Hacker News new | past | comments | ask | show | jobs | submit | ip26's comments login

The people selling models are actually just selling compute

Yes, fully agreed. Anything AI is discovering in your dataset could have been found by humans, and it could have been done by a more efficient program. But that would require humans to carefully study it and write the program. AI lets you skip the novel analysis of the data and writing custom programs by using a generalizable program that solves those steps for you by expending far more compute.

I see it as, AI could remove the most basic obstacle preventing us from applying compute to vast swathes of problems- and that’s the need to write a unique program for the problem at hand.


The "good old boys" network is a problem. But given how hard we all agree it is to interview effectively and determine who is a great fit for the role in a matter of a few hours, there's a lot of good sense in hiring people already widely known to be excellent by your team from years of past experience working together.

There’s tension between what is best for the company and what is most fair to applicants. I acknowledge that, but think that the onus should be on (large) companies to figure out a better interview process.

I don’t see why references have to come from current (or past) employees. Colleges don’t make you get referred by alumni, but they do require letters of reference (usually).

On a related note, it’s amusing to me when white men in tech on Reddit get mad about Indian men preferentially hiring other Indian men from their community. I assume that many of these same white men don’t see any problem when they preferentially hire their own friends using the rationale that you gave.


The mountains are loaded with metaphor, seem nearly insurmountable, filled with danger and excitement…

Turns out, of course, that they are less dangerous than in the past, money goes a long way, and the human body is pretty good at walking even without training.

Notice it’s never free climbing the nose in a day, which would require 5-7 years of dedicated training.


>Notice it’s never free climbing the nose in a day, which would require 5-7 years of dedicated training.

Would take way more than that for the average person. You can count the number of people who have done that as of now on one hand.


One did a lot of reasonable little stuff but never did anything with much impact again

This always stymies me. If I were to make it big, I’d want to go on to do other very useful things. But so commonly that doesn’t happen. Is it because they lost the fire? Because they were really lucky exactly once? Or perhaps their talents were suited only to that one thing?


That's always a good question. Were you smart, or were you lucky?

This is most often a problem with traders and investors who made a good bet. Less so with something that took a lot of effort to make work.

I know a married couple who blew through an amount in eight figures and had to get low level jobs. Lost their house.


Maybe somewhat less, but effort doesn't really imply value and value doesn't imply successful extraction.

Skill, luck, timing.

Few people achieve something big. Even fewer can achieve another big thing. Even less can continue achieving.


Personally I get immense enjoyment out of not doing anything useful, especially not on a big scale. So if I were to accidentally "make it big", I'd most likely not do it again and keep enjoying the small inconsequential things.

As a founder, the people relying on him would have been the employees at Loom. But now that’s done. Far from the first story about a founder feeling unmoored after a buyout.

artificially produced gaps in ancient texts

Someone deleted part of a known text.

This does require the AI hasn’t been trained on the test text previously..


They do mention that the missing data test was done on "new" data that the models were not viewed trained on in the article so it's not just regurgitation for at least some of the results it seems.

It does seem like an improvement. Six or twelve months ago, I recall a lot of crashes and even more basic problems. “If you tune it right, it’s awesome” is a big step forward compared to that.


people just want to travel, party, and drink instead of having kids

That’s something of a self-correcting problem, don’t you think?


Eventually yes, but it would be nice to avoid the Dark Ages this time. The crash will be brutal.


Is it? If it was, it would be fixed already. Parties didn't start in the 2000s.


Wait, are you suggesting they SEO'd the article on how to remove SEO filler? I'm sure that can't be true.


I see it more like, there are a lot of relationships balanced by conflict, such as Adsense advertiser and target audience. The advertiser wants you to buy product, the target audience wants to get free content. Both sides will exploit every tool at their disposal (Adsense metrics, adblockers) to get an upper hand in the relationship. It just so happens that this time, AI can play both sides.


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

Search: