626 karma · joined April 8, 2022
Every institution which could in theory check executive overreach is either already captured by ideological allies or is completely overwhelmed.
"Equipment provided by Raytheon as part of a multi-million dollar contract broke the winter after it was installed. One U.S. official said most of the detectors had been designed by Raytheon for the desert environment of the U.S.-Mexican border. The Kazakhs, on their own initiative, sourced equipment designed to withstand Siberian winters from a Russian military supplier; it cost half the amount of the U.S. contract, and easily survived the winter."
Good luck!
In terms of creative writing I think we need to accept that any proper assessment will require a short essay to be written in person. Especially at the high school level there's no reason why a 12th grade student should be passing english class if they can't write something half-decent in 90 minutes. And it doesn't need to be pen and paper - I'm sure there are ways to lock a chromebook into some kind of notepad software that lacks writing assistance.
Education should not be thought of as solely a pathway to employment it's about making sure people are competent enough to interface with most of society and to participate in our broader culture. It's literally an exercise in enlightenment - we want students to have original insights about history, culture, science, and art. It is crucial to produce people who are pleasant to be around and who are interesting to talk to - otherwise what's the point?
To experiment with BGP you could use a network simulator like what the author of this blog did. In my class we used something called gini[1] which I think my profs grad student wrote but the author apparently used gns3 which seems to be a cisco specific ns3 version. I used ns3 once and found it had a steep learning curve. The gini simulator has a more basic user interface but is probably less powerful.
[1] https://citelab.github.io/gini5/ [2] https://docs.gns3.com/docs/
It's hard to say that there is no internal organization since trillion parameter models are hard for us to summarize and we do see some semantic vector alignment in the GPT models but the toy example of the 2 skull image generator present a powerful anecdote of how current ML models find correct solutions but miss a potentially valuable property of having what the paper calls factored representation which seems to be the far more "human" way to reason about data.
It won't - that's not what this paper is about.
A lot of the headline advancements in AI place lots of emphasis on model size and training dataset size. These numbers always make it into abstracts and press releases and especially for LLMs even cursory investigation into how outputs are derived from inputs through different parts of the model is completely waved off with vague language along the lines of manifold hypothesis or semantic vectors.
This section stands out: "However, order cannot be everything—humans seem to be capable of intentionally reorganizing information through reanalysis or recompression, without the need for additional input data, all in an attempt to smooth out [Fractured Entangled Representation]. It is like having two different maps of the same place that overlap and suddenly realizing they are actually the same place. While clearly it is possible to change the internal representation of LLMs through further training, this kind of active and intentional representational revision has no clear analog in LLMs today."
This quote sums it up perfectly, the worst part is not the bias it's the false articulation of a grounded decision.
"In this context, LLMs do not appear to act rationally. Instead, they generate articulate responses that may superficially seem logically sound but ultimately lack grounding in principled reasoning."
I know some smart people who are convinced by LLM outputs in the way they can be convinced by a knowledgeable colleague.
The model is usually good about showing its work but this should be thought of as an over-fitting problem especially if the prompt requested that a subjective decision be made.
People need to realize that the current LLM interfaces will always sound incredibly reasonable even if the policy prescription it selects was a coin toss.
I've done only very simple ML stuff using R and it never feels like exactly the right tool.
It's pretty good considering it is his first not-non-fiction project and the narrative is a refreshing departure from typical sci-fi stories since it's written to sound like a true history with too many important figures to remember and historically disputed causes and effects of pivotal events.
The story doesn't not follow the conflict-rising-climax-resolution structure but it often refutes a listener's anticipation of satisfying narrative elements like true history many loose ends remain loose and plenty of important characters "disappear from the records" which leaves one wondering.
It's certainly unlike any fiction I had consumed prior and it's pretty good imo so I'm shining a light on it here.
Solar PV is great but is mostly grid-following so cannot operate on it's own. As I understand it you need a minimum fraction of power generation to be large spinning turbines.
I think this problem can be mitigated with add-on rotational mass style kinetic energy batteries or something like that. I don't think variable energy pricing will help if it's an issue with over-demand the grid managers can do rolling blackouts to manage while fixing the supply problems. The grid is just broken at the moment and the solar can't maintain the grid alone.
My feeling is that these over-water panels generating a not-insignificant amount of power are an ideal compromise between the sprawl of the built environment of (Taizhou, China) and the natural ecosystem.
Land use is a large contributing factor to climate change but this particular image seems to be the best case scenario for large solar installation the ecosystem (marshland?) appears to be relatively undisturbed (also considering it is centrally located in a city of 4 million people solar panels or not it seems pretty lush). And I'll parrot an often cited statistic: "the entire U.S. could be powered by utility-scale solar occupying just 0.6% of the nation’s land mass" this (imo) makes a good case for solar PV as a relatively large chunk of installed grid capacity it just seems like a better compromise than the alternatives.
If utility scale battery storage ever pans out (that or UHV transmission or both) then we could see renewable sources like solar+wind actually work as base-load capacity.
Others also comment on this thread that converted agricultural land (i.e. field to pasture) meshes well with solar PV installation the plants typically do not need full sun. I would guess that a small percentage of installed solar required any kind of land shaping or clearing although it does happen [2].
[1] https://www.scientificamerican.com/article/china-invests-546...
[2] https://www.cmigroupinc.ca/solar-farm-environmental-impact-b...
Like a small vision model combined with the size/measurements data from the AR sensors modern phones come with and an open source caloric values database should achieve the 90% accuracy they are claiming.
Ronald Wright writes about "progress traps" in A Short History of Progress. It's been awhile since I read that but I think about it more and more these days with AI products on the rise.
[1] https://en.wikipedia.org/wiki/Government_incentives_for_plug...
"The company's pullback on data center projects has raised concerns about demand for AI services and has weighed on global tech stocks, particularly chipmakers like Nvidia.
Summary by Bloomberg AI"
https://paulkrugman.substack.com/p/will-careless-stupidity-k...
The system does not represent ownership the system only tracks of the validity of transactions and if the North Korean government proposes a valid transaction of your BTC or ETH to an address they control and a mining-node includes that transaction in a block which a majority of the network accepts then those assets are no longer yours they belong to North Korea.
The properties of the crypto-asset ecosystem which allow it to be ungoverned also make it ungovernable.
The American government is proposing across the board on every import (>=$800 or something) tariffs which has never been implemented by the Canadian government on American goods.
Canada is highly protective of certain industries which are typically small and are engaged in by a large voting population but small portion of the economy (GDP). As a percentage of total goods that American companies might actually want to export to Canada import tariffs are typically quite low.
However to poke just a little fun this line stands out as a hilarious marketing claim that cannot possibly be true
"Almost all (98%) of UAE coastal residents are interested in riding a seaglider, according to our global consumer survey."
It's also interesting to note that the CEO has appeared on no fewer than 13 different podcasts between mid-2022 and last week which is I suppose how one would go about getting a high value US defence contract these days.
"A 1/4 scale model was successfully demonstrated in 2022 in Narragansett Bay"[1]
Also I assume radar-proof is just because it's a ground effect vehicle that will never fly high enough to show up on radar it certainly doesn't look all angular like a stealth bomber. In which case my bicycle is also radar-proof?
I’m not an expert but I think it has more to too with reducing vibrations along the bat and increasing the efficiency of energy transfer
Again not an expert
"On the other hand, we are also mindful of the low baserate for a US administration directly attacking Iran. We also considered modelling the administration as being dispositionally very over the top."
Of these 3 options sas datasets are my preference but I'll immediately convert to csv or excel, csv is a close 2nd once you confirm the quoting / seperator conventions it's very easy to parse. I understand why someone may find the csv format disagreeable but in my experience the alternatives can be so much worse I don't worry too much about csv files