nilkn

8 ms·

91.
by nilkn 8mo ago
Pangram itself looks like it was just generated by Google AI Studio.
92.
by nilkn 8mo ago
I'm very curious about this as well. This is the main thing that has held me back from a meaningful rebalancing. Eating a huge tax bill to avoid a theoretical future loss of unknown size and duration while also losing out on potential
93.
by nilkn 8mo ago
"Signaling" is just the information that your visible choices send to those around you, including strangers. That's why it's called "signaling" -- your choices are broadcasting an information signal about you t
94.
by nilkn 8mo ago
This comment is filled with speculation which I think is mostly unfounded and unnecessarily negative in its orientation. Let's take the safety point. Yes, OpenClaw is infamously not exactly safe. Your interpretation is that, by hiring
95.
by nilkn 8mo ago
My language was sloppy, but I'll admit I thought it was pretty obvious that we were talking about defining the rest of the reals starting from the rationals -- obvious enough that it didn't need clarification. I can't edit
96.
by nilkn 8mo ago
Yes. Someone making one contribution among many to a paper clearly does not deserve anything like sole authorship credit of the entire paper, which is what the title from OpenAI implies to me. I don't believe I'm being pedantic at
97.
by nilkn 8mo ago
Your statements are a couple months out of date. The space is evolving rapidly. It's definitely not a cost-efficient approach today, but models like Kimi K2.5 can be run on dual 512GB Mac Studios with performance rivaling (though still
98.
by nilkn 8mo ago
It would be more accurate to say that humans using GPT-5.2 derived a new result in theoretical physics (or, if you're being generous, humans and GPT-5.2 together derived a new result). The title makes it sound like GPT-5.2 produced a c
99.
by nilkn 8mo ago
My take is that the future will look something like this: - You pay for a personal AI assistant from a cloud vendor (most people) or you run it yourself on your own hardware (not yet common, likely somewhat common in the future as hardware
100.
by nilkn 8mo ago
Can you not already do nearly that right now with Gemini on any device, iOS or not? I just gave a similar prompt to Gemini. It activated Gemini's "personal intelligence" feature and gave me the kind of highly personalized iti
101.
by nilkn 8mo ago
I was talking about that, specifically, the relative difficulty of defining reals from rationals vs complex numbers from reals. You replied to me. :) Moreover, I disagree that you have imagined real numbers. I don’t think you’ve imagined
102.
by nilkn 8mo ago
I think you’re underestimating or not understanding why Macs have taken off so much for AI. It has nothing to do with fanboys shilling for Apple. You can get a MBP today with 128GB of unified memory or a Mac Studio with 512GB of unified mem
103.
by nilkn 8mo ago
I think you're assuming that no durable or at-scale changes in compute form factor will occur, so that their success pretty much just solely comes down to differentiated iPhone software features. That seems unlikely to me. I don't
104.
by nilkn 8mo ago
You're not wrong, but you can also get most (or even nearly all) of that right now today on any device by just getting a Google AI subscription. Gemini already does most of that through its own personal intelligence feature. You do nee
105.
by nilkn 8mo ago
With that approach, all anyone has to say is that you'd have to provide infinite information to specify an example and that the way these objects interact is completely undefined; therefore you haven't defined or done anything at
106.
by nilkn 8mo ago
Just my opinion: Apple doesn't need to solve AI. It's not core to their business in the same way that search engines aren't core to their business. What Apple does best lies at the combination of hardware, software, physical
107.
by nilkn 8mo ago
This is not a bad example. Tesla is indeed running a custom LLM, available in their vehicles, capable of acting as a general chatbot and issuing commands to the car, developed in-house. While Grok is not up-to-par with other frontier models
108.
by nilkn 8mo ago
“What I’m taking about” is that they are not easy to conceptualize intuitively. If I were a skeptic of real numbers, I’d tell you that talking about an infinite decimal expansion that never terminated and contains no repeating pattern is
109.
by nilkn 8mo ago
I'm sorry to say this, but the company does not need employees to respond enthusiastically. They'll just replace the people who resist for too long. Employees who resist indefinitely have absolutely zero leverage unless they'
110.
by nilkn 8mo ago
The competitive advantage belongs to companies, not engineers. That's exactly the conflict. What you're predicting -- voluntary adoption due to advantages -- is precisely what is happening, but it's happening at the company l
111.
by nilkn 8mo ago
That's false. Those things were in fact resisted in some cases. For instance, look up the swing riots of the 1830s.
112.
by nilkn 8mo ago
People will voluntarily adopt modest productivity boosters that don't threaten their job security. They will rebel against extraordinary productivity boosters that may make some of their skills obsolete or threaten their career.
113.
by nilkn 8mo ago
> Maybe you were able to cut at 10.000, but if you go more precise you'll start seeing other digits, and they will not be repeating. You just picked a real number. A reasonably defensible inference would be that adding a finite amou
114.
by nilkn 8mo ago
An infinite decimal expansion isn't enough. It has to be an infinite expansion that does not contain a repeating pattern. Naively, this would require an infinite amount of information to specify a single real number in that manner, and
115.
by nilkn 8mo ago
> I believe real numbers to be completely natural You can teach middle school children how to define complex numbers, given real numbers as a starting point. You can't necessarily even teach college students or adults how to define
116.
by nilkn 8mo ago
AI music on a radio station doesn't really make sense. The point is to be able to create custom music tuned to your own tastes -- music that's specifically for you .
117.
by nilkn 8mo ago
It really isn't. Nobody's migrating their payroll vendor to some vibecoded app. I agree with Jensen Huang that this reaction doesn't make sense at all. My own prediction is that reliable vibecoding will be additive. It's
118.
by nilkn 8mo ago
My read: Sam's goal is to create viral products. This has always been his goal. While his interest in and passion for AI are almost certainly genuine, he famously hedged his bets before committing to OpenAI up to the point of being amb
119.
by nilkn 8mo ago
Anthropic was founded by OpenAI defectors who said OpenAI's product strategy was too dangerous and needed more safety research. But in reality Anthropic has almost exactly the same product strategy. A lot of this is just marketing to r
120.
by nilkn 10mo ago
Multiple denied promo applications. Warm, caring language but no attempt at retention on resignation. Other companies unsure of hiring candidate even after 10+ interviews. The simplest explanation of these datapoints is simply that this per

More ›