it and specified what you wanted and how it should be prepared.

All generative AI is nothing but a giant plagiarism machine. It isn’t publishable, first and foremost because it’s not your own work (and robots don’t have imaginations so they’re simply regurgitating scraped text without crediting it) and secondly because you are doing your readers a disservice by not putting in any effort to write high quality original material.

Beyond the aesthetics generative AI is an ethical black hole founded on rampant plagiarism that will likely be hit with a tsunami of copyright infringement cases in the coming years. On the other hand people have been using photography for 200 years and drawing and painting since… well, forever, and neither of those have the issues or divisiveness AI has. It’s the difference between using something because it does something that couldn’t be done before and using something that does the same thing as perfectly good existing technologies just because it’s new.

We aren’t talking about fine art here. We are talking about utilitarian art. Photography largely outmoded painting as utilitarian art because it is superior to it in that field in every way. AI is not superior to photography as utilitarian art because people instinctively trust, even though image manipulation is as old as photography itself, that the “camera never lies”. And good-faith contributors on Wikimedia probably aren’t maliciously manipulating images. People associate AI with deepfakes, hoaxes, “hallucinations”, and nonsense misinformation like putting glue on pizza to make the cheese stick. And even good faith AI contributors can’t fully control those things.

There ain’t no such thing as a free lunch. The “cost” of creating and hosting AI generated material isn’t money (which as we know the foundation has infinite amounts of) or even storage space (the foundation similarly is in possession of the Server Farm of Babel) but rather supporting big tech’s campaign to enshittify the entire web by replacing all human content with AI slop plagiarized from human authors without compensation.

AI is not a magical way to recover unsalvageable photos. There is literally no way to generate an accurate photorealistic portrait from, say, a 100x100px image. If you loaded File:George Washington and his Instant Coffee.jpg into the slop-maker de jour it will give you an image, but that doesn’t make the image an accurate representation of George Washington.

I am continually baffled by how AI seems to have made people too lazy to look for free images for even the most common topics. Or has AI spoiled people to the point of them rejecting even 100% free (in every sense) human art because they can now have exactly what they want instead of merely good enough? What is the motivation here? At this point I am absolutely certain I am going to encounter an AI generated picture of the sky, something visible literally anywhere on Earth that we have countless high-quality images of, being used to illustrate an article on the sky.

w:Ethics of artificial intelligence has even more reasons why AI sucks, particularly the section on open knowledge platforms like Wikipedia

User:Dronebogus

Reasons and arguments why generative AI sucks I am sick of repeating over and over

Other stuff

links

Useful

Important

Sandboxes

Other

To do