🌟 Working on a pixelart RPG with a twist game On @thisisfrag.bsky.social Nostr npub1a7fxqeagxgndv9g8u7hfjartu593rjtn8puuaqlars8lgd44q7lqf2q22y

Brussels
Joined July 2011
2 senior devs. 1 Godot project. 0 monetization strategy. Infinite chaos. Looking for a pixel artist to make the cutest little world while we two super expert dinosaurs argue about code architecture and AI algorithms. Interested? You're probably crazy. That's a good sign. 🎮 DM
16
12
491
“Tokens are to software as bricks are to architecture: a necessary input, ... when produced faster than they can be inspected, a hazard.” The Mythical Token Month (book) available for purchase (ebook). Soon paperback on Amazon. mythicaltokenmonth.com/ #AI #software #engineering
1
30
Edge inference isn't just about speed; it's a sovereignty architecture. Moving computation closer to use ensures AI systems on sovereign infrastructure can't be disabled externally by geopolitical events or vendor pivots. Full episode on datascienceahome. #EdgeAI #Sovereignty
2
1
154
ASML, a Dutch company, holds a monopoly on EUV lithography machines essential for advanced chip production. Without ASML, TSMC can't make 3nm chips, and without TSMC, no one produces NVIDIA-class chips. #Tech #SupplyChain #AI
125
Safety talks about open source often focus on bioweapons, but the real aim is to target open models over closed ones. #OpenSource #CyberSecurity #AI
63
AI product launches aren't science communication; they're manufactured consensus. Demos are unreproducible, benchmark claims lack methodology, and the press runs continuously to manipulate stock movement. This cycle repeats every year. #AI #Tech #Marketing
1
51
The economics of AI at scale are dominated by inference, not training. As AI moves to production, inference demand skyrockets, making cloud providers crucial to the AI economy. They control the infrastructure that generates ongoing revenue and leverage. #AI #CloudComputing #Tech
89
The real AI threat isn't giant models from rivals. It's smaller, open-source AI that runs offline on your laptop. The full episode is available on datascienceahome.com, Apple Podcast and Spotify. #AI #OpenSource
2
66
Big labs won't fight this bill publicly, but expect them to lobby hard through trade groups to narrow definitions and delay its impact. They'll welcome it in principle while working to make it voluntary or capped. #TechPolicy #Lobbying #Regulation
44
US export controls on NVIDIA chips to China are a major AI policy move, making supply chain dependency a strategic tool. China's response? Accelerating domestic semiconductor capabilities. #AI #Semiconductors #USChina
57
Open source vs. open weights: a crucial distinction often overlooked, especially for financial reasons. #OpenSource #AI #Tech
75
Europe's AI Act tackles real issues like vendor-manufactured hype and behavioral extraction, unlike the GDPR. The focus isn't AI danger, but engineered dependencies. Full episode available on datascienceahome.com, Apple Podcast and Spotify. #AIAct #Regulation #Tech
1
37
Industry leaders like Sam Altman, Linda J. Hasen, and Elon Musk proposed AI regulation concepts. Now, they're being codified into law, making it harder to argue against policies they themselves suggested. #AIRegulation #TechPolicy
1
46
AI sovereignty matters. Systems on sovereign infra can't be disabled by sanctions or pivots. APIs to foreign data centers? You're a tenant, vulnerable to disruption. #AISovereignty #CyberSecurity #Tech
1
1
3
157
A contractor builds a custom kitchen over 6 months, using 20+ years of experience. You watch, learn, and then build your own extension, never calling them again. Big AI companies offered a similar deal, but open models are changing that. #AI #OpenSource #Tech
52
AI models are trained on global data, from books to code, across continents and languages. Should the benefits exclusively go to one nation? This frames a global commons as a national one, a legally shaky argument globally. #AIethics #GlobalCommons #TechDebate
43
The AI compute supply chain has two single points of failure: NVIDIA's 80% market share in AI accelerators (like H100/H200) and ASML's monopoly on EUV lithography machines. NVIDIA's advantage is compounded by its mature CUDA software ecosystem. #AI #Tech
54
Big AI companies aren't worried about losing safety control, but about losing billing control. When you run AI on your own machine, it's free—no usage data, no sales, no charges. The model becomes a tool you own, not a service you rent. #AI #Tech #Future
32
The EU AI Act goes beyond risk tiers. Focus on these key provisions that impact everyone, not just enterprises. Full episode available on datascienceahome.com, Apple Podcast, and Spotify. #EUAIAct #AIDevelopment
1
34