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

It's not exactly what it is; they now model an incredibly complex markov process, and harnesses that control how that thinking is done.

Is this any different than how a PM gets a programmer to work on a project? They think, then they deliver. If given more time, maybe they deliver something better. Maybe they consult some text and try to apply a design pattern.

The LLM in this use case is perfect because almost everything involved is text based, and the model is able to take in all the expressive that is language.


> Is this any different than how a PM gets a programmer to work on a project?

Yes, it's very different. You seem to be suggesting that the current frontier LLMs, when tied to their tools and harnesses, have emergent properties that are similar to human consciousness. If you truly believe that, I'm not sure how to have a productive discussion here.


It's not just that, but the core is just that, even with reasoning models. Harness can only get you closer to the good result, but can't save you from every pitfall. As for PM analogy - don't forget that models don't learn and keep doing same stupid stuff they were doing a month ago.

I don't think LeCun is saying they won't be able to program. I think he says we won't hit AGI. Programming does not require AGI; it's a pretty specific skill!

-- I think this article is COPE, if I'm being quite honest. I thought of putting cute analogies, like the C programmers saying the Python and Javascript programmers are not "hardcore" enough... but the truth should be obvious to anyone using LLMs effectively.

-- Current AI is a much better programmer than 100% of people and when directed by someone in that top 10%, it's a force majeur.


Isn't the thinking part the part that burns the tokens? You're just outputting tokens.

Totally depends, but I think this is mostly just an illustration of overall speed, regardless of the content.

Okay, but I think the realistic thing is * burns 18000 tokens thinking of the solution * outputs 1000 tokens of code

So you can easily follow the 1000 tokens of code, and the 18000 tokens of thinking is you sitting around waiting for your GPU to process the LLM.


Unfortunately, if everyone goes down people are understanding. If just _you_ go down, then its oddly less forgiveable.

It's sad that our government can't pass a bill without it being a katamari ball.

One thing, discuss, vote.

No "hey if we give you this, you give us this." just simple "do most of us agree on this?" level politics.

That's real democracy, not the crap we have today.


Well the popular argument is that it takes so long to pass any kind of bill that smaller bills would just mean more bills and a bigger backlog. I don't really buy that.

The real reason is that it's easy to sneak stuff into a bill, so why not? That and trying to attack political opponents by joining something politically disastrous to <their side> to an otherwise uncontroversial bill.


Yes; it wouldn't take so long if bills were simple, straightforward and easy to understand. Instead they're rubberband balls of horse traded, full of items that as you said someone has tried to 'sneak' in.

I wish our parties didn't view each other as opponents as much as collaboration partners ultimately responsible for the continued well being of the nation rather than only for their own political good.


The nature of a congress is that every bill gets balanced with the interests of a majority of the people there.

-- I actively seek out the Kirkland clothes.


"the solo adult men with AirPods in (also sold at Costco), listening to God knows what podcast: Are they thinking about a dead relative’s favorite Costco items, too?"

-- this is me. I am seen.


Anyone who is interested in how this sort of thing could work, or interested in how it has worked, could read the book "Barbarians at the Gate".

(https://en.wikipedia.org/wiki/Barbarians_at_the_Gate)

This sort of acquisition is typically called an LBO or Leveraged Buy Out. The story gets into the details of key figures involved, including Henry Kravis who people today would better know for his private equity firm KKR.

Basic formula is raise cash using junk bonds, buy company, fix up company or kill off costs, use money company earns to pay debt, sell company. Since you have leveraged, your payout can be large.

It's a similar style financing activity to a house flipper. example: buy house for 1M, but pay $200K + 800K debt (mortgage). Fix up house, sell for $1.2M. Pay off $800K debt. You're left with $400K, or 100% return!


Minus fixing + selling costs


Yep! Always helps when you have sweat equity or some clever idea.

LBOs can have other formats. Take Debt, assign debt to part of the company, auction off arms, legs, kidneys of company to pay off debt... you're left with a shell of your former company but a boatload of cash to pay off the debt and everyone gets their bonuses.

Unfortunately a lot of LBOs tend to result in job cuts. The Twitter deal was a bit of that sort of thing, and we all see where that went.


I have a GC but it is about to expire this summer. If I apply for citizenship, can I still travel? Do I need to apply to extend my GC as well?


Yes, the processes are separate so you will need to apply for a new green card to be able to travel internationally but after you apply, you should be able to travel on your green card renewal application receipt notice (along with your expired green card).


This is basically the same problem of products astroturfing reddit, or SEO optimizing google. You want a new X, and so they heavily go after the keywords associated with it.

This is sort of why "brand" matters; it provides a source of trust.

Encyclopedia Britannica used to be that source of 'facts'. Then it became whatever page-rank told you. Eventually SEO optimization ruined that.

News stories are the same thing. For certain groups, they have their 'independent' publication whose reporting they trust.


>This is sort of why "brand" matters; it provides a source of trust

it tells you more about who you are buying from than how good the product will be, so I guess it's like National ID/Internet ID


It's such a pity the Oxford English Dictionary decided to paywall themselves decades ago - they used to be THE dictionary in most countries, now nobody seems to know who they are.


They would have been better off going freemium or ad-supported. Or 501(c)(3) ala wikipedia?


The OED’s goal isn’t really to be every nation’s dictionary.


What would you say their goal is >_>

curious!


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

Search: