resters

11 ms·

151.
by resters 1y ago
It's off topic to mention this, but I've become convinced that the richest among us really don't care about improving the systems and institutions that would really elevate humanity and unlock tremendous human capital.
152.
by resters 1y ago
Imagine if we saw mostly headlines like this rather than things like "Trump threatens war on Chicago" or "Supreme Court approves racial profiling", or "Columbia students deported for protesting" or "Trump
153.
by resters 1y ago
If that is the case why create a separate govcloud and HIPAA service?
154.
by resters 1y ago
Not official laws that we can read, but things like that are already in place per the Snowden revelations.
155.
by resters 1y ago
It does to some but not to most of it, which is why Azure and GCP offer nearly the exact same core services.
156.
by resters 1y ago
Once a vector has been generated (and someone has paid for it) it can be searched for and relevant content can be identified without AWS incurring any additional cost to create its own separate censorship-oriented index, etc. AWS can also a
157.
by resters 1y ago
By hosting the vectors themselves, AWS can meta-optimize its cloud based on content characteristics. It may seem like not a major optimization, but at AWS scale it is billions of dollars per year. It also makes it easier for AWS to comply
158.
by resters 1y ago
Makes sense. Stablecoins in the new regulatory landscape offer significant efficiency gains in the provisioning of lots of innovative financial services (and also typical ones). Does Stripe have a perspective on the unique systemic risks th
159.
by resters 1y ago
Demand will shift to AI-literate knowledge workers whose productivity will be increased.
160.
by resters 1y ago
Middleware abstractions are challenging because third party libraries tend to abuse it and that can cause performance issues. Also, Vercel has optimized the middleware for its cloud business which includes various optimizations. I don'
161.
by resters 1y ago
True but morse keys are typically much lower latency!
162.
by resters 1y ago
I got a Begali and oddly I ended up preferring the feel of the plastic finger pieces (also included). I like a really light touch.
163.
by resters 1y ago
less strange ones: https://www.i2rtf.com/keys---paddles.html https://www.n3znkeys.com/ https://www.vibroplex.com/ https://www.westmountainradio.com/product_info.php?products
164.
by resters 1y ago
Consider some of the scaling properties of frontier cloud LLMs: 1) routing: traffic can be routed to smaller, specialized, or quantized models 2) GPU throughput vs latency: both parameters can be tuned and adjusted based on demand. What see
165.
by resters 1y ago
this is a step away from capitalism and it harms any startups destined to compete with Intel and discourages investment in the startup ecosystem that might compete with Intel, and punishes investors/customers in/of any of Intel&#x
166.
by resters 1y ago
It works as intended, it allows the US to pay bribes around the world while also allowing law enforcement to prevent criminal enterprises from competing with the regulated financial services industry.
167.
by resters 1y ago
AI will become a commodity -- see Deepseek. The handwriting is on the wall and the open source models and practices to get to Claude 4.1 or GPT-5 are all in the public domain. Apple could simply sell earpods with an AI voice interface, the
168.
by resters 1y ago
many aspects of business and financial planning are adversely impacted by excessive inflation. Sure, you can theoretically bake inflation hedges into any financial instrument, but a very small percentage of business and financial planning
169.
by resters 1y ago
not sure yet about either of those questions!
170.
by resters 1y ago
GPT-5 was able to fix a variety of bugs in some code that I'd been working on with Claude 4.1 (which Claude 4.1 had made and was not able to fix), and GPT-5-pro was able to offer some high quality critiques of some research I've b
171.
by resters 1y ago
Welcome to the world where Tim Cook has to present gold offerings to the US president and spend most of his time navigating massive new taxes on imports the company relies on.
172.
by resters 1y ago
This is a good and often overlooked point. Ai will be more like domesticated pets, their utility functions tightly coupled to human use cases.
173.
by resters 1y ago
Reading the comments it becomes clear how befuddled many HN participants are about AI. I don't think there has been a technical topic that HN has seemed so dull on in the many years I've been reading HN. This must be an indicati
174.
by resters 1y ago
consider the hardware DOOM runs on. 720p would only be a true test of capability if every bit of possible detail was used.
175.
by resters 1y ago
AI is going to cause a regression to the most anodyne output across many industries. As humans who had to develop analytical skills, writing skills, etc., we struggle to imagine the undeveloped brains of those who come of age in the zero-i
176.
by resters 1y ago
AI will be both a floor and a ceiling raiser, since there is a practical limit to how many domains one person or team can be expert in, and AI does/will have very strong levels of expertise/competency across a large number of doma
177.
by resters 1y ago
My iPad only lasts a few days even with zero use. I have not been able to figure out what settings to modify so that I can (for example) pick it up a month later and not find the battery dead.
178.
by resters 1y ago
Suppose there is an LLM that has a very small context size but reasons extremely well within it. That LLM would be useful for a different set of tasks than an LLM with a massive context that reasons somewhat less effectively. Any dimension
179.
by resters 1y ago
It’s a method of storing information that makes it far more useful than previous methods. Sure we can idealize feats of the human brain such as memorizing digits of pi. LLMs put more human behavior into the same category as memorizing digit
180.
by resters 1y ago
If you consider indirect costs I believe it ends up at $18T or more.

More ›