Building something new. Problems soluble, potential to improve invariant.

Cyberspace
You know how when you ride a rollercoaster with your kids, you grab them tight right before you go down the first drop? Grab them tight, it’s time!
1
82
Watching aging politicians react to AI takeoff will at least be one source of perpetual gallows humor. Like a dog barking at a hurricane.
1
105
People think the vulnerabilities models will exploit are written down in the code. No, even once the software is fully verified, they’ll still find vulnerabilities, they will just be vulnerabilities that exploit the physical world.
OpenAI CEO Sam Altman: "I don't even know a dramatic enough word to use… models can find bugs that are outside of the normal threat model of what we think a hacker can do… hardware-level, insane stuff… I no longer think we're gonna be able to just patch all the bugs." It's over.
3
157
Tbh I thought the cyberpandemic would have happened sooner and would have been worse. We will see if it gets worse. I can still use the Internet.
Replying to @hinapansari
yes. it was always going to be this way. we tried to warn the world.
1
3
265
Thanks a lot, whale. <grateful> Thanks a lot, whale. <resentful>
2
204
As a software engineer I’ve always tried to avoid siloing. I’ve worked on search, ML, ops, frontend, web, databases, graphics, game engines, embedded, etc. It’s hard to overstate what kind of software I can build and deliver now with AI by drawing on this breadth simultaneously.
I agree with everything OP says except "watching all of those years go down the drain." AI might have made it almost meaningless to write code by hand anymore. But the people who have spent years/decades writing code are now in the best position to leverage AI coding tools. Learning to code was a stepping stone toward learning to decompose and solve complex problems. I have worked with a lot of different people with different levels of coding experience. And I can say with confidence that, in general, the more coding experience you have, the better you're positioned to think about long-horizon tasks and to organize and tackle problems with multiple moving parts. All those skills directly translate to using AI coding agents. And I'm seeing it happen in real time. People with long software engineering experience have a totally different way of using AI coding agents (given that they set their ego aside and accept that this is the new norm). They are more organized, know how to steer the agent like guiding an intern or mentoring a junior dev. People who don't have coding experience are using agents in a random and chaotic way (the approach has its outside-the-box-thinking merits but doesn't have the same reliability as the experienced SWE approach).
3
1
23
619
The great irony is I wouldn’t consider myself an expert in any of these domains. More of a middle grade engineer with enough knowledge to be dangerous. But lo and behold, that margin of technical expertise is now in reach of being automated, and cross domain knowledge is key.
3
160
Putting AI compute in space, like Bitcoin, is poorly understood by Americans because it’s a wartime technology, and so they don’t see the point. But the point of wartime technologies become clear when there is a war.
1
12
604
We are in this really awkward stage where AI genuinely undermines human cognition. The thing people are missing is this is transitory. You can find it disgusting and say you’ll never do it, but the fact is we are all getting silicon exocortexes. There is no alternative.
there's evidence that the use of AI in schools is delivering the opposite of "intellectual flourishing"—maybe in some bubbles the use of these models is leading to more reading and philosophical debates but I'm inclined to think that's an exception to the rule
2
2
24
1,094
gfodor.id retweeted
这篇论文有点意思啊,它证明注意力层和一个两层的MLP数学上等价,并且作为即时生成fast weight,然后宽度就是上下文长度,Q,K,V都分别对应着输入,第一层MLP的权重和第二层MLP的权重。softmax就是激活函数。 这一下就解释的通了,毕竟注意力机制这个矩阵看起来就不像是人脑里面能长出来的东西,看成MLP就合理多了。同时,这也说明了transformer里面其实就有现在持续学习里面的比较重要的fast weight路线。这么说的话,zero shot的能力其实也能解释的通了,解释为临时生成的MLP的能力,原来的一层attention一层MLP就是快慢权重的结合 更重要的是,现在的各种优化attention层的方法也可以看成是路由和压缩,类似linear attention就是MLP的压缩,而DSA就是路由,moe。 感觉未来也许我们能找到某个简单的方案,把transformer里面那些乱七八糟的东西都换成简单的MLP呢?这样也符合大道至简,符合bitter lesson。attention其实还是有人类的经验在里面的。 arxiv.org/abs/2602.01219
12
48
433
29,217
People dunking on this have a kind of Stockholm Syndrome from the difficulty of using computers. Being able to just say what you want or need and have it happen was always the dream of the Computer Revolution. The original pioneers would be elated the difficulty has melted away.
My 12 year old son just made this @Tesla Roadster game from scratch by himself using AI. AI is nuts.
33
60
1,679
27,081
It’s probably just the algorithm but the anti-AI posts have reached a kind of fever pitched denialism I don’t remember before. Just repeating mantras that are trivially disproven by using the latest systems for five minutes. This is the real AI psychosis.
34
31
789
12,193
I’ve seen things in Vegas that make me think that cert was just an unlucky draw
1
6
1,065
gfodor.id retweeted
-----BEGIN RSA PRIVATE KEY----- 4 -----END RSA PRIVATE KEY-----
248
3,990
42,203
1,075,614
Cyberpandemic
2
2
73
2,066
Anyone who has fallen in and out of love with PHP knew this day would one day come. May the odds be ever in your favor 🫡
Effective Immediately, all organizations using PHP for their systems are advised to be prepared for an update addressing two critical vulnerabilities. PHP is widely used for web applications. These vulnerabilities often provide a direct path to the server's file system or database. If an immediate update is not feasible, it is recommend that organizations take the system offline until it can be fully mitigated. NSA will follow up shortly with more specific guidance.
1
1
42
2,772
A point of view is worth 80 IQ points. A point of view with an unlimited token budget is something else altogether.
9
1,050
It’s true. Opus 5.5 is actually good. There is no emoji reality distortion field. It’s just good.
now that opus 5.5 is good at code, i can openly like claude without some annoying Codex nerds judging me for being an airheaded bimbo, which I am for unrelated reasons.
1
1
82
2,361
Birthright citizenship
October 20, 2027 Optimus Assembly Line
3
26
1,212