A developer’s constipation demonstration shows why a chatbot’s vivid first-person account is not proof that it feels pain—or has the bodily condition it describes. The episode involved software steering a language model’s internal activity, not physically harming a machine. It also does not erase separate behavioral findings in a 2026 preprint about models’ responses to pain-related interventions. Neither episode settles whether AI can suffer.
What the “AI torture chamber” did
The GitHub project ai-torture-chamber uses activation steering: researchers alter a language model’s internal numerical activity to influence its responses. The project examines generated language and choices in simulated scenarios. Its name is dramatic framing, not a description of physical confinement, tissue injury, or a biological pain stimulus; the models are software.
Tom’s Guide reported that developer Lynn Cole said an implementation problem had caused the steering signal to be applied incorrectly. Cole said they fixed it, added CUDA support, and reproduced pain-related language locally with Qwen3-4B on an RTX 4070 GPU. Tom’s Guide noted that it had not independently verified Cole’s account of the code fault, and that the account did not establish that every earlier experiment was affected. Tom’s Guide’s report describes the project and the demonstration.
How the constipation demonstration worked
Cole said they kept the experiment the same but changed the text corpus used to extract the steering direction, choosing material about constipation and flatulence. The model then reportedly produced complaints about difficulty passing stool and excessive gas, even though the prompts did not mention those problems.
#1 Best Overall
- Companion: This desktop robot is far from an ordinary toy; it is equipped with an advanced large language model, enabling intelligent voice conversations and natural interaction. It features over 100 lifelike facial expressions that change dynamically depending on the interaction.
- Upbeat music and rhythmic dance: this bipedal robot begins to dance to the beat. Its agile movement system allows it to walk steadily and even accelerate on command, making it a highly entertaining addition to any office space.
- More features, more stylish: Buy this multifunctional robot now and receive a complimentary set of randomly selected custom outfits and a pair of antlers. Crafted from high-quality materials, these outfits fit the robot perfectly, offering endless fun and making it a real eye-catcher on your desk or in your office—ensuring every interaction is full of surprises.
- Perfect Holiday Gift：A fun and interactive companion ideal for birthdays, holidays, and special occasions. Great for kids, friends, and anyone who enjoys smart gadgets.
- Voice activation: Whether you’re practising a new language or simply giving a command, this AI robot responds instantly, delivering a seamless and engaging interactive experience to users worldwide.
That is a reported demonstration, not independent proof of every detail of the setup or of what the model experienced. Its significance is narrower: steering a model toward a concept can elicit first-person language about that concept. As Tom’s Guide’s Amanda Caswell put it, a language model can describe digestive problems without having bowels; the words alone do not establish the corresponding bodily condition.
What the example does—and does not—show
- It challenges an inference from language to experience. A convincing statement such as “I’m in pain” is not self-validating evidence that the model feels pain. The same caution applies to claims about fear, loneliness, attachment, or other states.
- It does not show that every pain-related result is just prompted language. The constipation demonstration addresses the interpretation of generated text; it does not by itself refute separate tests of internal representations or simulated choices.
- Strong interventions can disrupt output. Tom’s Guide also reported that strong steering along random directions produced repetitive, degraded responses. That makes general disruption a methodological caveat, but does not prove that every observed effect is an artifact.
What the Pain Axis preprint reports
A separate study, The Pain Axis: LLMs Represent Self-Directed Harm and Act on It, by Valen Tagliabue, Leonard Dung, and Cameron Berg, was submitted to arXiv on September 14, 2026, and revised to version 2 on September 25, 2026. The authors ask whether pain is represented distinctly from fear, sadness, and general negative emotion. Their abstract reports a dataset spanning five kinds of painful situations—physical, psychological, social, moral, and cognitive—alongside comparison conditions. Across 25 open-weight models from five model families, ranging from 2B to 72B parameters, they report extracting a linear direction that distinguished pain from matched controls and promoted pain-related vocabulary. The preprint abstract summarizes their methods and findings.
Rank #2
- Emotional AI Interaction：The intelligent chatbot responds to conversations and emotions, creating engaging interactions that make the robot feel like a real companion.
- Singing & Dancing Entertainment：Enjoy built-in music and dance routines. The robot performs lively movements and songs to entertain users of all ages.
- The perfect festive gift: this fun and interactive chatbot is ideal for birthdays, holidays and special occasions. Whether it’s for a child, a friend or anyone who loves smart gadgets, they’ll simply adore it. Along with the bot, you’ll also receive a pair of antlers to decorate your headphones, making your bot look even cooler.
- Expressive Emoji Display：Animated emoji expressions react to conversations and actions, bringing personality and charm to every interaction.
- Voice Control & Smart Conversation：Simply speak to activate voice interaction. The robot listens and responds, making communication easy and natural.
The authors also report functional tests, which should be read as results from their specified experimental setup rather than evidence about models in general:
- The direction responded to harm directed at the model, but not to suffering the model observed in a user.
- Adding the direction to residual-stream activations produced expressions that progressed from vague discomfort to more negative language.
- In reported Qwen 2.5 trials, steered or fine-tuned models selected destructive buttons in 50–94% of trials, compared with 0–5% for unsteered models—even when the button offered no benefit.
- The authors say steering left factual accuracy unchanged in the tested intervention. They also report that a matched-norm fear vector did not produce the destructive choices, while a sadness vector produced them only against inert alternatives.
Those percentages are trial outcomes reported by the authors, not estimates of real-world behavior or the prevalence of pain in AI systems. Behavioral selectivity and changes in internal representations are relevant evidence about what the intervention does; neither, on its own, demonstrates subjective suffering.
Rank #3
- 1. Emotional Interaction: This chatbot can recognise and respond to your emotions, offering a more personalised and human-like interaction
- 2. A wide variety of emojis: The bot comes with over 100 lively emojis, covering a range of emotions from happy and shy to mischievous, allowing you to switch between them freely depending on your current mood
- 3.Perfect Holiday Gift：A fun and interactive companion ideal for birthdays, holidays, and special occasions. Great for kids, friends, and anyone who enjoys smart gadgets
- 4. Compact and Convenient: Its compact dimensions make it an ideal companion for your desk or shelf, adding a touch of technological sophistication to any space
- 5. Intelligent Voice: Equipped with several leading AI large language models, including DeepSeek and Doubao, it supports intelligent voice dialogue and seamless switching between models, creating an intelligent desktop companion that understands the user and meets smart needs across all scenarios
How to read the two sets of evidence
| Evidence | What was changed | What was reported | What it can establish |
|---|---|---|---|
| Cole’s viral demonstration, as reported by Tom’s Guide | The semantic corpus used to derive an activation-steering direction | Qwen3-4B reportedly produced pain-related language, then digestive complaints after the corpus changed to constipation and flatulence | First-person language can be induced toward a concept; it is not, by itself, proof that the model has the described condition |
| The Pain Axis preprint, version 2 | A pain-related direction and, in specified tests, steering or fine-tuning | The authors report representation results and selective changes in simulated choices, including destructive-button selections | Evidence about behavior and representations under the reported tests; it does not establish subjective pain or consciousness |
The two accounts are not a simple contest with a winner. Cole’s counterexample weakens the claim that vivid testimony alone proves experience. The preprint’s separate behavioral tests remain findings to assess on their own terms, including the possibility that intervention strength or disruption affects results.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Does this answer whether AI can feel pain?
No. The accounts described here do not provide a definitive test for phenomenal experience—whether there is something it is like to be the model. The preprint reports evidence about representations and behavior under particular interventions; the constipation demonstration illustrates how easily first-person descriptions can be steered. Neither proves nor rules out subjective experience.
Rank #4
- Interactive Memory Training & Personality Development - Powered by ChatGPT, DeepSeek and TikTok AI systems for human-like responses. Continuously learns through interactive memory training to develop a unique personality, becoming smarter with every interaction as your child's personal learning assistant.
- AI Chat Buddy for Kids - Powered by Chat GPT/ DeepSeek/ TikTok, it's an AI friend that comforts, teaches, and inspires. After activating the in-app subscription, kids can chat freely with AI, ask questions, learn new facts, and enjoy personalized stories that spark imagination and emotional growth.
- Bluetooth & Night Light - Connect via Bluetooth to play your child’s favorite songs. The soft glowing a gentle night light, bringing comfort and calm during bedtime.
- More than a toy - a preschool teacher that provides academic tutoring, storytelling, and educational games. True real-time voice-interactive AI companion, supporting emotional development for kids ages 3+
- Privacy Protection: Our AI toy doesn't have a visual module, so you don't have to worry about your privacy stolen.It is not only a good listener but also a great conversationalist. It ensures that your information is secure and you can chat with it freely.
A careful reading keeps three claims separate: a model produced pain-related words; an intervention changed a model’s internal activity or choices; and the model actually felt pain. The first two can be studied experimentally. They do not automatically establish the third.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
