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

Looks good! I think adding timing grid would be cool, ala 1/4, 1/16 etc

Divided by 40 years it doesn't look like much...

What about just giving your kid $400,000 in index funds when they become an adult? What does that look like after 40 years?

It would be a ton with ~10% annualized returns on the SNP500 since 1985, especially if you reinvest dividends. Someone who lives very frugally can probably just live off the interest off that 400k alone and if they bank unitl a million, they can have a safe buffer and a comfortable lifestyle. If they reinvest all of it for fourty years since 1985 it’d be worth on the order of $10-25 million.

It’s actually quite stark how bad it’s gotten. I’d rather just give a kid that money to invest and have them live at home doing whatever work floats their fancy until they can live comfortably on interest alone.


Just like compounding interest, a VERY small percentage advantage each year snowballs to be a large advantage over a full career.

A college degree in the industry you work in also means your job security is MUCH higher than those without a degree.


$10k a year is a huge amount. The median US salary is $61k.

Also realistically, you're likely getting a loan, so you need to factor in interest.


Agree, and the $61,000 part is much of what this article is missing. Its a lot like "weep for those making $300,000/year" when most families are making less than half of that on average (almost a third, $122k if both family members are pulling average). However, taxes tell us they're not even doing that great.

If your family is reporting a family income of $122k, you're already at approximately the upper 25% threshold. The $61,000 number hides a really L-shaped distribution skewed heavily toward the $2,500,000+ crowd. Previously calculated 2024 statistics from another post.

in the most recent tax filing season data available (2024), there were tax returns of:

                                        Top 1%       Top 5%      Top 10%       Top 25%       Top 50%   Bottom 50%  All Taxpayers
  Number of Returns                  1,535,899    7,679,495   15,358,991    38,397,477    76,794,954   76,794,954    153,589,908
  Average Income Taxes Paid           $653,730     $187,468     $108,251       $50,963       $27,891         $667        $14,279 
  Adjusted Gross Income (Millions)  $3,872,395   $6,182,180   $7,745,525   $10,613,602   $13,191,209   $1,531,038    $14,722,247
If we then break those into the actual groups, and numbers per group, then we find their Average Per Capita Income

                                             1          2-5         6-10        11-25        26-50       51-100
  Number of Returns                  1,535,899    6,143,596    7,679,496   23,038,486   38,397,477   76,794,954
  Income Taxes Paid (Millions)      $1,004,063     $435,594     $222,966     $294,234     $185,068      $51,225 
  Adjusted Gross Income (Millions)  $3,872,395   $2,309,785   $1,563,345   $2,868,077   $2,577,607   $1,531,038 
  Average Tax Rate                       25.9%        18.9%        14.3%        10.3%         7.2%         3.3%
  Average Per Capita Income      $2,521,256.28  $375,966.29  $203,573.91  $124,490.69   $67,129.59   $19,936.70

Not sure why you're downvoted as this is true.

(and there's nothing wrong with you expressing this opinion)

Not everyone in the US makes 500k TC doing remote work.

400k for a degree is absolute bonkers!


To pay $400k off over even 20 years at 6%, which is below the average student loan rate is just over $2,000 a month and it is mostly not tax free. Our society is messing up with how expensive this has gotten. The bloat is facilities and administrators who earn two and three or more times the tenured professor salary. The guarantors have to be the schools themselves and those execs who take these bloated salaries. And that interest isn't deductable other than a pittance of $2,500 or less in a year.

Here is the IRS about this;

Student loan interest is interest you paid during the year on a qualified student loan. It includes both required and voluntarily prepaid interest payments. You may deduct the lesser of $2,500 or the amount of interest you actually paid during the year. The deduction is gradually reduced and eventually eliminated by phaseout when your modified adjusted gross income (MAGI) amount reaches the annual limit for your filing status.

https://www.irs.gov/taxtopics/tc456


Arc agi 1 that "got defeated" was published even before first mainstream llms and still stood the test of time


> "got defeated"

1.5 hours with Chollet on "LLMs won’t lead to AGI - $1,000,000 Prize to find true solution" -

Published June 2024, and by December, well...we can all agree there's an ARC AGI 2 now.

https://www.dwarkesh.com/p/francois-chollet


Pre chatgpt 3 it was beautiful, fresh. Not it's another race to the bottom


The desperate attempt to decouple the economy finally from reality entirely.


We should create a new and better economy.


With blackjack and hookers!


We may be forced to.


Oh boy! Some of these tasks are not hard, but require full attention and a lot of counting just to get things right! ARC3 will go 3D perhaps? JK

Congrats on launch, lets see how long it'll take to get saturated


ARC 3 is still spatially 2D, but it adds a time dimension, and it's interactive.


I think a lot of people got discouraged, seeing how openai solved arc agi 1 by what seems like brute forcing and throwing money at it. Do you believe arc was solved in the "spirit" of the challenge? Also all the open sourced solutions seem super specific to solving arc. Is this really leading us to human level AI at open ended tasks?


Strong emphasis on "seems".

I'd encourage you to review the definition of "brute force", and then consider the absolutely immense combinatoric space represented by the grids these puzzles use.

"Brute force" simply cannot touch these puzzles. An amount of understanding and pattern recognition is strictly required, even with the large quantities of test-time compute that were used against arc-agi-1.


Also there's no clear way to verify the solution. There could be easily multiple rules which works on the same examples


It's useful to know what current AI systems can achieve with unlimited test-time compute resources. Ultimately though, the "spirit of the challenge" is efficiency, which is why we're specifically looking for solutions that are at least within 1-2 order of magnitude of cost from being competitive with humans. The Kaggle leaderboard is very resource-constrained, and on the public leaderboard you need to use less than $10,000 in compute to solve 120 tasks.


Efficiency sounds like a hardware problem as much as a software problem.

$10000 in compute is a moving target, today's GPUs are much much better than 10 years ago.


> $10000 in compute is a moving target

And it's also irrelevant in some fields. If you solve a "protein folding" problem that was a blocker for a pharma company, that 10k is peanuts now.

Same for coding. If you can spend 100$ / hr on a "mid-level" SWE agent but you can literally spawn 100 today and 0 tomorrow and reach your clients faster, again the cost is irrelevant.


Are you in the process of creating tasks that behave as an acid test for AGI? If not, do you think such a task is feasible? I read somewhere in the ARC blog that they define AGI as when creating tasks that is hard for AI but easy for humans becomes virtually impossible.


If you aren't joking, that will filter most humans.


They said at least two people out of 400 solved each problem so they're pretty hard.


I don't think that's correct. They had 400 people receive some questions, and only kept the questions that were solved by at least 2 people. The 400 people didn't all receive 120 questions (they'd have probably got bored).

If you go through the example problems you'll notice that most are testing the "aha" moment. Once you do a couple, you know what to expect, but with larger grids you have to stay focused and keep track of a few things to get it right.


The "select" tool gives some help with tasks that require counting or copying. You can select areas of the input, which will show their dimensions, and copy-paste them into the output (ctrl+c/ctrl+v).


Isn't something about alphago also involves "infinitely" many possible outcomes? Yet they cracked it, right?


Go is played on a 19x19 board. At the beginning of the game the first player has 361 possible moves. The second player then has 360 possible moves. There is always a finite and relatively “small” number of options.

I think you are thinking of the fact that it had to be approached in a different way than Minimax in chess because a brute force decision tree grows way too fast to perform well. So they had to learn models for actions and values.

In any case, Go is a perfect information game, which as I mentioned before, is not the same as problems in the real world.


Japanese sort really well. Most European countries too


Interesting, but also true that intent can be much more accurately passed with words, than inferred with guessing? Visit apple.com, reorganize this list in alphabet order, find me a ...

You get it


Not true! Look at toyota, yaris, corolla! Even rumors of celica and mr2 making a come back


I'm not a dev but I was a UI/UX designer and manager at 100 people startup. Been fired and it since been exactly two years. I have not tried applying anywhere yet, so I'm trying to build something of my own and developed my first proper Saas.

I started writing here about it https://news.ycombinator.com/item?id=43107456


Good luck! Your SaaS website looks very nice. Are you doing sales or advertisements yet?


Thanks for kind words, not really promoting yet, just patching up some things before I can go ahead and pitch it to potential clients


Join us for AI Startup School this June 16-17 in San Francisco!

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

Search: