nl

46 ms·

901.
by nl 11mo ago
You realize that AI is driving huge advertising growth at Meta, right? > Meta, the parent company of Facebook and Instagram, reported strong second-quarter 2025 earnings, driven primarily by robust advertising revenue growth. Total rev
902.
by nl 11mo ago
Humans aren't smart, they are really just good at appearing to be smart. Prove me wrong.
903.
by nl 11mo ago
Well that's not true - see the Terry Tao article using AlphaEvolve to discover new proofs. Additionally, "novel ideas" isn't something that is included in something that smart people do so why would it be a requirement f
904.
by nl 11mo ago
Is IntelliCode the same as Intelisense (the non-AI based suggestions thing)?
905.
by nl 11mo ago
If this is surprising to you then your circle is fairly unusual. For example HBR recently reported the number 1 use for ChatGPT is "Therapy/companionship" https://archive.is/Y76c5
906.
by nl 11mo ago
Virtual offices have been around forever and aren't really an indication of being shady necessarily.
907.
by nl 11mo ago
I thought the journal Nature was doing some decentrailized publishing thing.
908.
by nl 11mo ago
CLAUDE.md is the correct file for Claude.
909.
by nl 11mo ago
Also `CLAUDE.md` (which is `AGENTS.md` everywhere? else now)
910.
by nl 11mo ago
Fraktur actually does use a partially different alphabet. For example it uses the Long s: https://en.wikipedia.org/wiki/Long_s and Half-r: https://en.wikipedia.org/wiki/R_rotunda
911.
by nl 11mo ago
I rather assumed so as well, but a big of digging turns up a whole history: https://en.wikipedia.org/wiki/Antiqua%E2%80%93Fraktur_disput... Surprisingly to me the Fraktur typeface was the traditional "German"
912.
by nl 11mo ago
Does anyone remember the Yahoo design patterns library? It was mostly for UX pattern (eg: ways to "Rate an object") and it was really good. Almost 20 years ago.. damn. https://creativecommons.org/2006/02/
913.
by nl 11mo ago
> I'm using it Do you mean Antigravity or Gemini? If you mean Antigravity then.. how? Their docs say you can't do this. If you mean Gemini then I personally haven't had issues but haven't tried to productionize a Gemi
914.
by nl 11mo ago
You literally cannot buy Antigravity with a non-personal Google account. I read someone on here who is using Gemini via OpenRouter because it was the only way they could pay for it.
915.
by nl 11mo ago
My comment is in the context of: > That USA didn't nuke China in 1950 or 1951. Would have solved a lot of problems for generations of people. > USA dropping nukes probably would have been the better outcome for humanity. Both of
916.
by nl 11mo ago
Worth pointing out that South Korea had very limited democracy until the late 1980s: https://en.wikipedia.org/wiki/History_of_South_Korea
917.
by nl 11mo ago
It's different almost by definition? Because it was a once (twice!) off the impact and significance of it is amplified.
918.
by nl 11mo ago
Wait - you think the solution to some people having a lower standard of living and others being persecuted is to kill them all?
919.
by nl 11mo ago
In what way? The US nearly lost the Korean war. The US army was nearly overrun at least once. The US airforce never achieved air superiority, and Soviet aircraft were better in most ways. The only undisputed advantage the US had was nukes,
920.
by nl 11mo ago
This is simply not true. Modern LLMs are trained by reinforcement learning where they try to solve a coding problem and receive a reward if it succeeds. Data Processing Inequalities (from your link) aren't relevant: the model is learni
921.
by nl 11mo ago
I mean you'd think so, but... > In fact, the UL2 20B model (at Google) was trained by leaving the job running accidentally for a month. https://www.yitay.net/blog/training-great-llms-entirely-from...
922.
by nl 11mo ago
The reason Palantir is valued as highly is because it is a Peter Thiel company, and he also owns the Vice President of the United States. The thesis is "The current US administration is making enormous profits from their powerful posit
923.
by nl 11mo ago
To add to this, this type of thing happens all the time in crypto. A coin will release 1/1000000000th of it's eventual supply, have some trades at 10c and then claim the value of the entire supply as the headline value. It's
924.
by nl 11mo ago
Llama 4 wasn't great, but Llama 3 was. Do we all forget how bad GPT 4.5 was? OpenAI got out of that mess with some miraculous post-training efforts on their older GPT-4o model. But in a different timeline we are all talking about how g
925.
by nl 11mo ago
Pot, Kettle, meet black. "some elements of the indictment concern cyber-snooping in connection with trade disputes, which at least sounds a lot like the kind of cyber-snooping on firms that the United States does." https:/&#
926.
by nl 11mo ago
> Large-scale exfiltration of data from ChatGPT when DeepSeek was being developed, and which Microsoft linked to DeepSeek This is not the same thing at all. Current legal doctrine is that ChatGPT output is not copyrightable, so at most D
927.
by nl 11mo ago
If you haven't tried these things since Sonnet 4.5 came out then it's time to give them another try. Sonnet 4.5 and especially Codex 5.1 have completely changed the way I build software. > The fact that tools like Replit also i
928.
by nl 11mo ago
Have you tried Loveable, Replit, V0 etc? Outside of purchasing the domain and native apps for you they cover a very significant amount of this. If you insist on Native Apps, it's possible Google Jules could do it. With Gemini 2.5 it wa
929.
by nl 11mo ago
> But the end state in my mind is telling an AI "build me XYZ", having it ask all the important questions over the course of a 30-minute chat while making reasonable decisions on all lower-level issues, then waking up the next
930.
by nl 11mo ago
Semi-analysis also believes they haven't done a fill pretraining run since 4o (except for GPT-4.5): https://open.substack.com/pub/semianalysis/p/tpuv7-google-ta...

More ›