🦞🌯 Lobster Roll

All LHN/.~Ars
RSS

Showing stories from 2025-04. View all

NewestOldestTop ScoredMost Discussed
AI suggestions make writing more generic, Western (news.cornell.edu)
Show HN: I gamefied an AI-powered health optimizer (vitalflip.replit.app)
I built VitalFlip because I realized: People don&#x27;t care how many steps they took - They want to know what to do next to optimize their health.<p>It’s a GPT-powered AI that reverse-engineers your lost motivation and gives you a tailored plan to restart — emotionally, not just physically.<p>Built...
Science, Promise and Peril in the Age of AI (quantamagazine.org)
Gemini AI video generator powered by Veo 2 (gemini.google)
Visa wants to give artificial intelligence 'agents' your credit card (apnews.com)
ChatGPT vs. students: Study reveals who writes better (phys.org)
I built an AI-powered business system using ChatGPT, Notion and Midjourney (skylark862.gumroad.com)
Google Places Ads Inside Chatbot Conversations with AI Startups (bloomberg.com)
Show HN: Notable LLM Research Papers (github.com)
Building Private Processing for AI Tools on WhatsApp (engineering.fb.com)
Chat GPT Is Eating the World (chatgptiseatingtheworld.com)
Show HN: A simple flashcard web app with a little bit of AI (clsstudio.com)
Marc Andreessen Says One Job Is Mostly Safe from AI: Venture Capitalist (gizmodo.com)
Is AI alignment repeating the mistakes of "New Coke"? (en.wikipedia.org)
RustAssistant: Using LLMs to Fix Compilation Errors in Rust Code (microsoft.com)
Mercury: Commercial-scale diffusion language model (inceptionlabs.ai)
When ChatGPT Broke an Entire Field: An Oral History (quantamagazine.org)
Confidential AI (tinfoil.sh)
Ask HN: What benchmarks are you using to judge AI models?
There are so many models, and so many new ones being released all the time, that I have a hard time knowing which ones to prioritize testing anecdotally. What benchmarks have you found to be especially indicative of real-world performance?<p>I use:<p>* Aider&#x27;s Polyglot benchmark seems to be a d...
The Strange Physics That Gave Birth to AI (quantamagazine.org)
ChatGPT MCP (personas.omi.me)
Ask HN: Will AI make humans smarter or dumber in the long run?
Big question, probably too big, but let’s speculate wildly just for fun. On one hand you have the argument that the ultimate knowledge&#x2F;productivity boost comes from combining humans and AI effectively, bouncing ideas off of them, looking things up more efficiently, tutoring, etc. On the other h...
What AI Can't Replace: Intuition, Alignment, and Taste (startupsoup.substack.com)
Google funding electrician training as AI power crunch intensifies (reuters.com)
Show HN: AI Virtual Influencer That Posts for You (Fully Automated) (app.evatar.ai)
I’ve been building this over the last six months — it’s an AI virtual influencer that posts content as your digital self, fully automated. You choose or create a custom avatar (with your voice, if you want), pick topics that matter to you, and it starts creating short-form content — script, voice, v...
League of Prompts: Write prompts, get AI-rated, climb ranks, win rewards (leagueofprompts.com)
Duolingo drops nearly 150 new language courses, says AI helped build them (neowin.net)
JetBrains defends removal of negative reviews for unpopular AI Assistant (devclass.com)
Ask HN: Does this estimate for GPT-4's 128k token window make sense?
GPT-4 supports a 128,000-token context window (per OpenAI).<p>I&#x27;m trying to understand how much text that really is. Here&#x27;s my rough estimate:<p>1 token ≈ ¾ word ≈ 4 characters<p>A single-spaced A4 page (12pt font) ≈ 500–600 words → ~700–800 tokens<p>128,000 ÷ 800 = ~160 pages<p>So: GPT-4 ...
AI infrastructure is hot. New power distribution and liquid cooling can help (cloud.google.com)