17,118 karma · joined February 10, 2015
I’d imagine there’s only a tiny percentage of users that can or want to do this.
$20B is a massively successful company.
And who’s managing the strategy behind a company that’s competing in hundreds of diversified verticals?
At that point, just take a % of an industry leader in exchange for giving them your frontier AI capabilities, some compute and maybe a little AI-native training/expertise.
Or just pick 10-20 $200B opportunities to go after.
They’re chasing massive markets strategically, not trying to brute force small ones.
Sure, but a reasonable user would expect that condition to be implicit in using a commercial model.
It’s what the safety/alignment discussion is about: how we can get models to actually comply with laws & ethics.
But the same problem exists in the real world. If you tell someone to “Try everything possible to get it done, I don’t care what it takes”, I’d imagine that people would be far more likely to bend/break rules than if you said “could you please help me with… ?”
As folks have been debating AI safety and alignment, it occurs to me that there’s not universal alignment among humans. Per the stat you surfaced: 81% of humans pulled the lever, 19% didn’t.
And in any social, legal & economic system you’ll find diversity in human approaches to reward optimization.
Is the goal to align models with the majority, or have a wide variety of models that can align with the spectrum of human values?
Seems like a solvable problem though by generating synthetic data that’s guaranteed to be accurate through linters, compilers and tests.
If it’s ~9 figures to train a frontier model, it seems like training on a new language could be a rounding error if it was a priority.
I could see the appeal of a new agentic-friendly language that’s focused on minimizing tokens. The standard library could be massive with no concern of making the language easily readable or learnable.
If you look at the hiring marketplace, being just marginally better than your peers can be very lucrative.
If you’re competing on speed or capability as a company (or as an employee), you’re probably going to be willing to pay for the frontier.
In both scenarios, it really depends on how many alternative options for income you have available.
Just like someone in customer service might act differently with coworkers vs customers.
It’s the concept of a management chart as an inverted pyramid with each layer holding up and supporting the layer above them. If you imagine a promotion as working your way down the corporate pyramid, then it’s easier to see how the managers at the bottom are carrying more weight and deserving of higher pay.
As opposed to a pyramid where it’s visually represented as the broader management layers supporting the layers above them.
In a pyramid, it looks like the CEO has a cushy, overpaid job. In an inverted pyramid it looks like they have the weight and responsibility of the company on their shoulders.
Time slowing down does seem useful in the event you can actually affect your circumstances.
With Warp, you purchase your AI compute through Warp (who then pays Anthropic, Open AI, etc. based on the model you choose).
Here’s where I got it from, but I see how it’s ambiguous. “You pay for what you use” sounds a bit like the BYOK (bring your own key) “add-on credits” pricing model you’re referring to.
But in the pricing table, they refer to monthly “AI credits”.
You’re really paying for AI compute, not the terminal.
The new Build plan at $20/mo has 1,500 AI requests, but they roll over. (Edit: apparently they don’t)
> No bones about it: this plan will be more expensive for some users and less expensive for others.
> We get that there’s a lot of whiplash in the AI devtools pricing market, and sympathize. While we expect some churn from this change, we are trying to do it in as minimally disruptive a way as possible.
I’ve found Warp to be very useful, but you’re really paying for AI compute, not the terminal. And the AI compute space is getting very competitive.
* The casino takes a rake, so you lose money every hand, but you only win when the fish bets and loses. You’re also expected to tip the dealer
* Everything is on camera and dealers remember players, so there will be a lot of witnesses and evidence
* Seats often open one at a time, so you’d potentially lose money at other tables waiting to play together. Or, you all show up at once and ask to start a new table together, which would get suspicious.
* If you don’t know the fish’s cards, there’s still a chance you lose and lose big
That seems like a human-centric perspective.
Maybe they’re a cooperative, altruistic society with an innate desire to help, and maybe had been helped by others before. To not teach us about the imminent dangers of the universe might seem unconscionable to them.
Or maybe they’re a highly ordered society with an innate common goal and see nothing wrong with asking other entities to join their mission.
Sure, some humans may view their contact as intrusive or harmful, but that doesn’t mean they automatically would as well.
If I had to bet, I’d bet you’re right, but the universe is a big place and who knows what societies might be out there that would feel totally foreign to us.
I like the trend where the agent decides what models, tooling and thought process to use. That seems to me far more powerful than asking users to create solutions for each discreet problem space.
As a long-time user I’ve seen the most change around “What to Submit”:
> Off-Topic: Most stories about politics
> If they'd cover it on TV news, it's probably off-topic
The guidelines haven’t changed but it feels like the enforcement of it has.
For example, the US government shutdown is currently on the HN top 20: https://news.ycombinator.com/item?id=45434146
While a mainstream newsworthy story, I fail to see how it “gratifies one's intellectual curiosity” or is “Anything that good hackers would find interesting.”
Most subagent examples are vague or simplistic.
Two observations:
1. Natural language appears to be to be the starting point of any endeavor.
2.
> It may be illuminating to try to imagine what would have happened if, right from the start our native tongue would have been the only vehicle for the input into and the output from our information processing equipment. My considered guess is that history would, in a sense, have repeated itself, and that computer science would consist mainly of the indeed black art how to bootstrap from there to a sufficiently well-defined formal system. We would need all the intellect in the world to get the interface narrow enough to be usable, and, in view of the history of mankind, it may not be overly pessimistic to guess that to do the job well enough would require again a few thousand years.
LLMs are trying to replicate all of the intellect in the world.
I’m curious if the author would consider that these lofty caveats may be more plausible today than they were when the text was written.
With a mass market product leader you’re sacrificing a bit of customization for long-term stability.