@pmddomingos Except it's a 2st year undergraduate programming challenge to write a program to solve NP puzzles like Sudoku and Nonograms, and the best AI models still get them wrong most of the time.
English
Chris Jefferson
3.1K posts
Chris Jefferson
@Azumanga
AI and Maths Professor. Not spending time on Elon’s hellsite any more. @[email protected] @heathercafe.bsky.social
Earth เข้าร่วม Nisan 2008
486 กำลังติดตาม594 ผู้ติดตาม
@joshwoodward @GeminiApp Why do you need 72 hours to keep Gemini safe and respond to us? What does keep Gemini safe even mean? When I close a private tab in Chrome or Firefox, it's just gone, instantly.
English
How to personalize your @Geminiapp, announcing 2 features you wanted:
1) Personalize based on your past chats, so you don’t have to repeat things
2) Use Temporary Chats for conversations you don’t want saved
Both rolling out now for everyone, starting on 2.5 Pro model… enjoy!
English
@d_m_d_m_d_d Maths people make not talk about how clever they uniquely are, but they are very happy to talk about how clever everyone in maths is, and how it is the hardest and most important subject in the whole world, more than any other subject, in my experience at least.
English
not trying to be that much of a hater, but i find it interesting that, among math people i know, going on about how uniquely smart and special you are is frowned upon, but it seems to be (conservatively) 60% of what programmers/engineers/finance people talk about
English
@chillbert32 @killian_arts Honestly, I don't believe people wanted bad languages.
When I was learning to program in 2000, I looked for a good, free Windows lisp that would let me make GUI applications. I couldn't find one. So I learnt C++ instead.
English
@killian_arts For better or for worse, Lisp is antithetical to that paradigm. You can do it very well of course but there is far too much flexibility, as if the plumber was allowed to route the sink pipes out of the house beneath the grass. That's a liability for a company.
English
One of the factors many people attribute to Lisp's fall from grace is the fact that all of the companies making Lisps were adamant about not sharing what they made and insisting that they get paid huge sums of money for their products (look at the insane prices Franz and LispWorks STILL CHARGE TODAY).
Because of this cutthroat competitive spirit and ego, the development of Common Lisp was slowed down and no successors were ever conceived.
Meanwhile, Java gave away the JVM, Python its interpreter, and C had legions of sweaty memory-management nerds hammering away at compilers.
Making money on software is innately challenging because it is infinitely reproducible, and the parts that aren't (like time and expertise) need to be more expensive to make building the "free" product worth the effort--making them unattainable and unattractive to the people who just want the free stuff.
Datastar CEO@DelaneyGillilan
Interesting discussion. We are trying a different approach. Making a non-profit with a Pro option for enterprise things. Either get tax free donations or a set of tools you pay for; up to you. It'll surely make some mad but they don't have better ideas youtube.com/watch?v=OrHorw…
English
@AngelicaOung Yes, having seen several people playing these kinds of games, at some point, possibly this very case, I've tipped over into "Never donate to people I don't know, it's almost certainly a scam or trick of some kind".
English
EAs scare me.
Not because they are harboring evil intentions, but because they do not see beyond the immediate consequences of their own personal actions.
This guy had one go-fund-mes. One for Givewell and one for himself. He then gave money from both to go-fund-me. That’s actually fine. He could have done that without dishonor…it was his money after all. He can do with it as he likes.
But he has to announce it to everyone, breaking the social contract and making our world just that little bit more low-trust. Believe me. That impact is real. But because he can’t put it in his little mental ledger it doesn’t count.
Nicholas Decker@captgouda24
With apologies, I lied. I gave it all away. Go support Givewell folks! It does the world a lot of good.
English
@simonsarris Have you tried printing a webpage recently? They are still basically unusable when it comes to printing out pages, chrome will often cut a line of text in half between pages.
English
(it also tells us that early browsers were bad at printing and early web tools were bad at authoring but that excuse has long since passed)
English
What's weird is that this is exactly why HTML and the idea of "links" was invented
This tells us something about academia and it's products
Tom Dörr@tom_doerr
Is there any good reason for publishing papers as PDF instead of HTML? Imagine how much more accessible academic knowledge would be if it were a Wikipedia style web of papers
English
@hecubian_devil So many things should be like this. I would do academic funding the same, rather than try to distinguish which 5 out of 20 “rated excellent” grants should be funded.
English
@jan_ruettinger @Linahuaa One way they pay for it is equipment.
I don’t see get a doctor appointment in China without first getting a blood test. It seems this greatly helps the doctor, they can see about 22 different measurements. The test takes about 20 minutes to do, from sample being taken.
English
That's cool but who pays for it in the end?
There is the public healthcare system in Europe which is underfunded and comes with long wait times. Then there is the private healthcare system in the US which is crazy expensive but has no wait times.
How does China solve it? Do doctors just not make that much money? If so, how do you make sure that you get smart people to become doctors?
English
Healthcare is so efficient in China, Chinese folks do checkups all the time.
Their arm hurts a bit- they go to the doctor.
Their blood fat is a bit higher than normal- they go to the doctor like 4 times in 2 months.
Their pee looks a bit cloudy for two days in the row- they go to the doctor.
I personally had more blood tests in the past 3 years in China than I had in 20 years in Germany.
Getting an appointment in Germany is such a hassle- I just don't bother to go, and hope whatever is plaguing me will go away by itself.
In China, whenever I feel uneasy, I can see a doctor within like an hour (including 20 min travel time) without any appointment.
English
@_trish_07 The fact that after years about the only thing they removed was std::random_shuffle, a perfectly fine function doing no-one any harm, while adding a pile of new footguns (like *v on std::optional being undefined when the optional is empty) was just amazing.
English
@thinkingshivers @Aella_Girl Both ends of that scale are embrassing. How are over 40% of british women losing to a housecat?
English
@maaruin210 @hecubian_devil Based on my limited experience as a foreigner, MAGA believe in the mission. Some of them don’t like Trump, or Elon, or other senior people — but they are willing to work anyone to achieve their goals.
Looking in this thread, the left seems to take the almost exact opposite view?
English
@patio11 Wait, doesn’t this graph literally show the poorest are paying proportionally the highest amount?
English
This paper also has the following result, which is underappreciated:
When you consider *all* sources of card income / costs, including interest and net interchange, *both ends of the spectrum* subsidize *the middle of the FICO distribution.*
English
The Atlantic has an interesting piece on credit card processing. The thesis is that interchange fees redistribute money from poor to rich.
I do not subscribe to this thesis.
x.com/TheAtlantic/st…
The Atlantic@TheAtlantic
“The credit-card market has quietly transformed into two credit-card markets: one offering generous benefits to wealthy Americans, the other offering expensive debt to the poor, with the latter subsidizing the former,” @AnnieLowrey writes: theatln.tc/sZZzl0mg
English
@JFPuget @ai_for_success One common thing I hear all academics say is that coming up with good questions is much harder than coming up with good answers.
So far LLMs seem are depressingly bad at coming up with good questions -- I, and others, have certainly tried.
English
@ai_for_success I'll wait for the day AI comes with something we did not know already.
English
AI will accomplish things we can't even imagine, we're just getting started.
🍓🍓🍓@iruletheworldmo
white pill
English
What a Nazi salute can achieve is dramatic.
Markets & Mayhem@Mayhem4Markets
Ouch! JP Morgan expects $TSLA to report its worst quarterly deliveries in three years: “We struggle to think of anything analogous in the history of the automotive industry, in which a brand has lost so much value so quickly."
English
@JFPuget my point is many people on the left know ellon is not a nazzii but yet still call him that because of one hand gesture, I guess slight difference would be to say (he did a nazzii salute but he is not a nazzii) but most people just straight call him nazzii coz of that one gesture
English
@tsoding I think it effects the libraries a lot.
Haskell libraries introduce a million types. C++ libraries get obsessed with adding another template specialisation to gain 2% more speed. Pure Python libraries tend to care more about good docs than speed, as its going to be slow anyway.
English
It's weird to me that people put so much emphasis on The Community of a language. When I learn a language I just read docs and source code. No where in this process The Community is involved. I think people looking for The Community are children looking for the kindergarten.
English
@rfsouza @rfleury Cool, I've not really used C#, but this motivates me to try it.
I realise using this too much could create a horrible mess, but when I've wanted it in the past (in Rust and C++), it's really annoying to work around, particularly when in the middle of a big rewrite / rearrange.
English
The bizarre part of this debate is that static typing arises first at the low level because it isn’t an extra “correctness” or “safety checking” feature, it’s a fundamental property of how two static codepaths agree on an encoding. You can literally never get rid of it.
A “dynamically typed system” is just saying “all input/output types are a generic object type—all accesses are dynamically checked depending on the value” (and usually in a language that fast paths these operations with succinct syntax).
It’s not that there isn’t value in introducing dynamism. You can do so within a statically typed system. It’s that you are constantly working with both styles, and in a statically typed system you can choose how much dynamism you want in each case.
Samswara@samswoora
People who hate strict typing/conpilers are literally so bad at coding and just want to run and watch their bad code fail rather than just make it work. Strong typing brings you closer to God
English
@ellie_huxtable You know your program is truly popular when Linux distros decide you are successful enough to deserve “special attention”, and they break everything. Congratulations! I’ve unfortunately not found a way to stop them :/
English
older linux distros making huge patches to our codebase
things break
users ask us for support
:(
English
@louiec @probnstat Thanks for analyzing the part of the question I also found unclear :)
English
@probnstat I prefer the 3/4 answer which assumes that he lies randomly.
We know nothing about the roll itself. All we know is that he says it was 6. Since he tells the truth 3/4 of the time, I conclude that the probability that it was actually 6 is 3/4.
English
A man is known to speak the truth 3 out of 4 times. He throws a die and reports that it is a six. Find the probability that it is actually a six.
English