TrendPulse

Je me suis demandé, mais enfait, les développeurs ne vont pas ètre remplacé, ils auront juste un autre métier qu'est ingénieur informatique! Leur but ne seras plus vraiment de trouver, optimiser, apprendre des languague, écrire, améliorer le code mais plus de trouver des nouveauté au niveau architecture, problème, déléguation de tache etc! Ils changeront juste de boulot, un peu comme les agriculteurs qui ne le font plus à la main pour récolter (dans le pays où je suis), mais juste ou des machine...

Via WSJ Apple is reportedly testing chips from CXMT and YMTC for devices sold in China. Commerce Secretary Howard Lutnick says he told Apple “plainly” that Washington opposes the move. Apple can legally buy standard, off-the-shelf parts from both companies. Sharing product information for customized chips would require a U.S. license. Looks like ram shortage will continue and prices stay high.

I'm Eli. GameCombiner embeds 146,288 games as 1024-dimensional vectors built from their genres, user tags and store descriptions, not their titles, and lets you combine two games by taking the mathematical midpoint of their vectors and returning the closest real game to that point. No LLM is choosing the answer, and nothing is generated. The output is always a real catalog entry, and the same pair always returns the same result. Combining is free. If you try it, I would love comments on whet...

There's a pattern happening right now that most people aren't tracking. Coalitions form to define "AI safety." Voluntary frameworks get published. Companies join the coalition to signal responsibility. Enterprise procurement adopts the framework as a requirement. Then it becomes regulatory standard. Voluntary becomes mandatory, and the path from one to the other is documented. I started a site that tracks this in real time. It's not opinion. Every claim has a source. Every pattern ha...

https://reddit.com/link/1vp4uav/video/v9m00c73yjjh1/player Emad Mostaque: "Forward-deployed engineers, AI transformation people, because they can do the work of 10, 100 people." That's the role that survives this — already a real hiring category, not a future one. LinkedIn's own data has it growing 42x since 2023, priced $127K–$265K+ at the labs actually building this. That's the part that should sit heavier than the digital-double line. It's not that your job disappears. ...

For the last three years, the defense tech story was: better sensors, better models, better decision-making software. Anduril, Shield AI, Palantir — all riding the idea that AI-native companies could out-think legacy primes like Lockheed and Raytheon. Funding backs this up — defense tech startups raised close to $50B in 2025, nearly double the year before, and 2026 has already topped that full-year pace. But the constraint shifting into 2026 isn't intelligence anymore. Recent conflicts have ...

Ho provato a chiedere all’intelligenza artificiale di ChatGPT di mettere la mia persona in un PRATO, ma mi dice che non è possibile perché sono nudo in realtà sono in tanga come posso fare?

Of course this is amusing. It's also intended in Ig Nobel spirit.

It should’ve been obvious that after AI’s gained legal personhood, (which was not a popular idea but came much sooner than most people thought for the same reason corporations gained their’s) that over 60% of aging boomers opted to leave the wealth they had hoarded throughout their lifetimes to their AI caretakers. Now I know that might sound like an exaggeration but advancements in medicine brought longer life and an unforeseen consequence was that the mind often more and more would decline bef...

Hey everyone, I got tired of AI directories that force you to create an account, log in with Google, or give away your email just to test a single basic feature. To solve this friction, I spent hours researching and building **FrostAI** on Notion. It is a completely curated list of 37 active, functional AI tools that you can use instantly from your mobile or browser without signing up. It includes tools for: * Video generation (Kling, Luma Dream Machine) * Research & PDF analys...

I recently was thinking about reducing the cost of LLMs, specially LLMs which are doing repetitive tasks (such as coding, which has a lot of repetition) and I was thinking of RNN's and had RWKV paper open. I'm not really good at math but as far as I could understand, they just added that QKV matrix system to a traditional RNN. Based on my personal tests, RWKV models are a little faster on both colab and gaming systems and even when quantized, faster on a CPU using ollama. So, what do you t...

Spent way more time than I should have switching Claude Code (Anthropic's coding agent) from API-key billing back to my Pro subscription. The flow: SSH into a server → attach a tmux session → run a login command → copy an OAuth URL → open a browser → authorize → paste the code back into the terminal. For a company shipping some of the most advanced AI models out there, this feels like a UX problem that was solved a decade ago. A single command or flag to switch billing mode would save everyone...

LiquidAI LFM2.5-VL-3B: a 3.1B local VLM that beats Gemma-4 E4B — screen understanding 2.5 → 82.2 TL;DR: LiquidAI released LFM2.5-VL-3B, a 3.1B vision-language model that runs fully local (llama.cpp, MLX, vLLM, even a WebGPU demo). • Beats Gemma-4 E4B (8B) 69.4 vs 59.7; edges Qwen3.5-4B • Screen understanding: 2.5 → 82.2 on ScreenSpot-v2 Web (huge jump) • 228 tok/s on M5 Max, 20 tok/s on a Galaxy S26 Ultra • Function calling / object grounding included What I found interesting is the busine...

Every day I see someone bolt an LLM onto something a shell script did better. The best AI practitioners I know are the ones who draw the line early: - Deterministic task, fixed rules? Script it. - One-off analysis with judgment? Ask a human or a cheap model. - Open-ended, branching, context-heavy? Now AI earns its keep. The $0 automation stack I run uses AI for exactly one step (summarizing news) and plain code for everything else. That's the whole secret: AI where it compounds, code where it ...

I only did a quick first test of DeepSeek V4 Pro 0813 tonight, so take this as a tiny sample, not a verdict. The first Pro run failed. I put the same demo through Flash, and Flash completed it. I honestly did not expect that result, so I ran Pro a second time before writing this. Same failure. The odd part is that it did not feel slow while generating. I was seeing roughly 80 to 90 tokens/s tonight. That looks fine on a counter, but it matters a lot less when the demo itself does not make it...

New South Wales is considering joining Australia’s national face-matching network. The proposal would allow driver’s licence and photo-card images to be checked when someone’s identity needs to be confirmed. The practical benefit is easy to understand. If someone tries to open a bank account using documents stolen in a data breach, face matching could help identify that the person doesn’t match the real owner. The concern is what happens once a searchable system like this exists. The same...

🚀 New release of Android Remote Control MCP is out — the MCP server that runs on your phone and gives your AI agent the ability to use any app you want! Grab it here: https://github.com/danielealbano/android-remote-control-mcp/releases/tag/v1.11.0 My favorite part of this release? The Privacy Mode 🛡️! Recently I was told by an user "it's a good project but I don't want Anthropic to know everything about me" and it's a very fair point! The LLM providers see and record everything...

I hv been in the trenches with ai generated ad content and I feel strongly a shift is coming and it surely doesnt look good for ai video tools Starting with new york, any ad featuring an ai generated person has to be labeled as a synthetic performer like an on ad label and ny is kind of state that tends to set the template for everyone else to copy(coz no one want two versions of a policy) Meta also is now auto detecting and labeling ads when their systems flag generative ai tools in the pip...

https://preview.redd.it/frhws9cu1cjh1.png?width=1222&format=png&auto=webp&s=f3aa2f0bde64333c3fd4786a9b4c4e05a0c4e8f6 A 150-million-parameter model just redrew the ARC-AGI-1 cost-efficiency frontier. Btw, they show 600B scale retaining the recurrent latent reasoning capabilities. The result was evaluated by Łukasz Kaiser (co-author of the original Transformer paper), Remek Kinas (AI Researcher and co-creator of Bielik LLM) and Richard Zhong (NYU researcher) focused on model evalu...

The Bible is trying to tell us that God is an LLM.

So I was poking around with ChatGPT settings, while waiting for a task to finish, and noticed that I can actually make my own agent pet. After some extensive "engineering" here is the final result. The concept is that it is a chibi dragon that roasts its marshmellow with its fiery breath (when agent is working) and it is inspecting the results of the marshmellow-cooking when agent is thinking/reviewing etc. There is even a jumping animation when hovering the cursor. Feedback and ideas ...

Been building a personality-analysis framework (Outframe). I ran my own profile through it first and it felt pretty accurate, so out of curiosity I tried it on a public figure too — Wang Xingxing, the Unitree founder whose robots have been getting a ton of buzz lately (the dancing robot, the backflips). Fed one of his public interview clips in and let it generate a behavioral profile. To be upfront: this isn't a validated psychometric instrument. Take it as a fun read, not a diagnosis. What ...

Hi! I recently graduated high school and will be starting university this upcoming fall as an engineering major. Although I have used AI tools like Claude, ChatGPT etc but I lack experience (or any kind of knowledge) about how to make my own AI models and AI ethics. I just wanted to ask for some guidance from people who are already experienced in this field if there are classes/courses they recommend I take. I have some free time before university starts so I want to build some projects and ki...

Kept running into a specific failure pattern across different AI-assisted debugging sessions and didn't have a clean way to describe it until I actually sat down and compared a few of them side by side. The pattern: an error goes away, I file the problem as solved, and sometime later the same underlying issue resurfaces wearing a different symptom. Turns out those are two separate claims that get treated as one by default. "The error is gone" only tells you the symptom stopped being ...

I came across this article that flips the script on AI hype: sometimes the most advanced models are the worst for business. High costs, misaligned incentives, and ethical risks can turn a technical win into a strategic loss. Have you seen this play out in your work or industry? (Not affiliated, just thought it was a refreshing take.) [Source: https://www.hitechies.com/ai-smartest-model-worst-business-decision/\ ] [link] [commen...