Top all-time · #582
Janelle Shane
Personal blog · aiweirdness.comaiculture
Janelle Shane writes AI Weirdness, a personal blog at aiweirdness.com about ai and culture. Since 2017, 103 different Hacker News readers have linked it 137 times, most often “Don't use AI detectors for anything important”. It is still writing: the last post was on 2026-10-30, and you can follow it by feed.
Sometimes recommended · 7 of 34 sampled comments recommend it?
Why HN links it
HN readers link Janelle Shane's AI Weirdness blog for its funny AI-generated results, such as GPT-2 novel openings, cereal names, paint colours and recipes. Commenters also cite it on AI failures like the invisible sheep.
- Recommended for · 7 comments
- Commenters call the blog delightful, funny and excellent, and one says it is good comedy. Others recommend it as worth looking at, or describe the posts as amus
Summarised by Claude from a sample of 34 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
A machine learning humor blog by Janelle Shane. Odd ice cream flavors, confusing pickup lines, cursed candy hearts, and the occasional phantom giraffe.
Most-linked posts?
- Don't use AI detectors for anything important4 people · 4 mentions · first linked 2023-07-03
- Letting neural networks be weird4 people · 4 mentions · first linked 2018-04-15
- Once again, a neural net tries to name cats4 people · 4 mentions · first linked 2019-06-03
- SkyKnit: When knitters teamed up with a neural network4 people · 4 mentions · first linked 2018-04-21
- New Breakfast Cereals from AI3 people · 3 mentions · first linked 2022-01-23
- AI Generated Valentine's Day Cards3 people · 3 mentions · first linked 2022-02-06
- Do neural networks dream of electric sheep?3 people · 3 mentions · first linked 2018-03-02
- AI Weirdness Blog3 people · 3 mentions · first linked 2020-04-13
- AI weirdness. The strange side of machine learning3 people · 3 mentions · first linked 2023-05-30
- AI Weirdness3 people · 3 mentions · first linked 2019-08-05
What Hacker News says?
Related: the technology behind this has a harder time with cats than it does people, and the results are hilarious: http://aiweirdness.com/post/182633984547/gancats
Comment linking /post/182633984547/gancats
on “This person does not exist”Submitted This AI is bad at drawing but will try anyways as a story · 313 points
HN discussionThis is a couple of years old now, but at one point Janelle Shane found that the only reliable way to avoid being flagged as AI was to use AI with a certain style prompt https://www.aiweirdness.com/dont-use-ai-detectors-for-anythi...
Comment linking /dont-use-ai-detectors-for-anything-important/
on “I'm Kenyan. I don't write like ChatGPT, ChatGPT writes like me”Submitted Minecraft with object impermanence as a story · 229 points
HN discussion#EBABEC is now "Pink Hork" in honor of Janelle Shane[0]. 0: https://aiweirdness.com
Comment linking /
on “Colornames.org – A collaborative effort to name all 16.7M colors”Submitted GPT-3 tries pickup lines as a story · 212 points
HN discussionThis reminds me of Janelle C. Shane's work on AI[1]. Computer vision's been learning based on the photos that people take, not the visions that people see, and this is giving AI a preference for photogenics instead of reality. People generally don't take pictures of rolling green hillsides. But they very often take pictures of rolling green hillsides with sheep on them . So if you ask the robot to draw a picture of…
Comment linking /
on “There is no evil like reCAPTCHA”Submitted How to Begin a Novel as a story · 162 points
HN discussionThe evolution of the AI weirdness blog ( https://www.aiweirdness.com/ ) over the past few years somewhat supports this idea, except that the author got plenty of funny results with LLMs - but mostly the early ones, up to around GPT3 (and smaller variants of GPT3). For example, cereal names generated by the Ada version of GPT were much funnier than the Da Vinci ones: https://www.aiweirdness.com/new-breakfast-cereals-…
Comment linking /new-breakfast-cereals-from-ai/
on “Markov chains are funnier than LLMs”Submitted Letting neural networks be weird as a story · 159 points
HN discussionIt seems that if you use Microsoft Azure for image recognition you would come to the conclusion that there are giraffe's everywhere: http://aiweirdness.com/post/171451900302/do-neural-nets-drea... Maybe it's because Microsoft worked on a project to help conservationists protect giraffes: https://blogs.technet.microsoft.com/machinelearning/2016/04/... This might be an example of how a specialized dataset that was use…
Comment linking /post/171451900302/do-neural-nets-dream-of-electric-sheep
on “Giraffes Added to the Endangered Species List (2016)”Submitted Dungeon crawling or lucid dreaming? as a story · 143 points
HN discussionAnd as a less dataset-quirk-specific example of AI fixating on backgrounds, here's Microsoft Azure dreaming of electric sheep: https://aiweirdness.com/post/171451900302/do-neural-nets-dre...
Comment linking /post/171451900302/do-neural-nets-dream-of-electric-sheep
on “Tesla Model 3 Drives into Overturned Truck in Apparent Autopilot Failure”Submitted Do neural networks dream of electric sheep? as a story · 139 points
HN discussionThe problem there isn't temporal consistency (although I agree that often sucks), it is over-reliance on context. The invisible sheep problem: http://aiweirdness.com/post/171451900302/do-neural-nets-drea...
Comment linking /post/171451900302/do-neural-nets-dream-of-electric-sheep
on “YOLO: Real-Time Object Detection”Submitted Play AI Dungeon 2. Become a Dragon. Eat the Moon as a story · 137 points
HN discussionJanelle Shane: https://wandering.shop/@janellecshane She runs the http://aiweirdness.com blog, which is delightful
Comment linking /
on “Ask HN: Who are your favourite people to follow on Mastodon?”Submitted AIs Named by AIs as a story · 133 points
HN discussionWhy the fuck does this link to the amp version of the site? Change http://aiweirdness.com/post/177091486527/this-ai-is-bad-at-d... to http://aiweirdness.com/post/177091486527/this-ai-is-bad-at-d... and it actually loads and looks proper.
Comment linking /post/177091486527/this-ai-is-bad-at-drawing-but-will-try-anyways
on “This AI is bad at drawing but will try anyways”Submitted Imaginary worlds dreamed by BigGAN as a story · 129 points
HN discussion> Sometimes after a night’s sleep, we wake up with an insight on a topic or a solution to a problem we encountered the day before. The current crop of models do not "sleep" in any way. The associated limitations on long term task adaptation are obvious barriers to their general utility. > When conversing with LLMs, I never get the feeling that they have a solid grasp on the conversation. When you dig into topics, th…
Comment linking /
on “The Case That A.I. Is Thinking”Submitted SkyKnit: When knitters teamed up with a neural network as a story · 84 points
HN discussionOne of the things I'm rather worried about here is AI's ability to learn on the most unlikely (to human cognition) features. Take the image recognizer that finds "sheep" by seeing pastures where it thinks sheep might graze http://aiweirdness.com/post/172894792687/when-algorithms-sur... , or take any of the adversarial perturbing examples like https://blog.openai.com/robust-adversarial-inputs/ . You're not allowed to…
Comment linking /post/172894792687/when-algorithms-surprise-us
on “Text Embedding Models Contain Bias”Submitted Three bad recipes generated by neural network (2017) as a story · 60 points
HN discussionI'm not so sure. Even prosaic, current AI, can be very unpredictable. See this blog post http://aiweirdness.com/post/172894792687/when-algorithms-sur... , which is a somewhat more causal summary of this paper "The Surprising Creativity of Digital Evolution: A Collection of Anecdotes from the Evolutionary Computation and Artificial Life Research Communities" https://arxiv.org/abs/1803.03453v2 The authors of the paper…
Comment linking /post/172894792687/when-algorithms-surprise-us/amp
on “Can “effective altruism” maximise the bang for each charitable buck?”Submitted This neural net would like to deliver these petitions as a story · 54 points
HN discussionI have the most bizarre sensation when reading through this. On one level, they're just boring brand names. But knowing that they were all created by Amazon for Purposes and their names chosen presumedly by marketing folks... Well, it sorta makes this read like the most genius piece of satire about today's retail marketplace. It would fit somewhere between McSweeny's lists[0] and AI-written paint color names[1] [0]…
Comment linking /post/160776374467/new-paint-colors-invented-by-neural-network
on “A list of Amazon’s businesses and subsidiaries”Submitted AI vs. a giraffe with no spots (2023) as a story · 42 points
HN discussionOh man, good question. I'm always up for swapping these stories. A lot of these came from a paper on weird AI tricks, and resulting best-of list on a blog collecting these stories.[1][2] Suffice to say, the people who think the orthogonality thesis is a weird hypothetical aren't keeping up with the state of things. - The aforementioned Tetris story: an undirected learner was set to maximize score at Tetris learned n…
Comment linking /post/172894792687/when-algorithms-surprise-us
on “Notes on AI Bias”Submitted Tench: When data is messy as a story · 40 points
HN discussion
