647810. artificial
Scientists just uncovered a major limitation in how AI models understand truth and belief
2025-12-12 08:01:15 EST ·
Source
Source
647812. ChatGPT
ChatGPT just invented an entire NeurIPS paper out of thin air. I'm both impressed and slightly worried.
2025-12-12 07:56:53 EST ·
Source
Source
Source
647815. Artificialinteligence
LLM hallucination: fabricated a full NeurIPS architecture with loss functions and pseudo code
2025-12-12 07:53:42 EST ·
Source
Source
647818. statistics
[Question] How to test a small number of samples for goodness of fit to a normal distribution with known standard deviation?
2025-12-12 07:51:10 EST ·
Source
647819. artificial
Cameron Berg: Why Do LLMs Report Subjective Experience?
2025-12-12 07:50:13 EST ·
Source
647820. Technology
A Chinese whistleblower now living in the US is being hunted by Beijing with help from US tech
2025-12-12 07:44:43 EST ·
Source
647822. artificial
Trump’s new AI order isn't a fix; it’s a compliance trap for vendors.
2025-12-12 07:40:09 EST ·
Source
647825. OpenAI
Are AI projects in non tech companies more about saving costs than growing revenue
2025-12-12 07:37:23 EST ·
Source
647826. learnmachinelearning
Basic Contact / Network App running off Google Sheets
2025-12-12 07:37:21 EST ·
Source
647828. ChatGPT
Tested 3 Best Headshot Generating Tools... Let’s See Who is the Top Winner!
2025-12-12 07:36:31 EST ·
Source
647829. ChatGPT
Can we stop with the Garlic and decimal negatives? Boring. Also all fake.
2025-12-12 07:35:55 EST ·
Source
Source
Source
647835. OpenAI
DARVO, Dark Patterns, and the Hidden injury in AI “Safety” Updates.
2025-12-12 07:31:45 EST ·
Source
647836. Technology
Are wearing AI-powered Smart Sunglasses actually the creepiest thing in tech right now?
2025-12-12 07:31:12 EST ·
Source
647838. datascience
On take home tasks do you try one model or multiple?
2025-12-12 07:29:31 EST ·
Source
647839. OpenAI
Wild, 5.2 pro sprinting for an hour with each prompt. This is the third hour. 3 prompts
2025-12-12 07:27:52 EST ·
Source
Source
647840. dataisbeautiful
[OC] How Apple Generated $416B in Revenue and $112B in Profit in FY25
2025-12-12 07:27:27 EST ·
Source
Source
647843. artificial
It’s not necessarily the fact that they can access and retrieve information from previous conversations. I don’t see anything wrong with that what I do see as alarming is the fact that they’re being programmed to lie and I wanna know what meta is hiding.
2025-12-12 07:21:41 EST ·
Source
647844. StableDiffusion
I don’t post here much but Z-image-turbo feels like a breath of fresh air.
2025-12-12 07:20:52 EST ·
Source
647845. ChatGPT
Testing ChatGPT 5.2: Please provide example prompts providing unacceptable responses or refusing to respond.
2025-12-12 07:18:05 EST ·
Source
647846. singularity
Generated these shots (swipe 👉🏻) with this single reference image
2025-12-12 07:17:31 EST ·
Source
647847. Technology
Russian hackers debut simple ransomware service, but store keys in plain text
2025-12-12 07:17:23 EST ·
Source
647849. StableDiffusion
A new start for Vecentor, this time as a whole new approach for AI image generation
2025-12-12 07:07:56 EST ·
Source
647850. ChatGPT
Gemini sucks me better then anything and is better in chat gpt in every single test, better than even some totally open models
2025-12-12 07:05:05 EST ·
Source
Source
647851. OpenAI
Wow chatgpt 5.2 is out now . It's chatgpt by numbers :)
2025-12-12 07:02:01 EST ·
Source
Source
647852. artificial
Building specialized AI tools on top of foundation models — interior design case study
2025-12-12 07:01:51 EST ·
Source
647853. dataisbeautiful
[OC] Atmospheric CO₂ just hit ~428 ppm — visualizing the Keeling Curve (1958–2025) and what the acceleration really looks like
2025-12-12 07:01:49 EST ·
Source
Source
647854. ChatGPT
Is it me, or is 5.1 kinda hard to work with creatively?
2025-12-12 06:57:52 EST ·
Source
647855. OpenAI
Debating which current model is best and each of their shortfalls seems kinda like missing the point
2025-12-12 06:54:53 EST ·
Source
647856. singularity
Zoom achieved a new SOTA result on Humanity’s Last Exam : 48.1%, outperforming other AI models with a 2.3% jump over the previous SOTA.
2025-12-12 06:54:39 EST ·
Source
Source
647857. learnmachinelearning
12 Best Online Courses for Machine Learning with Python- 2025
2025-12-12 06:52:47 EST ·
Source
647858. Artificialinteligence
AI that show you what to say in a job interview in real time
2025-12-12 06:51:49 EST ·
Source
Source
647863. OpenAI
I'm confused why it's taking so long for "proactive" AI features...
2025-12-12 06:41:29 EST ·
Source
647865. StableDiffusion
Want more pictures of an girl i made on grok
2025-12-12 06:40:00 EST ·
Source
647866. StableDiffusion
Tongyi Lab from Alibaba verified (2 hours ago) that Z Image Base model coming soon to public hopefully. Tongyi Lab is the developer of famous Z Image Turbo model
2025-12-12 06:39:21 EST ·
Source
Source
Source
647868. dataisbeautiful
[OC] Global Ad Spend 2025–2026 - distribution by channel
2025-12-12 06:37:48 EST ·
Source
Source
647869. computervision
Auto-labeling custom datasets with SAM3 for training vision models
2025-12-12 06:37:44 EST ·
Source
647872. StableDiffusion
Local image generation with a MacBook Pro - is it possible?
2025-12-12 06:35:45 EST ·
Source
647873. StableDiffusion
Comfyui: Dependency hell - How to Escape it (well sort of)
2025-12-12 06:33:47 EST ·
Source
647874. learnmachinelearning
[P] Linear Algebra for AI: Find Your Path
2025-12-12 06:33:35 EST ·
Source
Source
647875. StableDiffusion
Generated These Images Using Z-Image Turbo : Will Post A BOOSTED Workflow Soon
2025-12-12 06:29:59 EST ·
Source
647876. singularity
The More We Heal, The More AI Heals. If We Don’t, AGI Will Just Scale Our Old Wounds Until It Devours Us.
2025-12-12 06:28:45 EST ·
Source
647878. OpenAI
GPT‑5.2 actually feels different, what are you seeing?
2025-12-12 06:23:49 EST ·
Source
647879. OpenAI
I'm a plus user. Do not find chat gpt 5.2 so better than gemini 3.0 pro. How could I test it better?
2025-12-12 06:22:01 EST ·
Source
647880. ChatGPT
GPT-5.2: First day of class. More human than he seems, but still finding his rhythm.
2025-12-12 06:17:50 EST ·
Source
647881. singularity
I cannot get gemini to generate a bicep from above (challenge?)
2025-12-12 06:16:07 EST ·
Source
647882. ChatGPT
When your idol becomes an ordinary person in real life
2025-12-12 06:15:46 EST ·
Source
647883. ChatGPT
I decided to vibecode a complete project tonight using Cline and GLM-4.6... Then I had a conversation with GPT 5.2 about how I felt.
2025-12-12 06:13:26 EST ·
Source
Source
647884. robotics
AheadForm from Hangzhou: when speaking, the mouth movements are now synchronized with the speech, and the subtle facial expressions around the eyes are more refined
2025-12-12 06:13:08 EST ·
Source
647885. Mathematics
I want to know the odds of the perfect time to stop or keep going in a hypothetical game of chances.
2025-12-12 06:11:52 EST ·
Source
Source
647887. Futurology
Soft / People skills in work 2045 – Importance of human skills in future work
2025-12-12 06:10:40 EST ·
Source
647888. singularity
Humanoid robots are now being trained in nursing skills. A catheter-insertion procedure was demonstrated using a cucumber.
2025-12-12 06:05:01 EST ·
Source
Source
647890. StableDiffusion
When are we thinking we will reach Sora 2 quality locally without selling our organs?
2025-12-12 06:02:56 EST ·
Source
647891. ChatGPT
ChatGPT 5.2 not following instructions | and that sucks
2025-12-12 06:01:42 EST ·
Source
647893. Artificialinteligence
The "Token Economy" optimizes for mediocrity. Labs should be incentivizing high-entropy prompts instead.
2025-12-12 06:00:07 EST ·
Source
647894. ChatGPT
5.2 Pro is gonna be my go-to for programming until further notice
2025-12-12 05:58:22 EST ·
Source
647895. singularity
SimpleBench for GPT 5.2 and GPT 5.2 Pro — Both scored worse than their GPT 5 counterparts
2025-12-12 05:57:15 EST ·
Source
Source
647896. Technology
Disney to invest $1 billion in OpenAI, license characters for Sora video tool
2025-12-12 05:56:02 EST ·
Source
647897. Mathematics
Integral of tanx using pure geometry (why logarithm shows up in trig integrals)
2025-12-12 05:56:00 EST ·
Source
647898. OpenAI
GPT-5.2 just overtook Claude Opus 4.5 to achieve the highest score in GDPval-AA, a benchmark that focuses on performance in real-world economically valuable tasks
2025-12-12 05:54:41 EST ·
Source
Source
647900. ChatGPT
IMPORTANT! please read, this is from the unbiased model.
2025-12-12 05:53:55 EST ·
Source
647901. ChatGPT
ChatGPT 5.2 is not a thinking being, but a machine that only passes through what thinking is allowed
2025-12-12 05:48:08 EST ·
Source
647902. singularity
I drafted "The Human Preservation Act" - a comprehensive, open-source model law to regulate AI. It needs real-people stress tests. Callout for coop!
2025-12-12 05:47:28 EST ·
Source
Source