visarga

8 ms·

151.
by visarga 3mo ago
Meditation can also hype up your anxiety depending on individual circumstances and personality. And it is by no means easy, you have to invest much effort to get results. It's like refactoring your code base (mind).
152.
by visarga 3mo ago
Consciousness is what the body is doing to be viable. Without it we can't walk, eat, reproduce, or do anything. I like to think cost viability reasons explain consciousness. I know people prefer metaphysical or quantum magic explanatio
153.
by visarga 3mo ago
> How do you create that which is not definable or measurable? animals can have offspring without understanding reproduction
154.
by visarga 3mo ago
I thought it was going to be even shorter leash - code autocomplete with smaller local models. That raises the level of interactivity and leads to better code knowledge.
155.
by visarga 3mo ago
I think the obvious solution here is to beef up the test side of the app, much more than when writing code by hand. Tests represent project knowledge in executable format. The LLM does not need to be careful to remember every detail of the
156.
by visarga 3mo ago
Try thocky keys, Aula F75
157.
by visarga 3mo ago
We might quibble what to call it, but they use language productively.
158.
by visarga 3mo ago
> When you speak, what comes first, the idea or the word? Do you first feel a thought inside you, and only after that go searching for the right word to wrap around it? Let's take a feral human, do they achieve the same with languag
159.
by visarga 3mo ago
Plain old markdown with links is exactly what I use. But I split memory in 3 parts 1. a log of user comments, all of them collected in a chat_log.md file which is used to validate the agent still follows user intent 2. task level memory, a
160.
by visarga 3mo ago
> Very different things – an electric guitar doesn’t play music by itself. And a model does not play good music by itself. Only slop if your contribution is nil. Models are more like pianos than parrots.
161.
by visarga 3mo ago
Even if you generate the soundtrack, if you set the lyrics it can be great music. At least for one person, who chose the lyrics. I set some poems on music and I absolutely love the results.
162.
by visarga 4mo ago
I think he is busy making sure AI doesn't finish it first. Can't have AIs trample in his fantasy land.
163.
by visarga 4mo ago
Macbook M5 64GB - can run gemma-4-26b-a4b-it-4bit and Qwen3.6-35B-A3B-4bit at about 1500 tps prefix and 45 tps decode on contexts up to 100K tokens using MLX. It's faster than Claude. I was really surprised, chat quality is also simila
164.
by visarga 4mo ago
No, they do RLVR (reinforcement learning with verifiable rewards) like everyone else. And probably use claude data too, with human in the loop and tool feedback.
165.
by visarga 4mo ago
When you export your personal data Google hides all model responses leaving just user messages. So it's even worse
166.
by visarga 4mo ago
Yes, AI gives me superpowers but my competitors get it too. True, we are in a race to re-differentiate now. Companies too, imitation of anything we put out is approaching a very low cost.
167.
by visarga 4mo ago
But once you put something out, an app, repo, content, music or photo - it can be easily cloned with AI, changed to be different just enough to avoid accusations and customized as needed. The price of replication is dropping fast, and that
168.
by visarga 4mo ago
I have my own take on what is ours and can never be taken away by AI. When a task is initiated, it starts from a need, from a specific context. To work it out the AI needs to continuously interact with the context, and get feedback from it.
169.
by visarga 4mo ago
It doesn't reflect itself, we only see the UI of a complex process, not the real thing. We don't understand what happened in our brains any better for being able to feel conscious. We can only be conscious of what is cost effectiv
170.
by visarga 4mo ago
It's even worse when everyone around you is using it. How can you keep up? Companies face the same dilemma: investors, competitors, and users already use AI and have factored it into their expectations.
171.
by visarga 4mo ago
It's not unmaintainable if most of it is tests. Just have it write tests until it becomes safe for AI.
172.
by visarga 4mo ago
The relation is not qualia at the base and language on top, even if qualia is more primitive, because language directs action and action leads back to qualia, so they form a recursive loop which cannot be analyzed component by component any
173.
by visarga 4mo ago
I think we have a pretty good explanation today - it's like embeddings from AI models. Experience is both content and reference, we represent new experience in relation to old experience. That makes representation personal, being made
174.
by visarga 4mo ago
The ad hoc heuristics are the domain knowledge baked into the model by human experts, like features, architecture and loss function. "Evaluation" means environments or datasets, the model is supposed to discover its representation
175.
by visarga 4mo ago
> If anyone knows of any work towards more open-ended fitness functions, I'd love to read it. There is research in open-ended learning, see "Why Greatness Cannot Be Planned" by Kenneth O. Stanley. The core idea is that in
176.
by visarga 4mo ago
> There are more elements to discovery though. It is still not clear where the initial working model/hypothesis comes from or how the updates are selected That is a problem in RL, so we usually do supervised training first, teach it
177.
by visarga 4mo ago
Before they add AI they better fix the frigging search function in settings, it is horrible, you need to know their exact words, and Apple has a funny naming sense. Hierarchies nested so deep you never find anything. I come to use Claude or
178.
by visarga 4mo ago
> We are going to get near instant software from prompt, multiple ones and then choose the best one. If you extract the spec from first implementation and reimplement from scratch you get a free testing oracle. Where they diverge you sen
179.
by visarga 4mo ago
> All my finance and payment domain expertise, all the debugging intuition and distributed system knowledge earned through hours of sweat and tears, is now promptable. In ML it was even worse, we had to throw away a decade of experience,
180.
by visarga 4mo ago
I just use md files to plan questions, track my answers, and implement rehearsals for concepts that need more repetition from claude code. And I start from a good book or documentation as source material, first the agent reads the learning

More ›