Otterly99

The actual content yes, but the recommendation engine switching from "Here is a video that is kind of like this one, you might like it" to "Here is a video that we predict will keep you on this website for as long as possible" definitely soured the experience.

Related: Steve Brunton is also releasing a new book soon https://x.com/eigensteve/status/2055341831702057189

It is in the same vein but more applied to engineering and sciences. If you don't know him, I definitely recommend checking out his youtube channel, his lectures on Physics-inspired neural networks are top notch.

I am curious thought. If nobody is qualified to answer, how would your team have approached the problem without any AI?

May I introduce you to vocaloid concerts?

Do you even bother to click on links before asking a question?

There is a whole paragraph on "Statistical analysis" that even provides five supplementary methods (S1.5, S2.2 and S2.4-6) if you want detailed information.

This is impressive and, I think, complements well whatever benchmark is the hottest right now.

Unless you're working on private data, is it such a problem to let Cursor train on your data?

I personally work mostly on small, unoriginal projects. I don't really mind letting Cursor keep the data for training if it leads to a better product for me.

I think it can be more useful to think in terms of hours of lessons rather than time spent learning. My partner is learning french right now at a private school and they have occasional classes (1h30 once a week), middle ground (2h twice a week) or intensive classes (4h five times a week). The program is always 70h, but you can either do it in a month, a semester or a year.

Having learned czech, I only had 1h30 a week and it took me 5 years to get to B1. I agree that the steps to A2 are much simpler because it focuses more on simple phrases and vocabulary, so you also learn it outside of the classes by reading descriptions, signs, menus. For B1 and up, you definitely need to practice by having deeper discussions with people or reading articles.

The situation is kind of similar in my country (France).

My partner and I lived together in Czech Republic for 7 years where I was working as a scientist (EU to EU, pretty much no paperwork involved). She is american and obtained a "partner of EU citizen" long stay permit. We decided to move back to France and now we have to deal with bureaucracy and it is exhausting.

There are many paths by which an non-EU citizen can get residency in europe and it includes:

- Student visa (9 months, needs to apply long in advance) - Entrepreneur (need to pitch a business, needs 30k€ of savings) - Refugee - Married (the easiest one) - Family member of EU citizen (only applies when the EU citizen is not native to the country you apply to) - Family member of native citizen

We had to apply to the last one, because we only have a civil union (it's a french thing called PACS and it's not recognized as a "proper" union for permits).

We also needed to prove that we had lived together for more than a year (which we also had to do in Czech Republic) so that she could get her one year permit. The fees are opaque (the website says 350€ of tax + 100 € of processing fee but the final bill was 650€). Any document must be translated (cost us around 1000€) and official documents must be verified by apostille (which we learned when we celebrated our civil union in at the french embassy in Czech Republic).

At the end of her permit, she will need to validate an A2 level of profciency in french to apply for a multi year permit (2 to 4 years). Once she has lived 5 years here, she can apply for a permanent residency if she has a B2 level, pass a civil exam and show proofs of involvement in civic life.

All of this to say that the requirements to immigrate in Europe require either a lot of money or a lot of dedication. This really changed my view on the whole "Europe has free border, anybody can get in to steal our job" rhetoric.

Why should right wing populism be the alternative thought? Couldn't we have a decent alternative?

The only skill I genuinely find useful is the /strategic-compact from ecc (everything Claude Code) but you can get the same effect by saying something like "I am going to compact the session, give the next Claude session a summary of what was added or changed and what the next step is". The you can just use /compact <message>. Other than that, just barebone too. ECC is shipped with something like 300 skills, maybe I should just delete it!

I hard the pleasure to assist David Hu's talk a few years ago where he presented his work on feline tongues, frogs tongues and wombat poop. Really fun guy. I always wondered how he got his funding because studying the "physics of animals" must not be on the top of the list of most funding agencies.

I completely agree with this post. After I used it in one session of 300k tokens, I had maybe 3k tokens saved. Plus, if commits really are an issue for you in term of tolen consumption, you can always ask to hand over the reigns and apply the commits yourself as a rule (unless you're operating in a loop).

Sounds like you found peace. Good for you brother/sister.

What are your responsibilities at the co-op? I thought the idea was that people would work a few hours to get access to the product, but you said you work full-time.

Which to me, raises an interesting question:

- How many individual components do you need for it to emerge?

Most people would be okay with saying that individual cells are not conscious, maybe even that tree are not conscious even thought they are made of many cells. Neurons seem to be the determinant factor in deciding whether something could have a consciousness, but again how many do you need? Does growing 1B neurons count as a brain if they are not organized?

Maths skills have been slowly falling even before the advent of LLMs. I have a story but this is anecdotical so take it with a grain of salt.

I was in my 3rd bachelor's year studying physics (France) and overheard a conversation between two of my teachers. They were discussing how they should modify the 1st year program to now include math, because he had been noticing how more and more students were failing the more math-heavy subjects like body and newtonian mechanics. He said that they should now teach (or re-teach) calculus to 1st year students, which was not taught when I entered college (it was assumed that you learned it in high school and we would only cover linear algebra in 1st year).

I can imagine things are only getting worse with students that can now get under the illusion that they know math because they have a tool that can do it for them. Which raises the question: should programs adapt to this, like we adapted to having calculators?

There is also one thing I would like to add, and you can correct me if you disagree: coding benefits much more from thorough planning. Now, I exclusively work by first writing a plan that has well-defined steps and goals, which can of course change over time.

It seems to me like it would be more difficult to achieve with legal documents and, in my experience at least, writing a concrete plan has been the decisive factor that make my AI coding robust (plus all that you mentionned).

I also find it amusing. I also heard a lot of "4.7 is garbage, everybody hates it". Shows you how important proper validation techniques are, not just gut feeling.

Honestly, scientists too. I did 10 years of research after deciding that it was not worth it. There is a vast amount of research with no direct application that gets published under the assumption that more knowledge is always better, but in my experience scientists rarely question the usefulness of their research (because most of the time, they find it interesting, which is motivating enough).

Granted, I am talking about harmless subjects, but there is also the dimension of resource usage that almost no scientist considers (the amount of plastics and chemicals used in biochemistry and cell biology is... concerning).

I think it has taken a rather negative connotation with the development of psychology, marketing and "influencers" which are usually people that try to influence you to buy into something.

I also immediately thought about his book on creativity. Thanks for the talk. For me, instead of staring at a wall, I just take a short walk. I think doing any activity with low mental load helps creativity.

It's the same story since at least 2012. It is well documented in the book "The chaos machine" by Max Fisher.

Facebook employees, journalists and psychologists have studied the phenomenon and Facebook's (as well as Youtube's) response is always the typical "We have done something" to calm the protest, but it's never really the case. It's a constant game of deflecting, delaying, diminishing, denying.

It seems to me that it is broadly the same thing, except they give you the resources to do it and expert knowledge.

I feel like even the phrasing of the original assumption that "we have more bank tellers now that we had before", which seems to imply that ATMs didn't affect or even boost the number of bank tellers is flawed.

If you look at the graph, the number of bank tellers from 1980 to 2010 went from roughly 500k to 550k (a 10% increase). However, the U.S. population grew from 220M to 305M in the same period (a 40% increase). To me, that seems to indicate that less and less people were becoming bank tellers after the invention of the ATM. Although from the graph again, you can see that the correlation is quite poor anyway.

Not sure how that makes sense, it takes a third of the time, but you're only going 5 km/h faster?

Thank you for the link.

I admit, my knowledge of reinforcement learning is a bit outdated so it seemed to me that it was unattainable for a non-specialized model to train efficiently on something like chess, which has a huge state space.

Every scientist I ever met (and myself included) has a backlog of papers to read that never seems to shrink. It really is not trivial to stay up to date on research, even in niche fields, considering the huge volume of research that is being produced.

It is not uncommon for me to read a recently published review and find 2-3 interesting papers in the lot. Plus the daily Google scholar alerts. It can definitely be beneficial to have a LLM summarize a paper. Of course, at this point, one should definitely decide "is this worth reading more carefully?" and actually read at least some parts if needed.

Some jobs that I interviewed replied with an automated email saying that, if I wanted, I could ask for feedback. I always did and none of them replied... This somehow feels even more insulting.

I tried it for 2 days and honestly don't see the usefulness either. Although, the big reason is that I paired it with Claude, which only uses the per token billing method. Here are the few improvement on a simple Claude usage:

- As you mentioned, the message bot thing was kind of cool. - It can browse the internet and act (like posting on MoltBook, which I tried). - It has a a permanent "memory" (loads of .md files, so nothing fancy). - It can be schedulded via cron jobs.

Overall, nothing really impressive. It is very gimmicky and it felt very unsafe the whole time (I had already read about the security issues, but sometimes you gotta live dangerously). The most annoying part was the huge token consumption (conversations start at 20k+ because of all the .md files) and it cost me roughly $12 for a few hours of testing.

This is a typical evolutionary arms race, advertisers come up with better tool to fuck with us, we have to come up with better defense systems.