HNHacker News
TopNewBestAskShowJobs

curiousllama

5,638 karma · joined February 24, 2020

submissionscomments
curiousllama··on OpenAI’s CEO says the age of giant AI models is already over
I believe Altman, but the title is misleading.

Have we exhausted the value of larger models on current architecture? Probably yes. I trust OpenAI would throw more $ at it if there was anything left on the table.

Have we been here before? Also yes. I recall hearing similar things about LSTMs when they were in vogue.

Will the next game changing architecture require a huge model? Probably. Don’t see any sign these things are scaling _worse_ with more data/compute.

The age of huge models with current architecture could be over, but that started what, 5 years ago? Who cares?

curiousllama··on Ask HN: Why call it an AI company if all it does is call open AI API?
It makes perfect sense - if you reorient your thinking.

AI companies come in two flavors.

(1) Build a novel AI for a pre-existing purpose

(2) Use a pre-existing AI for novel purpose.

It's "I didn't realize _AI_ could do that!" vs. "I didn't realize AI could do _that_!"

Engineers found company type 1. Product people found company type 2. Both are valuable, but for different reasons.

curiousllama··on Laid-off tech workers choose stability over status
20k per person?
curiousllama··on Defamed by ChatGPT
This is a really interesting legal issue. What recourse does a private person have to algorithmically-generated libel?

Can I sue the company? They’re the ones who trained the model, but there’s no intent.

Could I sue the user? They’re the one who generated the response, but there’s no intent nor duty to fact check, if they don’t repeat the claim.

Do I have to subpoena OpenAI, reverse engineer the response, then sue the creators of the likely-responsible training data? They probably wrote something non-libelous.

It feels very all or nothing - either creators of LLMs are intrinsically negligent with regard to facts, or there’s nothing you can do.

curiousllama··on The Contradictions of Sam Altman
> In a winner takes all market

Just because Search was winner take all, doesn't mean AI will be. What network effects or economies of scale are unachievable by competitors? Besides, Alpaca showed you can replicate ChatGPT on the cheap once its built - what's stopping others from succeeding?

curiousllama··on Sam Altman didn’t take any equity in OpenAI, report says
Details, details
curiousllama··on Sam Altman didn’t take any equity in OpenAI, report says
Microsoft owns 49% IIRC. Prob a bunch of the other folks who pledged $ at the beginning as well (Thiel, Musk, etc)
curiousllama··on Facebook is going after LLaMA repos with DMCA's
The phone book argument is super interesting. I wonder if it would hold, though: Wikipedia suggests it relies on the re-user having a new “selection & arrangement.” Kinda tough to rearrange model weights!
curiousllama··on Facebook is going after LLaMA repos with DMCA's
That 4B # encompasses AI and Metaverse. Hard to tell where the $ is actually going (plus, metaverse requires AI advancement).

That’s one big reason the stock is up: $4B in the metaverse is nonsense. $4B in AI, though? Transformational

curiousllama··on UBS got Credit Suisse for almost nothing
> suppository profited

I think you mean supposedly

curiousllama··on U.S. lawmakers to examine merits of higher FDIC bank deposit insurance cap
I mean it is amazing if they do it well. If we don’t have a financial crisis soon, it would be a wonderful sign of responsible governance, effective regulation, and international cooperation.

So many doubt the ability of the US government to govern responsibly. It IS amazing when they demonstrate responsibility!

curiousllama··on U.S. lawmakers to examine merits of higher FDIC bank deposit insurance cap
If the bank collapses, you get wiped out. Incentive for the banker is the same. Incentive for the depositor shifts assuming perfect information. But in practice, information asymmetry makes depositor preference insensitive to bank risk, regardless of insurance.

This would just be an acknowledgement “we’re always gonna have to do it for the big guys, might as well make it fair for the little guys too”

curiousllama··on U.S. lawmakers to examine merits of higher FDIC bank deposit insurance cap
This is a good thing, no?

More turnover among sr folks leads to a greater focus on (1) institutional risk management, because you can’t trust the new guys view things the same as the old, and (2) consistent reevaluation of previous assumptions, because the guy who made them is gone.

I’d argue the long tenure of sr bank employees is a weakness of the current system, not a strength.

curiousllama··on Credit Suisse sheds nearly 25%, key backer says no more money
They owe folks money. If you buy them, then you owe those folks that money. So you'll have to pay the purchase price, and then pay the company's debts.
curiousllama··on Online tests suggest IQ scores in US dropped for first time in nearly a century
Yes, but the authors of this paper hold the test constant, so if people get smarter, it’ll show IQ rising
curiousllama··on Online tests suggest IQ scores in US dropped for first time in nearly a century
Yea pretty much. Those darned kids are just too smart. But your IQ percentile relative to your age group doesn't change.
curiousllama··on Online tests suggest IQ scores in US dropped for first time in nearly a century
This paper feels methodologically iffy. I'd be willing to bet money that, if you don't stratify by education level, the decline is not stat-sig, or IQ even grew. This trend, and the way they reported it, screams Simpson's paradox to me. I.e. the headline should be "more lower-IQ people went to college" not "college graduates' IQ is dropping"

Who the heck wouldn't report a number like "Since 2006, average IQ has dropped X%"?

curiousllama··on FDIC auction for SVB said to be underway, final bids due Sunday
I thought they were bidding negative amounts - ie the bidding is to find the lowest $ amount the FDIC needs to put in to make folks whole
curiousllama··on “Healthy obesity” doesn’t exist. It is unequivocally bad for you
> most men who go to the gym frequently are in the category of "healthy obese"

Healthy obese because of muscle mass is the size of an NFL linebacker, not “most men who go to the gym”

ETA: yea, just checked. TJ Watt has a BMI of 30.7 (barely obese BMI). BMI only matters here if you’re TJ Watt lmao

curiousllama··on Indirect Prompt Injection on Bing Chat
The point is not that [bad thing] isn't bad. It's that scolding is the wrong way to go about it. How would your friend's dad have reacted to a stranger at the gas station yelling at him about the kids in the bed?

A better approach is friendliness, and gently pointing out how dangerous it is. "Hey man - careful with those kids. My neighbor died that way growing up - might be worth pulling them into the cab" is more likely to get a "Really? Dang, yea, you're right"

curiousllama··on Uncle Sam to block Adobe absorption of Figma over monopoly fears
What’s the implied harm to citizens?
curiousllama··on It’s time for Alphabet to spin off YouTube?
That stuff still comes with special privileges.

E.g., yes they'll "pay" for compute, but they'll do so at cost. And they're also allowed to deploy a monitoring widgets to other orgs' data centers to find when they're at low utilization, and then argue "at cost" is actually near 0, because the compute resources were just sitting around anyway. And then have the search team agree because YouTube promises to build the new data API search has been asking for forever.

If you ultimately report earnings on the same P&L, USD is just monopoly money you trade around to get things done.

curiousllama··on Google search for a local restaurant returns scammer’s ad as top result
Yea, this is the answer. If it’s easy to file a complaint, scammers will just file a million complaints to (wrongly) win 1%, and walk away with the $
curiousllama··on Meta plans to cut thousands of jobs, after CEO predicted no more layoffs
Third? The article says this is the second, no?
curiousllama··on ChatGPT launches boom in AI-written e-books on Amazon
That ruling just says a person has to claim copyright. Whoever game the prompt to the AI is the "author" for copyright purposes
curiousllama··on Professor writes history essays with ChatGPT and has students correct them
> in the past I've given them bullshit proofs from real students in exams

I'd be so honored to be one of these proofs. "Your wrongness is an elegant balance of instructive error and subtle misunderstanding. Can I save it for posterity? (c-)"

curiousllama··on Professor writes history essays with ChatGPT and has students correct them
> "the education system" currently hyperindexes on rote memorization

Which education system, and in what subjects? I recall quite the opposite (rando public schools in the US). In hindsight, I would have benefitted a lot from simply memorizing more things.

My freshman calc grade would've been a lot higher if I had bothered to memorize trig identities in HS. Sure, I could derive them, but I had to notice "this is an identity" first - the lack of memorization actually hindered my intuition.

curiousllama··on Statement of Jewish scientists opposing the “judicial reform” in Israel
I’m not familiar with Israeli politics or government, but it’s certainly not oximoronic.

Strengthens democracy != is itself democratic

Eg, the US requiring a supermajority for impeachment is antidemocratic (the minority decides), but it strengthens democracy (because it shouldn’t be done by one side alone).

This is a pretty core principle of democratic government

curiousllama··on Bing: “I will not harm you unless you harm me first”
It seems like this round of AI hype is going to go the same way as voice assistants: cool, interesting, fun to play with - but ultimately just one intermediate solution, without a whole lot of utility on its own
curiousllama··on A little bit of slope makes up for a lot of Y-intercept (2012)
I like your model, but I love that you put Anticipate at the front. It's a loop so doesn't really matter, but I'm entertained by the idea of "Step 1: Take a wild ass guess at what's about to happen"
← PreviousPage 7 of 30Next →