Hacker Newsnew | past | comments | ask | show | jobs | submit | zer00eyz's commentslogin

> Typing 'Find me reservations at X restaurant' and getting unformatted text back is way worse than just going to OpenTable and seeing a UI that has been honed for decades.

Your conflating the example with the opportunity:

"Cancel Service XXX" where the service is riddled with dark patterns. Giving every one an "assistant" that can do this is a game changer. This is why a lot of people who aren't that deep in tech think open claw is interesting.

> We all learned the lesson that mass-market IT tools almost always outperform in-house

Do they? Because I know a lot of people who have (as an example) terrible setups with sales force that they have to use.


Admiral Grace Hopper is famous for using a length of wire to explain to others what a nanosecond was.

https://www.pbs.org/newshour/world/pentagon-embraces-musks-g...

Data centers in space make absolute sense when you want as close to real time analysis on all sorts of information. Would you rather have it make the round trip, via satellite to the states? Or are you going to build these things on the ground near a battlefield?

Musk is selling a vision for a MASSIVE government contract to provide a service that no one else could hope to achieve. This is one of those projects where he can run up the budget and operating costs like Boeing, Northrup etc, because it has massive military applications.


> Consumer can eat all the GPUs they have and more if we stop trying to force B2B

You should really crunch the numbers on buying and then running enough compute to run a leading edge model. The economics of buying it (never mind running it) just dont add up.

You still haven't factored in "training", the major problem right now that every one remains head in sand about.

I dont need a model to know who Tom Cruise is or how to write SQL if I am asking it "set up my amazon refund" or "cancel xyz service". The moment someone figures out how to build targeted and small it will take off.

And as for training, well having to make ongoing investment into re-training is what killed expert systems, it's what killed all past AI efforts. Just because it's much more "automated" doesn't mean it isnt the same "problem". Till a model learns (and can become a useful digital twin) the consumer market is going to remain "out of reach".

That doesn't mean we dont have an amazing tool at hand, because we do. But the way it's being sold is only going to lead to confusion and disappointment.


Consumer, as in B2C, not consumers buying directly. B2C companies will happily buy (or rent from people who are buying today) GPUs, because a huge part of the game is managing margins to a degree B2B typically doesn't need to concern itself with.

> I dont need a model to know who Tom Cruise is or how to write SQL if I am asking it "set up my amazon refund" or "cancel xyz service". The moment someone figures out how to build targeted and small it will take off.

I think people got a lot of ideas when dense models were in vogue that don't hold up today. Kimi K2.5 maybe be a "1T parameter model" but it only has 32B active parameters and still easily trounces any prior dense model, including Llama 405B...

Small models need to make sense in terms of actual UX since beating these higher sparsity MoEs on raw efficiency is harder than people realize.


> We used tech without letting it own us.

This was, and is, a personal choice.

You can call me a judas goat if you want for spending my whole life building the things I would not use, but that is the nature of the game.


For some context:

Nvidia: GAAP and non-GAAP gross margins are expected to be 74.8% and 75.0%,

Micron: MU (Micron Technology) Gross Margin % as of today (January 28, 2026) is 56.04%


"Abominable Intelligence"

I cant wait till the church starts tithing us mear flesh bags for forgiveness in the face of Roko's Basilisk.


> Current LLMs/ViTs have non-zero probability of producing something unsafe

Show me on the LLM where the "unsafe" result is that doesn't already exist inside its training data set? Go ahead I will wait.


https://en.wikipedia.org/wiki/Room_641A ... Then, years later every one acts like Snowden had some big reveal.

There is the old password for candy bar study: https://blog.tmb.co.uk/passwords-for-chocolate

Do users care? I would posit that the bulk of them do not, because they just dont see how it applies to them, till they run into some type of problem.


> Sometimes a law is just on its face and unjust in its application.

There is another side to this coin: jury nullification.

The fact that, most Americans, are unaware of the concept, or that it is a choice they can make is one of the tragedies of the modern era. Adams had much to say on the topic, and his take is still valid 200 years later.


I think that when you put most Americans in a jury box, they will learn that whether they vote to convict is their choice. Ask the guy who threw the sandwich.

https://en.wikipedia.org/wiki/Trial_of_Sean_Dunn


> Or rich people owning more vacation homes.

https://fred.stlouisfed.org/series/RSAHORUSQ156S

Home ownership rates have a 6 percent variance over the last ~50 years.

We dont have a housing problem in America, we have a utilization problem:

https://www.census.gov/library/stories/2023/06/more-than-a-q... as an example.

There is a conversation that needs to be had about housing, but no one is going to LIKE the medicine that comes with that.


There's some historical stuff happening in that graph that it's easy for young people to not have context for, like the fact that the peak home ownership around 2005 was caused by a subprime mortgage fiasco.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: