text
string | label
int64 |
|---|---|
Unemployment could hit 25 among recent grads and trigger 'unprecedented' social disruption thanks to AI, U.S. senator warns
| 0
|
Joining Valve's Gabe Newell at the altar of AI, Ubisoft CEO Yves Guillemot says the controversial tech will be "as big a revolution for our industry as the shift to 3D" Ubisoft is using generative AI "in all our studios and offices"
| 1
|
Top Economist Warns That AI Data Center Investments Are "Digital Lettuce" That's Already Starting to Wilt
| 0
|
Insurers retreat from AI cover as risk of multibillion-dollar claims mounts
| 0
|
Anyone here using AI as a coding partner? I tried building a small Python project recently with AI help, and it made the whole thing way less intimidating. Now Im trying to figure out which AI coding assistant is actually worth sticking with. Claude is great at explaining concepts, GPT feels better at reasoning through tricky logic, and Ive seen Sweep AI pop up for people who want project-level help directly inside JetBrains instead of switching back and forth with chat. Which model or tool gave
| 1
|
The 4 Layers of an LLM and the One Nobody Ever Formalized People keep arguing about what an LLM is, but the confusion comes from mixing layers that operate at different levels of abstraction. Heres the clean, operator-level breakdown the one nobody formalized but everyone intuye Layer 1 Statistical Pattern Engine the machine itself This is the physical mechanism token probabilities embeddings attention matrices gradient-shaped geometry Nothing here understands. It transforms input into output by
| 1
|
Google must double AI serving capacity every 6 months to meet demand, AI infrastructure boss Amin Vahdat tells employees
| 1
|
Major N.L. healthcare report contains errors likely generated by A.I. 1.6 million Health Human Resources Plan from Deloitte cites research papers that dont exist, making it the second major government policy paper called into question in as many months
| 0
|
freepik is straight up lying about 4k after comparing the outputs, its obvious the images are just cheaply upscaled, not actually generated in true 4k. i commented this under their post and instead of proving me wrong, they just deleted my comment. says a lot.
| 0
|
Nanobanana vs Nanobanana Pro First of all, please excuse me filming with my phone. I couldn't get Loom to run. Quick corrections to the video 1. I used Nanobanana AND VEO3 Nanobanana for the reference images and Veo3 to turn the images into videos 2. I said I created this ad "a few months ago". I fact checked myself. It was actually not "a few months ago", it was Saturday 1st of November but in AI-time it might as well be 3 years 3. This is not intended as a research-level, accurate comparison f
| 0
|
LLMs have a built-in mathematical ceiling on their creative capacity, meaning they will never rival the originality or ingenuity of the most creative individuals - report
| 0
|
Pinterest is leaning hard into AI. The strategy appears to be backfiring
| 0
|
If you had to explain to a superintelligent AI why humanity should continue to exist, what would you say? As AI continues to advance toward superintelligence, this question becomes increasingly relevant. What makes humanity worth preserving? Is it our creativity, our capacity for love and connection, our flaws and imperfections, our stories and cultures, or something else entirely? How would you make the case for human existence to an entity far more intelligent than us?
| 1
|
The Godmother of AI Didnt Expect It to Be This Massive Stanford scientist Fei-Fei Li talks about teaching machines to see as humans do, the US-China AI arms race, and what worries her about a more automated future.
| 1
|
Open Call for ideas My customer operates a kind of marketplace for custom products. The products have - generic properties that can be quantified and imported - geographic properties eg think of a hotel and you want a description for the area We want to feed the descriptions for a history of roughly 2.000.000 products and end up with a machine that produces - a title - a short description - a long description What would be the ideal approach to start? Significant hosting environments are present
| 0
|
90 of Advice You Get Is Wrong Here's What AI Can Do Is the advice you receive from friends leading you in the wrong direction? Paul Allen, founder of Soar AI, believes that 90 of the advice we receive, even from the people closest to us, isnt actually right for us. Its shaped by their strengths, experiences, and perspective. But with AI and psychometric tools, we can map our own patterns of thinking, feeling, and behaving to get guidance that fits who we really are. The future of personal growth
| 1
|
New research shows how AI could transform math, physics, cancer research, and more A new report from OpenAI and a group of outside scientists shows how GPT-5, the companys latest AI large language model LLM, can help with research from black holes to cancerfighting cells to math puzzles.
| 1
|
Big tech's AI deals are creating one giant machine Steven Levy argues in Wired that the artificial intelligence industry has evolved into a single interconnected entitydubbed the Blobthrough a web of partnerships, investments, and cloud agreements among major players like Nvidia, Microsoft, Google, OpenAI, and Anthropic, despite originally being founded to prevent profit-driven control of AI. The recent Microsoft-Nvidia-Anthropic deal exemplifies this consolidation Microsoft commits up to 5 bill
| 1
|
Writer says AI taking jobs is good 'Automation Puts Us Humans On The Right Path To Live A Life That Corresponds To What We Really Want A Life Of Leisure amp Creativity.'
| 1
|
No bailout will be provided when AI bubble bursts Trillion dollars may be vanishing in thin air.
| 0
|
Were in an LLM bubble, Hugging Face CEO saysbut not an AI one Theres been a lot of talk of an AI bubble lately, especially regarding circular funding involving companies like OpenAI and Anthropicbut Clem Delangue, CEO of machine-learning resources hub Hugging Face, has made the case that the bubble is specific to large language models, which is just one application of AI. I think were in an LLM bubble, and I think the LLM bubble might be bursting next year, he said at an Axios event this week, a
| 0
|
I tested AI with videochat and it was concerning how good it is Ive been testing this since it dropped a few days ago because Im a sucker for anything new in ai. The premise is video calls with ai that can read expressions and body language, not just respond, I decided to test it by lying about stuff to see if it could tell I told Charlie, I was excited about a project while deliberately looking stressed and tired, it said "you don't seem excited, you seem exhausted, what's really going on". I f
| 0
|
Teachers and parents weigh benefits and risks of artificial intelligence in schools PBS News
| 0
|
How the EU botched its attempt to regulate AI The AI Act was designed to use Europes economic heft to force companies to create trustworthy AI for its 450mn consumers through a risk-based approach banning the most harmful uses, controlling high-risk systems and lightly regulating low-risk ones. But the laws complexity, its rushed inclusion of AI models such as ChatGPT and its chaotic implementation have turned the AI Act from a symbol of European leadership into a case study for those who say th
| 0
|
AI Slop Has Turned Social Media Into an Antisocial Wasteland Uh, one does not need to read an article to realize social media is more responsible for creating an antisocial wasteland then AI ever could.
| 0
|
U.S. Approves Deal to Sell AI Chips to Middle East. Agreement follows talks between President Trump and Saudi Arabias Crown Prince Mohammed bin Salman.
| 0
|
One-Minute Daily AI News 11192025 1. OpenAI and Target partner to bring new AI-powered experiences across retail.1 2. UN calls for legal safeguards for AI in healthcare.2 3. Trump-MBS meeting brings AI money.3 4. Nvidias record 57B revenue and upbeat forecast quiets AI bubble talk.4 Sources 1 2 3 4
| 0
|
Nvidia's earnings could answer the AI bubble question and upend global markets in moment of truth for Magnificent 7 Fortune
| 0
|
Predictions on what'll burst the bubble Drop here your predictions on what's gonna be the needle that pops the bubble. I for one, doubt it'll be the legislative approach, worst case scenario if the EU gets strict with it, is that it just won't be that Open-Access there. I think something really specific and punctual has to happen. Any ideas?
| 0
|
Adobe bought Semrush as an AI acquisition.. for 1.9 billion.. should that be surprising? Maybe I am not as connected as I thought but Adobe and Semrush seem like a surprising pairing and 1.9 billion on a platform whose core is SEO which is... dying?
| 0
|
Florida nonprofit news reporters ask board to investigate their editors AI use. Suncoast Searchlights four reporters told the board their editor-in-chief was using AI editing tools and inserting hallucinations into drafts. The next day, one of the reporters was fired.
| 1
|
Is there a free AI Image Generator that actually works? As my title indicates, there is frustration. I am not looking for anything crazy, just making some basic still images. The problem I have with many of them They aren't actually free, after 1 image or 1 edit you need to pay. Lame. It rarely gets it right the first time so it needs edits. Additionally, oftentimes the edits fail hard. I'll ask it to say add a flag in the background, and it opts to change the foreground entirely, even when told
| 0
|
D Self-Promotion Thread Please post your personal projects, startups, product placements, collaboration needs, blogs etc. Please mention the payment and pricing requirements for products and services. Please do not post link shorteners, link aggregator websites , or auto-subscribe links. -- Any abuse of trust will lead to bans. Encourage others who create new posts for questions to post here instead! Thread will stay alive until next one so keep posting after the date in the title. -- Meta This
| 0
|
D NeurIPS 2025 Mobile App NeurIPS 2025 is beta-testing a new mobile app this year. Personally, Ive had really good experiences with Whova app at past ML conferences 1. The UI is clean and makes it easy to browse the schedule 2. Lots of active social channels and events pop up weeks before the conference 3. Tons of job postings 4. Easy to reach out to attendees with similar interestsinstitutes But the new app feels pretty dead so far very few attendees downloaded the app, no channels, no activiti
| 0
|
D Transitioning from physics to an ML PhD Hey everyone! Im a physics undergraduate American applying to PhD programs next year, and my research interests are in theoretical neuroscience, mech interp, and physics of learning type work. Theres a couple American university professors in math and physics departments doing research in these fields, but the majority seem to be CS professors at top departments. This worries me about my chances of getting accepted into any program at all planning to app
| 0
|
P My Agents Crashed the Economy, So I Taught Them About Salads I just tried implementing RL in the wild and it was very satisfying seeing agents learn to optimize prices. The implementation is a bit clumsy and uses MDP and value iteration built from scratch so performance is not that good. But am very proud and I envy people who get to work with ML as their 9 to 5. Here is the code
| 0
|
D Vision Transformers and positional encoding Padding the ALIBI tensor to account for the CLS token? Working on visual transformers for images, now experimenting with positional encoding in the form of "Attention with Linear Biase" ALIBI, 1, more specifically 2D-ALIBI 2. Say our image is cut in 3-by-3, resulting in 9 patches. Ignoring batch and head dimensions for simplicity. a Each patch is linearly projected, then the ltclsgt token is concatenated, resulting in a tensor of 10, embedding size.
| 1
|
D Why arent there more multimodal large foundation models out there? Especially in AI for science? With all the recent work out on multimodal foundation models etc, why arent there more foundation models that utilize data in different modalities maybe even all possible available modalities for the data of interest? I think there are some interesting success cases for this AlphaEarth, so what are some of the barriers and why arent more people doing this? What are some frequent challenges with mul
| 1
|
P Are the peaks and dips predictable? I am trying to make a model that can predict future solar energy generation even few hours with great accuracy is a good start. The problem are the constant change of clouds, although clearsky variable is present in the model, clouds create dips and peaks in energy generation you see in the image. Any suggestion on how the model can predict them better? Alternately, is there model already build that can better predict? Edit For more context Model is trained
| 1
|
D New results on ARC 12 challenge, overfitting? Never heard about this company, Poetiq, apparently their system used gemini 3.0 and was able to get accuracy to above human baseline levels. Crazy if true. Waiting for confirmation from ARC people. Source The github shows some of the tricks they used, to be honest it looks a little like overfitting, there are numpy transformation hardcoded into the prompts Seems slightly against the spirit of the challenge since it is encoding specific priors to be
| 1
|
D Question regarding CS Phd admission Hi all, I recently published a paper in ICLR datasets and benchmarking track and it got positive reviews, i enjoyed the research process and im thinking of applying for phd programs in t30 universities in usa. However i come from a tier 3 college in india and the paper i published is self advised i didnt have anyone to guide meadvise me through. And i dont know any well known researchers who can write me a recommendation letter. How do i tackle this issue? I
| 0
|
R SAM 3 is now here! Is segmentation already a done deal? The core innovation is the introduction of Promptable Concept Segmentation PCS, a new task that fundamentally expands the capabilities of the SAM series. Unlike its predecessors, which segmented a single object per prompt, SAM 3 identifies and segments all instances of a specified concept within a visual scene e.g., all "cats" in a video, preserving their identities across frames. This capability is foundational for advanced multimodal AI
| 1
|
R Segment Anything Model 3 SAM 3 is released Abstract We present Segment Anything Model SAM 3, a unified model that detects, segments, and tracks objects in images and videos based on concept prompts, which we define as either short noun phrases e.g., yellow school bus, image exemplars, or a combination of both. Promptable Concept Segmentation PCS takes such prompts and returns segmentation masks and unique identities for all matching object instances. To advance PCS, we build a scalable data en
| 0
|
D Typical processes for ICLR review responses I'm responding to ICLR reviews for the first time and I had a quick question on what the typical protocol for review responses are. I have not had the opportunity to run sufficient experiments to respond to reviewer comments. I know ICLR recommended responding within a week i.e., by tomorrow. What should I do if I can't fully respond to reviewer requests? Should I a Respond to their comments, with results that I have done so far, and just say that I
| 1
|
D Scale-out is the silent killer of LLM applications. Are we solving the wrong problem? Everyone's obsessed with cold starts. But cold starts are a one-time cost. The real architecture breaker is slow scale-out. When traffic spikes and you need to spin up a new replica of a 70B model, you're looking at 5-10 minutes of loading and warm-up. By the time your new node is ready, your users have already timed out. You're left with two terrible choices Over-provision and waste thousands on idle GPUs. U
| 0
|
D Is it worth the time to publish and prepare for archival ACLEMNLP workshops? Is it productive as a grad student currently master's and applying for PhD to spend time working on an archival workshop at venues like NAACLACLEACLEMNLP? I see opinions around that you shouldn't even consider workshops as papers will not be as highly regarded as main conference papers. Is there any advantage to attending and submitting to archival workshops? I see many relevant workshops to my work, and I am thinking
| 0
|
D Has anyone used ONNX Runtime ORT CUDA for multilingual embedding models e.g., LaBSE on GPUs? I have a project where we have to use an LLM to generate similarity matrices for semantics. I am doing this in PySpark using AWS EMR, and Googles labse model. I converted the labse model to onnx runtime so i can keep my spark ML pipeline lightweight without installing PyTorch, TensorFlow or Sentence-Transformers. My experiments have been successful so far and I then read the labse model into my ML pipe
| 1
|
R Unlocking Out-of-Distribution Generalization in Transformers via Recursive Latent Space Reasoning 1. arxiv 2. openreview I found this paper both really interesting and clear. No one part is very novel, but It composes disparate threads to obtain what looks like strong results in OOD length generalization. Even for the toy task, and using a DSL vs. being an LM, length-generalizing on simple math gt4x is impressive, from what I've read. This also fits my priors for the key elements of unlocking
| 1
|
D Comparing Giskard and Rhesis for LLM evaluation looking for experiences I'm evaluating different open-source tools for testing LLMs and RAG pipelines. I've come across Giskard and Rhesis, and they seem to take different architectural approaches. Here's what I understand so far, corrections welcome Giskard Built-in test suites and quality checks Python-first with inspection UI Strong focus on model testing and guardrails Established ecosystem with documentation and examples Rhesis Integrates mu
| 0
|
P How can your AI skills help solve one of the worlds biggest challenges access to clean water? Around the world, billions of people face obstacles in sourcing clean and safe water for their daily needs. But with innovation, collaboration, and advanced technologies, we can change this trajectory. Thats where the EY AI amp Data Challenge comes in. Join the challenge to develop cutting-edge AI models to forecast water quality using satellite, weather, and environmental data. Your models will provi
| 1
|
P SumGPT to test robustness of the attention based modules Hello, sumgpt is decoder only gpt model inplemented from scratch that learns floating point summations like '2.41.3'. My aim was to test attention based modules for critical applications. You can easily create data, verify ground truths and monitor and detect hallucinations and errors, increase context length as you wish etc. Also it can be easily trained on desktop and results are actually interpretable whether it learned the summation
| 0
|
D Is a PhD Still Worth It Today? A Debate After Looking at a Colleagues Outcomes So I recently got into a long discussion with a colleague about what actually counts as a successful PhD in todays hyper-competitive research environment. The conversation started pretty casually, but it spiraled into something deeper when we brought up a former lab-mate of ours. Research area Clustering and Anomaly detection Heres the context By the end of his PhD, he had three ICDM papers and one ECML paper, all f
| 1
|
Beyond Hyperparameters We're Now Quantifying and Steering the Internal Physics of AI Training. R This morning, I've been validating a core concept from my AGI research the Vector Space Mapping VSM protocol. The theory? To truly understand Transformer models, we must first quantify the specialization of their attention heads. Initial tests were paradoxical our "specialization" metric sigma_a was flat, even as the model learned. This wasn't a bug, but a discoveryour measurement tool was at the wro
| 1
|
R Generative Flows on Weight Space for Covariate Shift Detection AAAI 2026 Workshop gtAbstract Flow-based generative modeling provides a powerful framework for reasoning about uncertainty in weight space. In this work, we explore model uncertainty and distributional anomalies through weight space learning, where a generative meta-model learns a distribution over neural network parameters that achieve comparable performance. Leveraging flow matching, we capture the geometry of weight space to ena
| 1
|
D Resources for Designing Out of Distribution Pipelines for Text Classification Hey all, I am looking into designing an automated system for evaluating data points as being out of distribution. This would be for a transformer classification model , multi-class setting. I am finding good resources very hard to come by. Currently the ideas I have had are maximum classification score, entropy of probability distribution and some measure of embedding similarity compared to the training dataset. Does
| 1
|
R LeJEPA New Yann Lecun paper Abstract Learning manipulable representations of the world and its dynamics is central to AI. Joint-Embedding Predictive Architectures JEPAs offer a promising blueprint, but lack of practical guidance and theory has led to ad - hoc RampD. We present a comprehensive theory of JEPAs and instantiate it in LeJEPA, a lean, scalable, and theoretically grounded training objective. First, we identify the isotropic Gaussian as the optimal distribution that JEPAs embeddings s
| 1
|
D Let's discuss World Models Hey everyone, I've been reading about "World Models" for a while now and wanted to share my understanding of them, as well as why I think they're such a big deal, especially for general-purpose robotics and potentially a major step toward "AGI" What is a World Model? A world model is a system that builds an internal representation of the physical world, much like a Large Language Model LLM builds an internal representation of human knowledge, logic, and culture as ex
| 1
|
'We cloned Gmail, except you're logged in as Epstein and can see his emails' is the most impressively cursed tech project of the year
| 1
|
Google tells employees it must double capacity every 6 months to meet AI demand
| 1
|
Major N.L. healthcare report contains errors likely generated by A.I.
| 0
|
France will investigate Grok chatbot after Holocaust denial claims
| 1
|
Icelandic is in danger of dying out because of AI and English-language media, says former PM
| 0
|
US Mulls Letting Nvidia Sell H200 Chips To China, Sources Say
| 0
|
White House threatens states with lawsuits over AI regulation
| 0
|
Ubisoft says AI is 'as big a revolution for our industry as the shift to 3D,' its magical AI NPCs are out of prototyping, and you bet it's got 'all our studios' working with the tech
| 1
|
No bailout should be provided when AI bubble bursts
| 0
|
Yann LeCun believes that in the future, AI will have emotions and subjective experience. Im putting this here for those who still think Yann LeCun is a Gary Marcus 2.0.
| 1
|
"Taming the Long-Tail Efficient Reasoning RL Training with Adaptive Drafter" "The emergence of Large Language Models LLMs with strong reasoning capabilities marks a significant milestone, unlocking new frontiers in complex problem-solving. However, training these reasoning models, typically using Reinforcement Learning RL, encounters critical efficiency bottlenecks response generation during RL training exhibits a persistent long-tail distribution, where a few very long responses dominate execut
| 1
|
A Figure.ai safety chief says a humanoid robot once cut a 14" gash into a steel fridge door during malfunction. He was fired for warning the robot could kill and sued the company. Figure.ai denies it.
| 0
|
Gemini 3 turned my book into a video game in 2 minutes wow!! I uploaded my book to Gemini 3 and I heard about vibe coding I tried a bunch of stuff but I really am not very good at it. I tried Gemini 3 when it came out inserted my book into the build section and told it to make a RPG utilizing all of the power of Gemini Now just random people like me can create full-blown video games on their own material and have it actually be really fun and impressive I am completely blown away.
| 1
|
Gemini 3 Pro solves IMO 2025 P6 with some prompting no hints or tools involved. Doesn't look like training data contamination since GPT-5.1 High, OpenAI's unreleased internal model, and even AlphaEvolve all fail on it.
| 0
|
How far can human cognition be enhanced in the future ? What future so you see ? With mental training we sill can achieve some stuff. But with genetic editing how far can we go ?
| 1
|
OpenAI Locks Down San Francisco Offices Following Alleged Threat From Activist A message on OpenAIs internal Slack claimed the activist in question had expressed interest in causing physical harm to OpenAI employees.
| 0
|
It will be funny if... If in 5-10 years, when there will be all sorts of photovideo AI that have reached perfection, the limits etc. And people will start using this plastic face effects that current AI models are making. It's like a today's point-and-shoot camera vibe from 90's
| 0
|
"Evolution Strategies at the Hyperscale" "We introduce Evolution Guided General Optimization via Low-rank Learning EGGROLL, an evolution strategies ES algorithm designed to scale backprop-free optimization to large population sizes for modern large neural network architectures with billions of parameters. ES is a set of powerful blackbox optimisation methods that can handle non-differentiable or noisy objectives with excellent scaling potential through parallelisation. Naïve ES becomes prohibiti
| 1
|
Gemini 3 pro places 8th in EsoBench, which tests how well models learn and explore unfamiliar programming languages. Full disclaimer, this is my own benchmark I played around with Gemini 3 pro before testing it and was quite excited as it felt really clever, and I was expecting it to be a new SOTA. But, it failed to beat Opus 4.1, or even o4 mini from April this year. However, in the heatmap you can see Gemini 3 pro has a yellow square in column 5. The columns are the question number, and the ro
| 0
|
"PertAdapt Unlocking Single-Cell Foundation Models for Genetic Perturbation Prediction via Condition-Sensitive Adaptation" "Single-cell foundation models FMs pretrained on massive unlabeled scRNA-seq data show strong potential in predicting transcriptional responses to unseen genetic perturbations. However, existing approaches insufficiently transfer pretrained knowledge and overlook the imbalance between perturbation-sensitive and insensitive genes, yielding only marginal improvements over non-
| 1
|
Gemini 3 is bonkers I don't mean that in a good way. I mean that if you examine it's thought chain, it is clearly all kinds of crazy. I tried to switch from Gemini 2.5 to 3, and when I asked Gemini to confirm it worked it was weirdly evasive and very slow to answer for such a basic question. I took a peak at it's logic, and... hoo boy. It spent 75 if it's time trying to decide whether I or someone else was trying to deceive it by implicitly lying to it about the date being the probably-imaginary
| 0
|
Elon Musk Could 'Drink Piss Better Than Any Human in History,' Grok Says Elon Musk is a better role model than Jesus, better at conquering Europe than Hitler, the greatest blowjob giver of all time, should have been selected before Peyton Manning in the 1998 NFL draft, is a better pitcher than Randy Johnson, has the potential to drink piss better than any human in history, and is a better porn star than Riley Reid, according to Grok, Xs sycophantic AI chatbot that has seemingly been reprogrammed
| 1
|
Gemini 3 is on another level when it comes to writing stories One of my "benchmarks" is to prompt a LLM to write a short story about a side character from a moderate popular franchise. Until now every LLM got most of it wrong. They didn't seem to know the side character and even after giving them more informations or task them with research, they failed to get the character right. Their relations with others characters and the whole placement in the setting were not fitting. On top of that, the
| 0
|
AI Video Benchmarks Levels 1 to 10 1. Flickering, unstable video with major frame inconsistencies. 2. Mostly consistent video with fewer flickers but still unstable over time. 3. Consistent short video with basic or imperfect audio. 4. Characters remain consistent and audio is clear and synced enough. 5. Characters keep identity, have their own voices, and show believable acting. 6. Multi-shot scenes stay coherent with realistic motion, sound, and lighting. 7. Near-photorealistic scenes with str
| 0
|
Nano banana pro is the first image generator that can consistently produce the likeness of famous people, and which they dont look slightly off I consider it a genuine GPT 3.5 level breakthrough. I just made both of these photos with the same prompt, the first photo is with nano banana pro and the second is with regular nano banana, the third photo shows the prompt I used, the difference is dramatic. Every other AI image generator is dogshit in comparison, the non pro version of nano banana incl
| 1
|
Nano Banana Pro Infographic. The improvement is simply impressive. Prompt Create a detailed Infographic of the functioning and flow of an automatic coffee machine like a Jura. From bean basket, to grinding, to scale, water tank, boiler, etc. Id like to understand technically and visually the flow.
| 1
|
OpenAI podcast hinting at more powerful internal models Very interesting. They seem to flip flop on the messaging of whether or not they have better internal models, but this is the firmest confirmation Ive seen so far. Not sure if this means yea were already running 5.5 internally or if theyre just doing the same models as us but with 2-10x the amount of thinking compute. Ik gemini3 already confirmed were far from a wall, but its nice to hear this as well. 2026 is going to be wacky as fuck. Hur
| 1
|
Lidar Companies I am looking at investing in one or more companies that manufacture Lidar specifically for automotive. A quick search gave me these companies - Hesai HSAI - Luminar LAZR - Innoviz INVZ - Oyster OUST Of all these companies only Hesai looks profitable. What do you guys think? Has anyone experienceinsights worth sharing? Thanks in advance!
| 1
|
Anyone else feeling like the markets vibe totally shifted this month? Not sure if it's just me, but the market has felt weird lately. Tech was carrying everything for so long, and now we are suddenly getting hit with AI bubble talk, rate cuts, and random sell-offs that don't even line up with the earnings. The Fed basically hinting we are not getting as many cuts as we wanted and big AI names cooling off, it feels like we are entering a mini risk-off phase. The tech strategies arent as comfy as
| 0
|
In leaked memo, Altman is panicing about OpenAI's future after Gemini 3.0 release No Paywall Altmans message marks a rare moment of vulnerability for a CEO known for his relentless optimism. He explicitly described the current atmosphere as having rough vibes, a departure from the triumphalism of its 2025 DevDay Dominating the admission is a concern over technical leadership. Acknowledging Googles resurgence, Altman conceded that OpenAI is now in a position of catching up fast. Independent bench
| 1
|
My top growth stocks and why To not sound like a complete idiot - I have 10 into spy, 20 into GOOG, 10 into MSFT, 10 in AMZN and 10 in META. The remaining 40 are however my chance to rise from poverty. I live in a country with a life expectancy of 63 years, so please dont blame me that I cant just take my time and wait blue chips to compound. I have 10 of my net worth in AUR Aurora Innovation - the future of trucking is autonomy and the company is likely to become profitable in 2027. Another 10
| 1
|
Amazon wants to repalce 600k workers with robots which will lead them in saving 30 cents for every item they store and ship from warehouses Source Personally I think a big job cut like this is not a good thing and these moves are a starting point for a mass displacmene of works which risks destabilizing society just to give more profit to a few companies.
| 0
|
Amazon cut 1,800 engineers in its record layoffs last month, filings show Amazons 14,000-plus layoffs announced last month touched almost every piece of the companys sprawling business, from cloud computing and devices to advertising, retail and grocery stores. But one job category bore the brunt of cuts more than others engineers. Documents filed in New York, California, New Jersey and Amazons home state of Washington showed that nearly 40 of the more than 4,700 job cuts in those states were en
| 0
|
Shares in GoodRX? Disclaimer i am fairly new to the stock world and dont know much, but Im down to learn. I currently own about 40 shares in GoodRX. My reasoning behind buying this resides more-so in fact that the government may not expand health care subsidies, so as a way to save money, the average American that is effected by this may turn to sites or apps like this one which would then bolster the company itself and increase share prices. Maybe im wrong, I dont know. But is this a logical co
| 1
|
Eli Lilly hits 1 trillion market value, a first for a health-care company Eli Lilly reached a 1 trillion market capitalization on Friday, the first health-care company in the world to join the exclusive club dominated by tech firms. Eli Lilly briefly hit the 1 trillion mark in morning trading before retreating. It was last trading around 1,048 a share. The companys stock has climbed more than 36 this year as investors applaud the gains it has made over chief rival Novo Nordisk in the GLP-1 drug
| 0
|
US mulls letting Nvidia sell H200 chips to China, sources say gt Nov 21 Reuters - The Trump administration is considering greenlighting sales of Nvidia's H200 artificial intelligence chips to China, people familiar with the matter said, as a bilateral detente boosts prospects for exports of advanced U.S. technology to China. gt The Commerce Department, which oversees U.S. export controls, is reviewing a change to its policy of barring sales of such chips to China, the sources said, stressing tha
| 0
|
Everyone is talking about AI causing the continued fall. It's not. It's the September jobs report. If it was only the AI trade cracking, you'd see much more rotation into other sectors, esp. defensives. That's not happening. The major catalyst today was an unusually strong September jobs report, released after the shutdown. Given it's one of the last major data points a hawk-postured fed will get before December on the labor market, participants are degrossing risk assets in case that rate cut i
| 0
|
Strategys Saylor defends business model amid MSCI index concerns gt Strategy Chairman Michael Saylor defended his companys business model on X, emphasizing that the firm is not a passive investment vehicle amid concerns about potential removal from major equity indices. gt Saylor stated that Strategy is "a publicly traded operating company with a 500 million software business and a unique treasury strategy that uses Bitcoin as productive capital," distinguishing it from funds, trusts, or holding
| 1
|
AI bubble concerns continue to impact US stock market performance Nvidia opened higher but closed lower. Today's US stock market performance wasn't what the bulls wanted. On what appeared to be a high-volume day, many stocks left an ugly candlestick on the charts. I've been arguing all week about a bubble in Nvidia, and the performance of Google's Gemini 3 further reinforces that argument. Google proved it can build incredibly large language models LLMs without Nvidia chips, meaning other compan
| 0
|
Data vacuum, crucial FOMC meeting! Below is my brief analysis and personal trading strategy. We can discuss any better suggestions or perspectives. 1. Timing is critical Due to data reporting delays, we may see a large volume of employmentCPI data between late November and early December. This means the Fed may receive a large amount of new data when it makes its decision on December 10th. In short Both the market and the Fed are in a semi-blind spot market volatility may not be over yet. 2. My
| 0
|
So markets have been rolling over for 10 days, thousands of headlines about bubbles, and Nvidia posts a beat and we're giga-rallying again? And it wasn't just junk rags coming out with bear sirens. Equity strategists from nearly every major bank except Morgan are on record warning of "correction", "more pullback ahead", "trees don't grow to the sky" and more. Surely it can't be healthy that trillions of dollars of portfolios and options bets are turning on the whims of one stock...
| 0
|
September NFP drops today, job gains expected at 50K, unemployment at 4.3 Today the U.S. Nonfarm Payrolls NFP data for September hits the tape after being delayed by the government shutdown. Economists forecast a gain of 50,000 jobs, up from just 22,000 in August, with the unemployment rate held at 4.3. Even though the numbers are backward-looking, this release matters more than usual because it may be the last full jobs print ahead of the Federal Reserves December meeting. With no full October
| 0
|
general question about how a market crash happens hi.. i got to thinking after the unusual rise and reversal yesterday that nobody seems to know exactly who began the reversal.. how it works. does anyone have any insight into this regarding 1 who triggers the reversal is it one large fund with a person deciding it or are market makers colluding to stage a large broad reversal.. or was it all algorithcomputer driven? 2 how was such a broad selloff across all segments triggered and maintained to t
| 0
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.