HNHacker News
TopNewBestAskShowJobs

parodysbird

201 karma · joined September 25, 2024

submissionscomments
parodysbird··on AI is eroding code quality states new in-depth report • DEVCLASS
The yearly dots are the individual points of data; so the lines between e.g. 2021 and 2022, despite being the main visual marker, do not indicate a change that happened in 2021 that leads to 2022, but rather a change that happened in 2022 compared to 2021.

GPT 3.5 came out early in 2022, chatGPT later the same year. Lots of change and community discussion that year around this sort of thing.

parodysbird··on Andrej Karpathy: "I was given early access to Grok 3 earlier today"
No, we do not try to answer the original trolley problem in earnest. We immediately reject and move on.
parodysbird··on The Generative AI Con
He's not a developer. He's really talking about consumer tech.
parodysbird··on The Generative AI Con
This is not a criticism he really mentions explicitly, but the issue I see with the valuation is that these products cannot be used in a production system by a responsible engineer. As in, I can't have LLMs autonomously plugged in as part of my product. The failure modes are not ever going to be predictable enough. Now, Microsoft will probably be able to charge a fortune at enterprise level, and managers will dream of a day that the LLMs can replace all those weirdo devs at their company, but that'll stay a dream.

All of the valuable uses are personal. It makes me personally feel more productive at work. It helps me personally understand some topic better. It gives me an idea in a personal project.

That's all really cool, but that is not what the valuation is about. The valuation is about a false science fiction and hype bubble about agentic this or that or AGI or whatever, and this is driving very questionable decisions for wasting possibly trillions of dollars and tons of energy.

The plus side is that there is some really cool personally useful tech here, and we will probably end up with very good open source implementations and cheaper used GPUs once the bubble bursts.

parodysbird··on The Generative AI Con
This is a great point. Now I have extra reason to cheer on the bubble bursting.
parodysbird··on X users are unable to post “Signal.me” links
He's just queried records with dead=FALSE. It's not active recipients of social security payments. All its showing is that some portion of people die without the death being officially logged on their record in the social security database.

Edit: Also, it's not clear that the death field is the sole criteria for determining the eligibility for payments (i.e. determining the recipient is alive)

parodysbird··on Why blog if nobody reads it?
Academic papers are supposed to be about the author... They're meant to be an author's work put forth to an intended community of "colleagues", not students or general public. No one should think that a general learner is supposed to turn to academic research papers as their main vector of learning content.
parodysbird··on Bayesian Epistemology (2022)
> Invertebrates were adjusting their beliefs to observations (science) long before the first philosopher.

To underscore the bad science you are led to in terms of assumed truth, let alone hypothesis: there is very little evidence or justification or explanation that any of the processes used by the invertebrate here execute calculation that obeys the very specific axioms of probability and updates to a state in accordance with Bayes' theorem. Stimulus response is not Bayes' theorem. Updating a state from new inputs is not Bayes' theorem.

parodysbird··on Bayesian Epistemology (2022)
> Bayesian epistemology is probably the best description of how we know things based on science.

This is wrong, and it's a bit ironic you are so adamant on a point that is bad philosophy and leads to bad science as a way of insisting that philosophy has no relevance for science.

parodysbird··on Bayesian Epistemology (2022)
Bayesian epistemology is not used in almost any domain in science, it does not model why science works, and it does not rely on science: it relies on metaphysics.
parodysbird··on Bayesian Epistemology (2022)
Because the use of Bayesian models does not require Bayesian epistemology, which, unfortunately, a lot of people conflate.

Here is a decent paper by a statistician about this issue: https://arxiv.org/abs/1006.3868

parodysbird··on Bayesian Epistemology (2022)
He's talking about Bayesian philosophy of science, not science, which ultimately does not rely on Bayesian epistemology.
parodysbird··on Bayesian Epistemology (2022)
I don't think you know what you're responding to, but in any case, regarding Deutsch, he "laid the foundations of the quantum theory of computation, and subsequently made or participated in many of the most important advances in the field, including the discovery of the first quantum algorithms, the theory of quantum logic gates and quantum computational networks, the first quantum error-correction scheme, and several fundamental quantum universality results."
parodysbird··on RLHF Book
RL on any production system is very tricky, and so it seems difficult to work in any open domain, not just LLMs. My suspicion is that RL training is a coalgebra to almost every other form of ML and statistical training, and we don't have a good mathematical understanding how it behaves.
parodysbird··on 'AI-powered judge' takes boxing closer to brave new world it appears to seek
Are we sure the CompuBox punch stats weren't part of the input for the model after each round?
parodysbird··on 2007 Boston Mooninite Panic
What evidence is there that someone intentionally used the context that the installations would be seen as bombs? And on what planet should the first reaction by police to seeing a bunch of obvious LED street art installations as being a massive bomb threat be seen as reasonable? If police think something stupid, it does not mean that stupid thought must be taken as legitimate and acceptable just because they are police.

The police should have apologized and taken responsibility for instigating an unnecessary panic.

parodysbird··on Large language models in national security applications
Warfare is not a chaotic system. We don't think outcomes are highly sensitive to marginal tweaks in initial conditions of the model. Hostile actor aren't modelled as chaotic systems but as agents in some game theory model. None of these agents has a monopoly on violence or power or information, or else it wouldn't be warfare.
parodysbird··on Revealing causal links in complex systems: New algorithm shows hidden influences
Correlation also assumes model of the data generating process, but you are correct in thinking that talking about causal links imposes even stronger assumption on the model and data structure for making inference. And then further you have to take a very narrow and convenient interpretation of what causality means (e.g. can't be at the actual level of individual samples, can't manifest through cycles or loops in the variables etc), which is even more of a vexing philosophical question than even the thorny questions in classical statistical inference
parodysbird··on Interview with Terence Tao in Barcelona
The author is the science writer for El Pais and many of his articles come out in both English and Spanish (no translator credited? Maybe that's normal, or maybe he writes them in both languages at once). So I would expect him to write with a very Romance prosody in English.

This article is not in the Spanish edition (yet) from what I can tell, probably because the interview is in English, so you would need a full professional translation rather than an author translation.

parodysbird··on Why don't we use awnings anymore (2022)
I had an awning and a pool enclosure in South Florida. So did most houses in the neighborhood. Then the 2004 hurricane season happened, and there was neither of each around anywhere ever again.
parodysbird··on South Korean Author Han Kang Wins Nobel Prize
> They are quickly followed by internet trolls who derail every discussion by insisting that these are masterpieces certified by big-name committees, claiming that we, the lowly masses, must accept the decisions of the great authority as the absolute truth of our lives.

Who does this with the Nobel Prize on literature? I've not even heard of any of the previous winners on until Bob Dylan in 2016.

Even with the Oscars, like who is demeaning the lowly masses that they must view as absolute truth that e.g. CODA or The Whale or The Power of the Dog etc are masterpieces?

parodysbird··on South Korean Author Han Kang Wins Nobel Prize
I enjoy Oscar bait films and Michelin star restaurants. Oscar bait film season (winter) is the main season I look forward to going to to the movies, not summer (blockbuster season). But it's not some attack on mainstream culture just because there are awards for niche/elite/artsy/etc culture. Similarly, just because the Nobel for literature goes to a niche literary author you don't know/like doesn't demean whatever your reading preferences are.
parodysbird··on Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton [pdf]
Well yeah, so are neural nets. I just meant that these are engineering accomplishments, not scientific per se. Of course experimental science will often take advantage of cutting edge technology, including from computer science.
parodysbird··on Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton [pdf]
Some other recent cases of the prize being given to an engineering contribution:

- 2018 was for chirped pulse amplification, which is most commonly used in medicine (LASIK surgery for example)

- 2014 was for basically for LED lights

- 2010 was for a method for producing graphene

- 2009 was for both charge-coupled device, which is a component for digital imaging (including regular consumer digital cameras), and fibre-optic cables

parodysbird··on Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton [pdf]
No you couldn't. Windows doesn't have any bearing on outcomes, whereas machine learning methods directly impact the data and probability inference.
parodysbird··on Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton
https://www.sciencedirect.com/science/article/pii/S037015731...

"Jet substructure at the Large Hadron Collider: A review of recent advances in theory and machine learning"

parodysbird··on Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton [pdf]
> but they have made no contribution to understanding physical laws or phenomena.

Neural networks are used in tons of data pipelines for physics experiments, most notably with particle accelerators.

The Nobel Prize is also occasionally awarded to engineers who develop tools that are important parts of experiments. 2018 for example was awarded for chirped pulse amplification, which is probably best known for being used in LASIK eye surgery, but it is also used in experimental pipelines.

parodysbird··on Legalizing sports gambling was a mistake
Polymarket isn't legal in the US
parodysbird··on Why Most Published Research Findings Are False (2005)
> But at the same time, science landed humans on the moon.

That was engineering. Closely linked to science, but not the same process of inquiry.

← PreviousPage 3 of 3