veunes

11 ms·

181.
by veunes 9mo ago
The analogy is great, but it breaks down when we talk about professional use. An aimbot in multiplayer is evil because it ruins the game for others. But an aimbot in game development (e.g., procedural aiming animation) is just a tool. The p
182.
by veunes 9mo ago
The problem runs deeper: AI doesn't just remove execution, it replaces it with probabilistic averaging. Mastery of execution in film or code consists of thousands of micro-decisions, like light a bit to the left or pause a bit longer.
183.
by veunes 9mo ago
The author nailed the problem but missed the technical root cause: lack of controllability. Current models (Sora, Veo, Runway) are probabilistic generators. They are optimized to output a plausible image, not the specific image a director n
184.
by veunes 10mo ago
Point 1 (document everything) is the utopia that killed the project. In any complex system, documentation is a lossy compression of reality. The actual truth about how to fix bugs doesn't live in Confluence; it lives in senior heads, S
185.
by veunes 10mo ago
The problem goes deeper: verification is harder than generation. When writing an answer yourself, you build the logic chain from scratch. When verifying AI, you have to deconstruct its logic, cross-reference facts, spot hidden hallucination
186.
by veunes 10mo ago
The idea of formal verification works great for code or math where clear rules exist, but in customer support, there is no formal specification. You can't write a unit test for empathy or for "did we correctly understand that the
187.
by veunes 10mo ago
They shouldn't have tried to force LLMs into doing something current models aren't designed for: semantic understanding of "unknown unknowns". Tier-2/3 support isn't just about picking an answer from a knowledg
188.
by veunes 10mo ago
If you ever open-source it, I suspect a lot of musicians who don't think of themselves as programmers would still find it surprisingly approachable
189.
by veunes 10mo ago
It's kind of nice that both modes coexist
190.
by veunes 10mo ago
Max is basically a programming language already: message passing, scheduling, state, abstraction, even metaprogramming
191.
by veunes 10mo ago
Csound is funny because it's everywhere in the lineage, but relatively few people seem to arrive at it organically now
192.
by veunes 10mo ago
This list is such a time capsule in the best way
193.
by veunes 10mo ago
There's definitely a performative vibe to a lot of it right now
194.
by veunes 10mo ago
The space is moving fast enough that everything feels provisional
195.
by veunes 10mo ago
If skills really can move across platforms, that's a meaningful push against lock-in, at least in theory
196.
by veunes 10mo ago
If a quick start with AI is inevitable, then mentorship and review programs need to be re-evaluated. Seniors shouldn't just check for functionality; they should actively ask juniors to explain why the AI suggested a particular solution
197.
by veunes 10mo ago
Understanding "why it works" is one thing, understanding "why it should work this way and not another, and what the alternatives are" is entirely different. AI shows you just one of countless correct implementations. You
198.
by veunes 10mo ago
I agree, the routine is gone, but at what cost? Understanding "how our system fits together" means solving problems, reading code, and debugging. If those fundamental skills aren't built through "humiliating and boring&q
199.
by veunes 10mo ago
The main problem Garman overlooks is skill degradation. AI is excellent at helping a junior quickly draft boilerplate or find the right API, but it doesn't teach the essentials: debugging, systems thinking, and reading complex code. A
200.
by veunes 10mo ago
Cool project. Would be interesting to see how this changes if someone runs the same analysis on arXiv or other open datasets.
201.
by veunes 10mo ago
Yep, if robots really become that cheap and widespread, it could dramatically shift the balance of power in society
202.
by veunes 10mo ago
I think the current debate is less about whether automation happens and more about where the gains go
203.
by veunes 10mo ago
The end game you're describing (abundance with minimal labor) only works socially if we deliberately redesign distribution, not just production
204.
by veunes 10mo ago
If AI really is just another capital amplifier, then the problem isn't new, AI just makes the existing imbalance harder to ignore
205.
by veunes 10mo ago
Taxes are paid by people and legal entities, not tools. The real issue isn't that machines aren't taxed, it's that our tax base is still heavily tied to labor
206.
by veunes 10mo ago
That's fair for brute force (running a model on the GPU), but that's exactly where NPUs come in - they are orders of magnitude more energy-efficient for matrix operations than GPUs. Apple has been putting NPUs in every chip for ye
207.
by veunes 10mo ago
That's true for current LLMs, but Apple is playing the long game. First, they are masters of quantization optimization (their 3-4 bit models perform surprisingly well). Second, Unified Memory is a cheat code. Even 8GB on M1/M2 a
208.
by veunes 10mo ago
That is exactly why Apple's on-device strategy is the only economically viable one. If every Siri request cost $0.01 for cloud inference, Apple would go bankrupt in a month. But if inference happens on the Neural Engine on the user&#x2
209.
by veunes 10mo ago
Sooner or later, everyone will realize Apple isn't building another ChatGPT - they don't need to. They're working on the world's largest distributed inference network. With hundreds of millions of Apple Silicon devices,
210.
by veunes 10mo ago
This is the tension that makes the whole situation so hard to fix

More ›