anonylizard

Joined 5 years ago

Karma
1,717
Submissions
321
Member since
Mar 29
View on Hacker News

Recent submissions

Comment · 6 months ago
It is just sad that commentary like this even exists.

I sincerely am curious of the education that produces sentences like this. On one hand it is articulate and educated, on the other hand its amazing that one can think China is doing this out of charity and not wiping out its competitors one after the other.

Comment · 1 year ago
Why not?

Zuckerburg has already realized that local social networks will be utterly replaced by AI chatbots, which are superior in like 99% of cases.

It doesn't matter if others find "Facebook" icky, because the people who would find it icky, don't use facebook anyways. The critical factor is getting a new group of users onboard, to form a sustainable core of users.

Comment · 1 year ago
You can pay for long term support from private companies.

And your scientific context is a distinct minority for python now. Most new development for python is for data/AI. Considering LLMs get updated every quarter, and depreciated every year, there is no appetite for code that doesn't get updated for 5 years.

Comment · 1 year ago
Because python dominates AI, and python is dominating because of AI. And prompting really, really benefits from f-strings and templated strings. LLMs as a whole means the rise of unstructured data, and flexible string manipulation is really important for handling that.
Comment · 1 year ago
Its a gigantic bet on user stickiness in AI, and the monetizable value of AI users who don't pay for subscriptions. Aka low-end consumers vs high-end consumers.

Nvidia and AMD were low-end vs high-end. In the end Nvidia won a total victory by ditching low margin distractions like building GPUs for consoles, and focused solely on higher end PC GPUs that could dually act as accessible research chips.

Comment · 1 year ago
Do you worship the posting guidelines or something? Are you that offended by someone adding information to a post? The forum is a public one, not a 1-1 conversation.

The poster added valuable information, that is interesting and not self-evidently obvious to the average person who doesn't think much about restaurants, that makes the forum more useful to others?

Comment · 1 year ago
AI just represents a new class of business, that doesn't need consumers or revenue or business plans.

Traditionally you want to build a moat with name recognition, established customer base etc. OpenAI has absorbed huge losses to serve consumers at below inference cost.

But its not clear if that has advantaged OpenAI or not. Turns out most AI customers are only loyal to the most intelligent model. So if a competitor releases a better model, the shift away can be extremely rapid.

So SSI probably is a bet on that, to just develop the model in private, and not waste time and energy on deployment, name recognition, go to market, enterprise sales etc. Have a decisive model advantage, and customers will crash your door trying to get it.

Comment · 1 year ago
If one compares O3-mini's coding abilities to the original GPT-4. It is as large as GPT-3 to GPT-4 gap

GPT-3: Useful as autocomplete. Still error prone, but vastly better than any pre-AI autocomplete

GPT-4: Already capable of independently coding up simple functions based on natural language.

O3-mini: Can code in say top 5% of codeforces.

There's a 2 years gap between each of them.

More over, intelligence has a superexponential return, 90IQ->100IQ < 100IQ->110IQ in terms of returns.

Comment · 1 year ago
Most of social sciences today is 'degenerate' by your definition, at least a plurality of those researchers openly admit that following the evidence is not the highest virtue, but arriving at the most socially correct result is.
Comment · 1 year ago
Indeed you can. This only affects the heavily subsidized usps shipping process. If your r&d cannot afford fedex you shouldn’t be doing r&d.

For all the supposed panic and importance of this process, the poster didn’t do even a google search of this issue

Comment · 1 year ago
Government is exactly a codebase. Government bureaucracies is essentially constricting human judgement to more robotic code-like behavior, that's the only way to build large systems.

You say government is not like code, then what exactly is it? Can you describe it in an effective way? Or are you just going to raise your hand up and say there's nothing we can do about it, nothing we can do about the $2 trillion/year titanic deficit?

Historical governments often needed little beyond an army and a tax collection system. And tax collection system was primarily data gathering and analysis, since if you knew how much property someone owned, you can easily tax them for an appropiate amount.

The tech way of thinking has proven extremely successful in many industries already. That's why tech companies (and tech adjacent ones, like say quant trading, or even index fund trading) have been so economically dominant, and utterly kicked out the traditional MBAs from their pedestals.

Stop being a self hating programmer who despises the mentality of tech.

Comment · 1 year ago
Compelled speech is far, far, far worse than constricting speech by every conceivable dimension.

Even in day to day interaction, forcing someone to be silent, is far more of a gentler 'social action' than forcing someone to speak.

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton