[1] https://www.youtube.com/watch?v=JQdhL-XuKLY
[2] https://www.youtube.com/watch?v=TCVWDHJN288
[3] https://www.youtube.com/watch?v=co7KgV2edvI
2,593 karma · joined September 29, 2015
[1] https://www.youtube.com/watch?v=JQdhL-XuKLY
[2] https://www.youtube.com/watch?v=TCVWDHJN288
[3] https://www.youtube.com/watch?v=co7KgV2edvI
I have been enjoying Nier Automata's game soundtrack recently. EDM, Porter Robinson and related is also a favorite of mine. Citypop, jpop, latin pop, and kpop all help too. The music feels familiar but I don't get distracted by the words because I don't speak the languages well or at all.
When I need to focus for a long time, I put on a live performance of an artist I like and know the songs of. Daft Punk's Alive 2007, Porter Robinson's Worlds, and Yo Yo Ma's performance at the 2015 BBC Proms of Bach's Cello Suites are my main three. DJ Sets from things like Ultra, EDC, Coachella, BBC Radio One are nice as because the songs flow really well from one to the next.
I found that slowing down music has helped it be less distracting to me. Not all music will sound good, but slowed classical generally does. 75% to 80% speed playback of the Cello Suites is really nice.
I am enamored with, what feels like, youtube originated genres. "Dark Academia" is a vibe and playlist genre that has helped me discover a lot more classical and adjacent music. I enjoy the playlist titles and creative writing in the comments too.
High schools and middle schools have started giving out a laptop to every student in my district. The teachers use google classroom. I don't know if a $1500 device will be better than a $800 device for grades, but even a $600 used device will be much better than going pen and paper only. There is group pressure to get an apple product, but that also depends on the major. In bio or psychology classes there was some elitism when I was the only one who pulled out a thinkpad, some classmates associated thick to business to slow and old. After I switched to CS the topic didn't come up again.
Every single one of my classes in college provided power points, even math and cs. I didn't need to worry about copying down the PPT while the teacher was lecturing. I took one class with a teacher who refused to provide a copy early and I meaningfully struggled more to learn the material. I switched sections/teachers in order to make sure that never happened to me again.
Depending on the comfort student, 2 in 1s have gotten good enough to where they could replace paper and pencil completely. The only time they would need to write on paper is for exams. I personally think iPads are a bad investment regardless of the student and their major. The software just isn't there, regardless of the model.
My tags for Hindi songs were always in English. Because of the way Hindi songs are credited, searching by tags like artist was always annoying, but that's not a problem specific to google music. I did have times where Japanese track names got translated to English in the old Google Play Music, but I haven't experienced that with Youtube Music.
The other alternative is self hosting, plex and plexamp are popular.
https://support.google.com/youtubemusic/thread/52908732/faq-...
There is a place for AI art generation and there is a place for artists. NFTs were interesting in how they overvalued otherwise mediocre art. These models are interesting in how they bring down the cost and experience needed for making derivative art.
To me, the creativity still lies in someone being able to produce something meaningful. Art is about being able to convey ideas in a way that's impossible to communicate in some other way. An artist is someone that makes art. In that sense everyone who has generated art is an artist. Oversaturating the world with derivative art will only make novel things stand out more.
It's very hard to share a nuanced take on this topic because this argument has become framed in such a binary way. With something like medicine, the value of a doctor's opinion is very clear to a layperson. But when it comes to art, the value of an artist's perspective is not clear at all. However, I think making parallels to music makes it clear for me. AI generated music will replace elevator music at best, but I don't think the public fears ai models will ever replace musicians. At most ai will complement the art creation process. The "soul" and novelty in art will always come from an idea another human wants to communicate.
This feels like the question they expect you to finish in the given time, the "easy" question. The question I was ask was a bit harder, but in hindsight my unfamiliarity with the question made it feel harder than it was. The minimum expectation is a mostly complete and actually working implementation. I can't confirm this but my understanding is missing major edge cases means failing the interview. I am surprised the interviee started typing as quickly as she did, I believe I was asked to write more detailed pseudo code first.
There is a second more advanced technical interview asked after lunch. I didn't interview online, I interviewed in person so I am sure that changes some things.
I did not pass every stage of the onsite, but it was one of my better interviews experiences. No one applying to Jane Street does so without knowing how rigorous the process is, I don't hold it against them. It felt like everyone I met wanted me to pass the interview. I can't say that about every interview I have been on.
The only time I have heard of "rampant" abuse of stimulants for work, it's been in online forums. To me, these stories sound as real as the ones about micro dosing LSD. I know people that have abused stimulants as college students to pull all-nighters in the week before finals. I haven't heard of anyone abusing stimulants after graduating, I imagine most people realize no job is worth abusing medications and taxing their body in that way.
I absolutely hate how I am made to feel guilty for needing stimulant medication, it feels like a version of imposter syndrome where I constantly question my own diagnosis. There genuinely are no good representation of the struggles of ADHD or the benefits of stimulant medication in media. The stigma that Adderall is "meth for kids" shows a severe lack in empathy or understanding of what ADHD is and how medication/chemical groups work.
The people I know, even diagnosed, in NYC finance and FAANG tier companies told me they don't want to use anything stronger than coffee for crunch time. With every stimulants redose used to push away tiredness, the side effects like blunted emotions become exponentially worse. Some of my diagnosed friends choose to avoid stimulants completely because of emotional blunting and other side effects that happens to them at therapeutic doses. Sleep deprivation makes all stimulant medication a lot less effective. No drug can help you recover the way sleep does. There have been morning where I fell asleep after taking my daily dose of Adderall.
People are more comfortable now sharing their diagnosis and if they are medicated. The stigma around adult ADHD is starting to improve but isn't great. Many experts in the field used to think that people outgrew ADHD as they grew up but studies have shown that's completely untrue.[1] There is still a very strong stigma regarding ADHD and being medicated. ADHD isn't visible and in the last year there have been people that have told me I don't look like I need to be medicated or I am "smart enough" unmedicated. To those people, me taking my prescribed dose is the same as me abusing stimulants. There have always fears and stigma around people requiring medication for mental health conditions. Unless you know examples personally, I would take anecdotes with a huge grain of salt.
Stimulants make things easier to do but they don't give you momentum or energy to start something new, that's still something you have to do for yourself. Adderall feels close to a strong coffee taken together with ephedrine or Sudafed, not meth.[2] In movies and TV, a character starting stimulants is used as a plot device to help them overcome and obstacle or as a source of comedy, sometimes a deux ex machina.
There's a shared theory among me and my friends. The field of CS is naturally more attractive to people with ADHD. Software engineering work is structured in a way that would make it easier to manage symptoms for someone with ADHD. Things like hyperfocus, a symptom of ADHD, can even make things easier to learn. I don't know if the field itself is generally more accepting of individual differences. I am sure there's selection bias happening, but it seems reasonable to me. It would explain why there's a larger subset of software engineers have an ADHD diagnosis compared to some other professions.
[1] https://www.nytimes.com/2020/11/13/well/family/is-it-possibl...
Maybe the difference between a typical classroom and my tutoring sessions is a student feels some sort of intrinsic motivation to be there and pay attention. Once the student has an understanding in the value of paying attention and actively chooses to pay attention, they are naturally able to have self control.
I appreciate you pushing calculator literacy. My school district is old school in both technology and methods. I know teachers that actively encourage using a ti84 over 89, so most people don't touch a real CAS. My go to party trick is still the solve() fuction. We have a very strong CS program, we have honors weighted classes in data structures, mobile development, and game development which has AP CS as a prereq. We also have AP weight classes for Multivar Calc, Linear Algebra and DifEQ which all come after AP Calc BC. The students are smart but the calculator literacy is just not there.
I think Excel literacy is the perfect bridge for getting students to start using database logic without a database. Index/match or vlookup are just constrained database queries. Once a student is able to connect their calculator CAS knowledge to excel, there are very few excel sheets they will come across that they won't understand. We have basic excel literacy because most AP science teachers expect typed lab reports, the exception is chem lab notebooks, there's usually one lab where there's some stat analysis and every lab requires some sort of graph. To me excel is the next logical step after mastering a calculator CAS.
Desmos is such a great tool, I haven't played with this specific feature too much but I am so impressed by the animation feature. https://teacher.desmos.com/activitybuilder/custom/6062093999...
The average speed of the available internet has increased exponentially since 1998. On dialup and even dsl, a 480p video is a struggle if possible at all, but over covid it became mandatory for home internet to support 3+ simultaneous video calls every single day, with some 4k Netflix/Youtube streams during lunch or for background music. My sister actually had to do a video call demonstrating a golf swing from her phone for online gym class. In 1998 99% of consumer internet speeds were too slow to stream an audio file in high quality without buffering.
Many middle and high schools have started using fleets of Chromebook for things like google classroom, textbooks, and submitting daily homework. Using computers isn't a special event that requires a class visit the computer lab. Everyone has adapted to these new trends for communication. It's fine to feel nostalgic and appreciate certain benefits in how things were in the past, but it doesn't make sense to apply the standards of 1998 to today's classrooms.
But on the other hand, I don't agree about a phone free environment being more productive.
The students that completely check out, often check out regardless of if there's a computer. Computers don't encourage that behavior, computers make it easier to hid it.
Teachers have always had a hard time detecting terrible things that happen. Even the school from the article, a very expensive, very small, low teacher student ratio boarding school had a fight break out between students.
I don't know if students are less likely to socialize with other students outside their preferred social group based on phones. I could argue that phones allow group chats and social networks of much bigger friend groups. The network will weigh a classmate you share a single class with equally to a classmate you share every class with.
It feels intuitive to blame phones as somehow causing maladaptive behavior, but if you include iPad and laptops into the argument, it no longer feels binary. Using individually assigned iPad and laptops have become as essential as textbooks were in the past. Even for middle schoolers, my school district has shifted to google classroom for assignments and digital textbooks.
I remember I was one of maybe a handful kids who brought a personal laptop to every class, senior year in high school. I was taking all AP classes except gym, so my teachers gave me the benefit of the doubt and weren't concerned about my work ethic. My calc teacher even introduced me to latex. Most of my notes were typed up that year. I know I would have had a much much harder time paying attention in calc if I was handwriting my notes. I didn't get distracted in classes just because I had a laptop, I got distracted when I was in a class where the teacher clearly didn't care about what they were teaching or when I didn't see the value in learning it.
Personally I am much much more productive working with some background music. Absolut silence and a phone/computer free environment makes it easier for me to lose focus. A caveat, I have ADHD, I know my experiences won't translate directly to the general population.
Unless you believe that students are objectively unable to learn how to self moderate technology usage, I see no point in delaying it. It's a skill they will need to learn at some point in their life. For those who struggle to learn it, the lesson will be much costlier in college. I believe that if it's normalized and introduced early, more students would pick it up quicker. I recognize this is ignoring the issue of how to build work ethic and passion to learn, those things aren't in direct control of a teacher.
I think there needs to be a different much bigger discussion on reorienting teaching and testing to be computer aided. Senior year, in physics I got to do a computer aided lab where I used a depth sensor to measure and plot the displacement of an object. From there it was possible to see the gravitational constant. That tangibility is so hard to get in physics and it would be impossible by hand.
It feels appalling to me that calculator/excel literacy is so poor for graduating high schoolers and undergrads. Part of it stems from an early association of the usage of calculators with cheating. There is no significant benefit in learning things like multiplication or division through rote arithmetic algorithms. Students enter a workforce where knowing hlookup, vlookup, and sumif in excel will cover 99% of the calculations they will need to do. But many of them graduate not understanding how to use a calculator and ratios to figure out unit conversion for things like baking.
It's fair to ask, how can teachers effectively grade students for take home assignments/essays in a time when students can use wolfram alpha or run a pre-trained GPT3/OPT models? I don't have an answer for that. But on the other side, standard teaching tools are showing their age. Textbooks are the worst examples. There are online programming textbooks that have embedded exercises where you compile and run code inline with the text and some algorithms even have interactive animations. Why don't we have the same for math exercises? 3Blue1Brown videos are prime examples of how much richer teaching and learning can be if technology is used properly.
My perspective is that for all current and future generation of children, removing them from social media will be harmful than helpful because of how isolating it will be. Some foundational social skills are now built online instead of outdoors. With the hypoconnectivity, kids today have to learn skills that adults today didn't need to earn. Anecdotally I gained an appreciation for online social skills when I got a smartphone the senior year of HS. I didn't realize how much I had been left out of things simply because I wasn't able to join social networks built on Snapchat and Instagram. Discord has become the main network for most kids now.
The school did this as a response to a student live-streaming a fight. Banning phones is not solving the underlying problem there, it's just covering it up. It's not a big school, with only 57 students.
I also strongly dislike the continued narrative of pushing "dopamine" as an "addiction" that you control with screen usage. Dopamine system dysfunction most commonly presents as ADHD, it's not a bad habit that can be fixed. I understand phones become an easy way to escape and it becomes a negative coping habit for some, and it's especially important to teach teenagers how to address it, but pretending less screen time is always a good thing is a very pre-broadband internet mindset. It's not the direction modern society is headed.
I got two different tracks to work, and it's clear that one was a lot harder to process than the other. It took noticeably more time on the second one, to start and the CPU utilization was higher as well. They were both instrumental tracks in the same format and around the same length. The one simpler to process was the instrumental of Britney Spear's Baby One More Time. The harder one was Porter Robinson's Divinity.
Neither audio had an effect similar to the one from the demo video, but were interesting regardless. They both looked like how I imagine sound waves echo and bounce around if contained in a cube shape.
I appreciate the notebook writeup where you described the goals because the visualization wasn't inherently intuitive with the sound. I chose much more complex tones than your demo. I imagine the feature extraction is much easier on isolated sounds. This reminds me a lot of project milkdrop and so I was expecting it to be closer to that but in 3d. That was probably a misunderstanding on my part of the goals for this.
I think exposing more parameters about how features get mapped and scaled would be really helpful in making it feel more intuitive. Zooming the cube in and out is nice but didn't seem to help convey more information with the tracks I chose. If anything it got in the way because on my computer the zoom sensitivity was very very high.
I look forward to seeing where this goes.
AGI isn't defined as strictly as it needs to be. The current test, as well as the article, qualify strong AI as being or superseding "human level". Every single category, the AGI is being contrasted with the skills of a human. I am arguing that for some people, the current state of weak AI is useful enough that it could be mistaken for the first steps of strong AI.
Weak AGI will be the first language model that is able to somehow influence the thoughts of the person communicating with it, I think that is the milestone of AGI. From my experience with GPT-Neo and OPT and using it to help write stories or make chatbots, the responses are still very reactionary. In that sense, adding more parameters helps the model give a more coherent response, but it's still a response.
For many people AGI is already here. They have Siri and Alexa, AI art, GPT3 based therapy/chat bot, a chat bot that will help them write a book, and "soon" will drive their car for them. The Google Duplex assistant demo where it booked an appointment made it clear to me that for some people, that's the smartest they need AI to be. Anything more is just extra.
I am really excited about how far we're going to push AI in my lifetime, but I also realized that for many scenarios, weak AGI is enough. People will project their own expectations and essentially help fool themselves. I don't know if testing a model to perform the same as a human matters in some ways.
There's one big skill that I personally value the most when it comes to qualifying hard AI, and that's the ability of it to make me laugh based comedic irony. I wonder what that model would look like.
Elon Musk may have questionable attributes but hiring or assessing engineers poorly isn't one of them.
https://www.bmj.com/content/364/bmj.k5233
This article is arguing ECT as acute treatment for resistant depression. In most, if not all, cases a patient would become eligible for ketamine treatment at the same time if not sooner than ECT. Memory loss directly due to ECT is the reason most people are strongly against using it. Treatment options are based on minimizing risks, not guaranteeing success. Unless ketamine is contraindicated, it will always be a safer treatment than ECT.