Tarq0n

4 ms·

31.
by Tarq0n 4mo ago
Rising demand induces inflation because more money is chasing the same goods.
32.
by Tarq0n 4mo ago
It's not just minimum wages. If productive sector employees' demand for goods causes inflation, prices will rise. If you want to keep your nurses/doormen/teachers you'll have to increase their wages regardless of wh
33.
by Tarq0n 4mo ago
I've been to staffless hotels (numa) that were good and noticeably more affordable than their neighbors. I don't know if the housekeeping was outsourced though, I imagine that, without at least some local management staff, conditi
34.
by Tarq0n 4mo ago
Labour was cheaper back then. Even valuable jobs can stop making sense if the costs outweigh them. That's the difficulty with automation making other sectors more efficient, wages get driven up while your productivity stays the same.
35.
by Tarq0n 4mo ago
Data.table is a masterclass in bad API design. Its lack of success despite its technical merits is entirely of their own doing.
36.
by Tarq0n 4mo ago
Sounds like socializing the harms instead of requiring these companies to bear the burden themselves. Could still be a valid approach but I'm afraid it will make them take less responsibility, not more.
37.
by Tarq0n 4mo ago
Recently finished the Starfish quadrilogy and adored it as well. One thing I particularly like is how this theme of evolutionary pressure applies to everyone alike; characters, software, ideologies, neural nets. Also Watts manages to conjur
38.
by Tarq0n 4mo ago
REAMDE's description of MMO economics is farcically bad. It undermines the whole book.
39.
by Tarq0n 5mo ago
The conclusions seem overly broad. Just because these languages are Turing complete doesn't mean they aren't massively hampered by expressiveness and amount of batteries included. To attribute all of this to training data memoriza
40.
by Tarq0n 5mo ago
Of course they do, error messages get tokenized and put into the context window just like anything else. This isn't a Markov chain.
41.
by Tarq0n 6mo ago
The aggregate effect of the chemical gradients in the antheap controls the ants. It's similar in these giant companies where the combination of all the incentives makes for a gradient towards certain outcomes. I can see little intentio
42.
by Tarq0n 6mo ago
Language is intersubjective (its meaning is in the minds of the participants). Referring to the history or composition of a word is interesting but entirely insufficient to justify its use.
43.
by Tarq0n 6mo ago
State governments are very much not under "management" of the federal government.
44.
by Tarq0n 6mo ago
Not definitively. LLMs are stochastic with respect to input, temperature and the exact prompt. It's possible that the model was already capable of it but never received the exact right conditions to produce this output.
45.
by Tarq0n 6mo ago
That's not how it works in many countries. You can have regional governments that raise their own taxes and aren't beholden to the central government organizationally, just legally.
46.
by Tarq0n 6mo ago
We're trying to have a discussion about facts, not opinions.
47.
by Tarq0n 6mo ago
Ecological footprints are more about the distribution of resource burdens between first and third world countries. Predictions about resource exhaustion have proved mostly false.
48.
by Tarq0n 6mo ago
That's the definition of a tree though. Everything has a parent, no cycles allowed.
49.
by Tarq0n 6mo ago
That sounds like people making an informed decision about their body to me. Doctors don't need to be inserted into every risky decision.
50.
by Tarq0n 6mo ago
You can't trust what a model says about itself. It has no ability to introspect.
51.
by Tarq0n 6mo ago
A lot of enterprise orgs are completely helpless without Microsofts' identity solutions. That's what makes it easy to just adopt more and more Microsoft products.
52.
by Tarq0n 7mo ago
It seems likely that they buy data from companies who don't obey the same constraints however, making it easy to launder the unethical part through a third party.
53.
by Tarq0n 7mo ago
If it works for predicting the next token in a very long stream of tokens, why not. The question is what architecture and training regimen it needs to generalize.
54.
by Tarq0n 7mo ago
I don't like the "those in power" framing because it implies that they all participated and that such a homogenous group even exists.
55.
by Tarq0n 7mo ago
Yes their translations offer negative value, which is annoying because at work you can't usually choose your locale settings. And the errors are really basic, like translating shortly to short, not the same thing at all!
56.
by Tarq0n 7mo ago
In machine learning, ensembles of weaker models can outperform a single strong model because they have different distributions of errors. Machine learning models tend to have more pronounced bias in their design than LLMs though. So to me
57.
by Tarq0n 7mo ago
OP has his retirement prepared. That might increase their perception of the upsides and negate some of the downsides of adopting AI.
58.
by Tarq0n 7mo ago
At this point theists often try to smuggle God back in as the source of morality through culture. But I agree, empirically religion and moral behavior seem at best uncorrelated.
59.
by Tarq0n 7mo ago
Transcendental moral principles can still be secular. One that I find compelling is that Rawls' veil of ignorance lets us imagine that we might be on either side of a conflict, and that therefore moral actions are equitable to both sid
60.
by Tarq0n 7mo ago
No it's just stochastic like everything about LLMs. The md file will bias results towards a certain set of outcomes.

More ›