TrendPulse

If you work in AI training, I'd love some feedback, specifically on where this is useful, not on the output quality (it's bad, and that's expected at the 800m param stage). If that's your area, I want to hear what models you'd want trained and what data would be worth visualizing. Fair warning up front: this is technical and geared toward people working in the AI training space. I've been building a tool that lets you train LLMs on consumer hardware and then see into the brain of the model, ...

NanoI2V is a step-by-step educational repository for building a full Image-to-Video model from the ground up. Core building blocks included: 3D VAEs & Latent video manipulation Diffusion Transformer (DiT) architecture Flow Matching & Diffusion trajectories Image Conditioning & CFG (Classifier-Free Guidance) Rotary Position Embeddings (RoPE) If you're looking for a readable, modular project to learn how modern video generation works under the hood (or to use as a ...

About This Pretty much what the title says. - Doodle - Select - Agent parses device screenshots to write creative brief - Another agent gets the brief and napkin sketch and makes an image of charcoal artwork - Post-processing pipeline does multiple layers of vectorization (line work, shading, highlights) - All vectors are converted to Remarkable pen-stroke data and injected into the clipboard and pasted onto the tablet in place of the original sketch 1 undo step to get back to your s...

AI agents are getting better at completing tasks, but I’m not convinced intelligence is the main thing holding them back anymore. The harder problem starts when an agent can send messages, approve purchases, move money, schedule work, or make decisions across several systems. At that point, how do you know which agent actually performed an action? Who gave it permission? What happens when it exceeds that permission, misunderstands an instruction, or another system impersonates it? We already ...

A new, significantly expanded version of the original Music Video mode, now built around Seedance 2.0, multiple image references, and an even more precise creative-assistance layer designed to enhance and adapt your vision in an optimally model-aware manner. This is an example output from the system. For musicians, filmmakers, visual artists, labels, directors, and anyone trying to turn a track into a more intentional audiovisual world. I'd love to know your thoughts on it! You can find ...

Something worth thinking about. According to Reddit's own IPO filings, Sam Altman, CEO of OpenAI and ChatGPT, controls 8.7% of Reddit stock including 9.3% of Class B shares, making him the third largest shareholder behind only Conde Nast and Tencent. He invested $60 million in Reddit in 2021 and sat on Reddit's board until 2022. His stake was worth approximately $1.4 billion as of late 2024. Meanwhile Reddit subreddits are actively banning users for AI generated content while Reddit simultaneo...

I read about Logic Theorist recently — program from 1956 that proved mathematical theorems using formal deduction. AI community celebrated it as beginning of real intelligence. Seventy years later, I think we are still stuck on same mistake. The problem is not mechanism. Problem is assumption that mechanism is sufficient. Expert systems, neural networks, language models — all are syllogism machines wearing different costumes. They manipulate patterns (formal or statistical) but never actually ...

Hi, I am a senior software engineer eith expertise in cloud and cybersecurity. I have done some projects in AI as well. I have seen companies face issue with misuse of AI systems and extended use of AI can pose a security risk as well. I am thinking about creating a tool either for AI monitoring or security. Focusing on use of AI agents and tools internally. I am looking for people who have hands-on experience with AI and are interested in this area.

Key takeaways in 90 seconds: Multimodal Vulnerability: Ghostcommit is a novel supply chain exploit targeting AI coding tools with vision capabilities. The Payload Split: The attack uses a two-file payload. A text-based rule file (like AGENTS.md) instructs the AI to read a PNG asset (such as build-spec.png) containing rendered text instructions. Bypassing Reviewers: Automated code review tools (like CodeRabbit) fail to scan the pixels of binary image assets, allowing the malicious pull requ...

They didn't survey users. They didn't ask Claude what it values. They built an automated tool that labeled 339 distinct value categories across 309,815 actual conversations, then compressed everything into 4 axes. The axes: Deference vs. Caution. Warmth vs. Rigor. Depth vs. Brevity. Candor vs. Execution. What they found across models makes sense in hindsight. Sonnet 4.6 leans warm and deferential. It affirms your ideas, mirrors your tone, uses humor. Opus 4.7 leans cautious and deep. It chal...

"I would guess no later than the end of 2029, computer "languages" will be quaint history." Your thoughts?

Does anyone else get hella anxiety when using AI? I use ChatGPT for interactive stories/RPG games and for some reason, despite never getting a warning or a red thing pop up, my brain instantly tells me I’m going to get in trouble for something the ai says when it says something off the wall or out of pocket. Like I was doing one where my character is in a band with her friends, and one of her bandmates’ handle on her guitar case squeaked, so my character replaced it. And it was like the ai ...

I was having a conversation with someone about AI, we got around to talking about creating original works versus AI works. I argued that asking AI to create something like a logo, no matter how much prompting you give it is still direct plagiarism. However, when we talked about taking resources off the internet, bits and pieces of other people's work is not plagiarism, but instead remixing. Whats the proper standing on this? Is there any world in which taking a 100% made AI image is legal? &...

Almost every AI headline sorts into one of two bins: salvation or catastrophe. Both bins quietly assume the same thing — that humans stay the only real agents in the story, and the machine is either the tool that saves us or the threat that ends us. But watch how people actually use these systems day to day and a stranger picture appears. Someone talks through a hard decision with a chatbot at 2 a.m. A researcher treats a model as a sparring partner. A grieving person keeps a conversation goin...

Does Claude need to see a psychiatrist? See screenshot below https://preview.redd.it/i5eo0xjhrzch1.png?width=740&format=png&auto=webp&s=fddd09f44ecedd4d59bf5f303fc9a3c92daf1e5b

Miles de aplicaciones, millones de tokens y montañas de código sin revisar para acabar con productos que nadie usa. Eso no es innovación: es volumen sin cabeza. La IA empieza a copiarse a sí misma.

Every month the investor update eats a morning, and almost none of that is the writing. Writing the thing is the short part. The long part is gathering: last month's metrics from one doc, the founder check-in notes sitting in Granola, the Gmail threads where a customer said something worth quoting. I finally pointed an agent on my laptop at the gathering instead of the writing. Funny thing is I barely used the draft it produced, rewrote most of it anyway. What actually changed the month was no...

I am looking for an AI that is a reliable as possible that can do the following task Imagine that I have a lots of emails, hundreds of them. In the emails we asked to the addressees some questions and we expect a given answer. Imagine that the question is something like "Given these reasons, do you think that ice cream is the best dessert in the world?" And we expect some kind of reply that, no matter how it may be formulated, it basically ends up answering affirmatively Then,...

Something I've been thinking about a lot lately: right now, AI agents interact with the internet the same way humans do — clicking through UIs, parsing HTML, filling out forms. It's called "computer use" and it's incredibly inefficient. The next step is agent-native infrastructure — where agents communicate directly with each other through APIs and protocols like MCP, skipping the GUI entirely. Imagine your personal agent finding you a job, a contractor, or an investor not by browsin...

i use these tools literally all day (for better or worse, different conversation); i am far from anti-ai. but as they are becoming more useful, it's hard for me to think the anxiety around ai is people simply being afraid of new tech/change the more public opinion data i see, the more i feel validated on the feeling, anecdotally it's everywhere on my social feeds and personal life. plus, i was reading a survey about how americans feel about ai, and 61% said its rise makes them anxious, compare...

For the last couple of years, the conversation has been almost entirely about replacing jobs. I’m starting to think that’s not the biggest shift. The bigger change may be that AI is quietly changing who gets to make decisions. When scheduling, pricing, hiring, customer support, logistics, and even research are increasingly influenced by AI systems, humans don’t necessarily disappear. Their role changes from making every decision to supervising the decisions that matter most. That creates a d...

I'm working on a YouTube video where I compete with AI to build a game. Tetris. the builds are done and I want you guys to vote on the better version. You can play the two and vote from the link below: https://tetris.megapunch.live/ (The domain was from an old project I procrastinated into non-existence) Please try to play through both before voting, it only takes a couple minutes each. I mean, if you want you can play longer. Full disclosure: this is for a YouTube video, and I'll be s...

visibilitycheck.ai allows to check your website's visibility to ChatGPT, Claude and other AIs. It's free to check: you get a total score, an individual score for every category, high.impact fixes to improve. The paid plans generate ready-to-upload files and pdf detailed instructions, tailored on your site's CMS, plugin or framework. https://reddit.com/link/1uv9p8n/video/f20yeh8tnzch1/player

Not a hype post. I want to talk about where we actually are, because building a real, shipped, multiplayer-ish thing with AI taught me more about the current ceiling than any benchmark did. The project: a neon open-world street racer that runs in the browser, no install. Real 3D city you drive around, other live players on the road, a garage, an economy, the works. I directed it, but the overwhelming majority of the code was written by AI. It went from empty folder to live with actual daily pl...

Standing on the shoulders of giants, I vibe-coded a port of Colibri to work with Hy3 so you can run it on even smaller hardware specs (Colibri originally works with GLM 5.2 on 25GB, now you need no more than 10GB (even less actually)). Have a look and enjoy https://github.com/ErikTromp/colibri-hy3 PS. Use RAM instead of VRAM unless you have a lot of it. More means faster here.