Sorare is a fantasy sports gaming experience and marketplace featuring officially licensed digital player cards (minted as NFTs).
Tech Stack: PostgreSQL / Rails / GraphQL / React
We're hiring a replacement for a colleague who's leaving for a world tour.
If you want to read the objective offer you can find it below, but here is the subjective view of what makes me enjoy my job as a sofware engineer there:
- very competent and kind coworkers
- small team that moves fast and decides quickly
- passionate user community
- top work conditions
- bonus: as a sports fan this is heaven
Don't hesitate to reach to me if you have any question:hugo [at] sorare.com.
For me, as someone who has to use stock photography often, this doesn’t make much sense. What I really value is an AI system that sorts pictures fast. I don’t need exactly the depiction I’m looking for, but I want to discover a look and style that fits well with the project. This is not what Dall•e is capable at providing. The first stock photography dall•e service has a big chance of turning into the Microsoft clip art of photography more than a disruptor of the market.
From an artistic point of view, the value is zero. Dall•e pictures will never have any Aura to them. And for an artist to use them artistically Dall•e is “too easy”, as the artist won’t have produced or worked on the model itself.
All in all, Dall•e is the most exciting, technologically advanced, and beautifully useless AI project I’ve ever seen.
Today my personal biggest use of AI is for generating photos of faces as profile pictures for tabletop rpg character sheets. I don't need them to be perfect, I just need quick, easy, cheap content that's good enough.
Tomorrow I might use Dall-E to create illustrations for my blog posts, or depict environments for my tabletop campaigns.
They might be "too easy", but they'll be cheap and quick and good enough for the purpose.
You’re thinking about it from the perspective of a professional artist. But put yourself in the shoes of someone who’s not an artist and would like free art. The implications for artists who want to get paid might be bad, but even as an occasional artist myself, I have to admit most of the results in the article would be hard to distinguish as computer generated, if shown on their own in a different context. I don’t agree that AI pictures will never have any Aura - in some sense the opposite is true, some amount of Aura may be guaranteed because the inputs have it. The Aura is only lost when you assign the ‘machine generated’ narrative, when you know the story and already have opinions about whether it’s good.
People who want cheap but unique abstract art for their house, small business owners who need ad clip art but don’t have the budget for a designer, even artists using DallE queries to generate launchpad ideas, I can imagine many reasons DallE would get used. But also don’t forget to imagine the future, this is one of the first and it’s bound to be improved until you can’t distinguish it from “real” art.
There’s a valid point in your comment about how artists currently look at AI, and a valid question about what AI produced art actually means, so I’m throwing you an upvote because I don’t think you deserved to get downvoted for sharing your opinion. Maybe more importantly, AI generated art could challenge and rekindle the age old debate about what art actually is, and bring a new perspective to the table.
Thank you. I actually basically said that I'm in awe at what Dall•e is capable of, it's just that we've seen a lot of these watershed moments for AI creativity, and to be fair it's always the people who don't really work with art that call them this way.
You know what's really impressive, that just got out? Adobe's Premiere Pro Auto Color feature based on Adobe Sensei. Ask a colorist or a filmmaker how it works for them. Is that something that was possible because of fringe research like Dall•E? Honestly, no.
What you say about supposed watershed moments is true, but I don't think it leads to the conclusion you want it to.
As someone who designs and makes fine furniture (not the way I make my living, but I do know the business), I know that there will likely always be some demand for custom, hand made pieces, because people assign value both to getting exactly what they want, and to the fact that something is human and hand made. When people who value those things have enough money, they'll spring for that value. But I'm well aware that the bespoke and hand made will never be more than a 3rd decimal place rounding error in the ocean of factory made furniture that fills our homes and workspaces. Images can go the same way, and will, if the price differential is right. Most images in the "image marketplace" are not art, after all, but rather anonymous commodity visuals augmenting the presentation of an idea, or product.
My partner is an hobbyist painter and often looks for hours on Google image search for a source image to paint off of. He was excited to get access to Dall-e because it would give him a vast new source of stock images he could do style transfer on to give him a better source for his art. It is very common in art classes to mash something up in Photoshop to use as reference. This tool is far from useless.
think of dall•e like stepping stones to create the finish product.
this isn't supposed to make designers useless but rather make them more creative.
dall*e can be used as an inspiration to combine multiple images & create their own. it can be used to steal color combination that is aesthetically pleasing on the eyes.
if you think dall•e is useless, you haven't seen the images shared.
the realistic ones are hidden by openai but they can generate a lot of stuff that we won't know until an open-source version comes along.
The company did not hire me to do scientific computing, rather build the beginning of an ambitious software platform which, beside other things, will leverage a scientific computing component. As we are a start up, I'm more than happy to go out my comfort zone to work on the first steps of this component, before we hire specialist in the years to come.
Thanks for your advices on MATLAB and Python, and I'll take a look at OpenMPI. Who knows, maybe I may have the option to understand what the code is doing if I put some work into it :-)
"And quite frankly, I hope I don’t have to give another live lecture ever again. I’ll tape it; I’ll have them watch it — I know most of them are going to watch it at 1.5 or 2x speed. I’ll sound like Alvin and the Chipmunks, but that’s OK. Then we can spend the time in class actually working on problems or discussing some of the issues that I brought up."
Beside the joke, I find the concept of ingesting content at your pace and then talking about it in class very interesting. No idea if the knowledge will stick in mind as well as classic lectures, but I would definitely have been seduced by this proposition when I was a student.
", I find the concept of ingesting content at your pace and then talking about it in class very interesting."
It's called a "flipped-classroom".
The teacher gives you material to read or watch at home before the class, and then in the class it just asks if anyone had any questions, and they can go over these together.
As a teacher it is super useful because you get to learn first hand which concepts the student didn't understand.
And as a student you can read or watch something whenever you want at the speed that you want, and if you don't understand something you can ask about it in class.
With some experience, teachers can also use the flipped-classroom model to focus only on the hardest part of a particular lecture. You can rely that some student is going to have trouble with it at home, so you can prepare before hand to the lecture to how to address that particular question (and once you have done the class a couple of times, you get pretty good at it).
There is no need to waste teacher and student time during 1:1 interactions on "easy" stuff that everyone can just read at home.
How do you deal with shy students, or ones who don't like to participate? This model sounds interesting, but I can't imagine exclusively using this method.
> How do you deal with shy students, or ones who don't like to participate?
With students who aren't participating in the current model? They will (have the opportunity to) do better than they are now, as when they watch prerecorded lectures they can skip back to parts they didn't understand. As opposed to their current behavior of not speaking up and, therefore, not being able to go over the content again.
> How do you deal with shy students, or ones who don't like to participate?
This method allows students to learn at home at their own pace, and use face-time with experts to asks questions.
Just learning at home at your own pace is a net win for all students.
Those who are a little shy, come to class, and learn something from the questions of others.
Those who are more shy, don't even come to class. They can read the stuff at home, and have ~2 hours extra per week to do other stuff.
We have 1:1 tutoring 2 per week (students have to get an online ticket), and they can also ask us privately per email, but in my experience, students that are "too shy" to go to class and ask questions in front of others are also often "too shy" to asks questions privately.
The other aspect of these is homeworks. Our homeworks allow groups, but they are thought of as "individual homeworks". We recommend people do them on their own. This means that non-shy people as a group get them done faster and often better, but it also means that shy people doing them on their own, end up learning more and performing better in the exams.
We don't grade homeworks, just pass or not pass, but passing them is required to be allowed to take the exam.
Still, we do internally grade them, and people that do them well alone do perform better in the exam.
---
In my opinion, the only "downside" for "shy" students, is that since we give them the freedom to not attend _any_ classes, they might decide not to do so. That means, no access to new friends, new colleagues, in person learning groups, isolation, and well understanding the hardest parts of the material might cost them more time if they are "on their own".
Most still make it. Some do find learning groups or new friends in the class forum, some do better than many students that do attend classes, etc.
Having taught many flipped and non-flipped (classic) classes, I never had the feeling that forcing people to attend class fixed any issues related to "shy-ness".
Students are how they are, I'm neither their parent, nor counselor, nor psychologist, nor am I qualified to do any of these things. I'm their teacher for a specific kind of content. I want the content to be useful and applicable to them, so that they can use it in their careers to be successful, whatever successful means to each of them. I also want them to enjoy the content, so that they can learn it easier.
For me this means giving them as many options as possible, so that they can master the content in whatever way makes the most sense to each of them.
I may be in the minority but I really don't care for listening to things sped up. If it's really just about maximizing data transfer, I'd probably rather read a good transcript at that point.
Unfortunately, as I'm sure you've discovered, "good transcripts" don't exist unless someone puts in the effort. :(
I don't mind audio, but I definitely don't understand the value of "talking head" style videos. They're, if not the worst possible information density possible, pretty darn close. A couple hundred megabytes to see someone speaking a few dozen kilobytes of text, and I still can't search it or skim it to see if it covers what I want.
A good transcript isn't really that hard. You just need to pay someone--the ML ones just aren't good enough to publish--and do some light editing. I usually do them for my podcasts. (Though the ML transcripts are good enough for skimming content and I use them if I'm just going to be using some excerpts from the audio.)
I actually find that, for talks with slides, I do like having the inset speaker video. What I tend to do is to start off with a full screen talking head intro with mostly the slides/multimedia and a small speaker video.
Totally agree that a good transcript is way better in almost all cases (except where actual moving visual demo actually helps).
However, are there any good automated transcription tools that can make a good transcript from an imperfect YouTube video? Most of what I've seen is good for humor, and I've even seen some used in a legal deposition context that are downright dangerous (inserting most likely word or phrase in place of what was actually said, so actively decreasing and corrupting info content). This might not be as disastrous for a lecture where there are also office hours, but... anything good out there?
Maybe you'd like to try us out at https://3scri.be. Like I offered above, if you'd like to sign up and drop us a line at support@3scri.be mentioning this thread I can add a hour on top of the 30 minutes trial.
Well, we're new and we think we've the most accurate transcription engine on the market. Why not try us out at https://3scri.be. We give half a hour for free but if you sign up and drop us a line at support@3scri.be I can add on an extra hour for you.
I have had maths lectures where that just doesn't work, so I'd question the "anyone". You need the extra time to fill in the steps the prof left out, because they are obvious to him.
(And yes, I usually also go through presentations at 2x).
There is a forcing function to in-person lectures; I can't tell you how many events I've basically skipped in the past year because I can catch up on the videos anytime--and, of course, I mostly don't. But, in principle, depending upon the size of the class there's a lot to be said for watching the lecture on video and using class time for discussion, project work, etc.
There's also something ritualistic about in-person lectures -- in a good way, like having a morning cup of coffee. We know that humans are not good at multi-tasking, so there is tremendous value in a dedicated time and place to think about one thing only in the presence of other people who are also thinking about that one thing.
Of course, it's easy to squander that opportunity by delivering a rote, feelingless lecture. And if you're just going to lay out the facts, why not record them? In that case, videos have many advantages. But a live lecture is also an opportunity to get people excited about what you're teaching.
Post-pandemic, my plan is to take a hybrid approach. Technical details, like proofs, belong in pre-recorded videos watched out-of-class. But the main conceptual thread should be delivered in live lecture, where I can give it the energy and life it deserves.
Pre covid, we had one professor who did lectures with videos (of him presenting) combined with in-person tasks and discussions. Each week you had to take a 10-minute test on the content of that weeks videos. It worked quite well.
The biggest disadvantage was that with a live audience the professor got cues from the audience if we understood him or if he had to slow down. In his videos he constantly assumed he had to repeat everything slightly differently, so you had to watch at 1.5x speed to get to something bearable.
Yeah, particularly with a smaller audience that can definitely be a problem. I'm better at it than I was a year ago but I still find doing a video without an audience can be a bit challenging. And I definitely can't course correct the way I can if I see a bunch of puzzled expressions in the front row of a room.
On the flip side, you can redo sections and easily insert multimedia and just mix up the talking head format.
The best class I ever had was one where the teacher explained things for half the time and then we worked on homework for the other half.
I think that is probably the best model. Letting students watch lectures on their own time is hardly better than expecting them to read on their own time.
Years ago I heard of experimental elementary/high school classrooms where students were expected to watch lecture videos at home and what would be “homework” would instead be the focus of the whole time in the classroom.
Students were expected to help each other with the material and only go to the teacher if they were collectively stumped.
I can imaging all kinds of bad scenarios were someone tried to switch a disruptive class to this routine overnight. But, with good onboarding and ramp up, that plan sounded like a dream to me!
College classes are often split into lectures and precepts/discussions/office hours. I think the thing holding back your idea is the expectation of being taught in-person by a professor. For large classes, that’s only really possible in a lecture format.
My personal favorite method of learning was a TA that went through the material, meticulously calling on each person in order. You could skip, but it was still obvious if you were playing a game or browsing Reddit. Eventually you learned that you really didn’t have all the answers and that paying attention was probably a good idea. It also got people comfortable giving a wrong answer and that really helped everyone with the learning process.
For a novice like me, it sounds a bit like quantum computing, where rather than "digitalizing" a physical problem like we do with classic computing, we use physical properties of elements (like the spin of an electron) to solve it. How far does this analogy go ?
About halfway between digital and quantum computers.
Wholly quantum - uses both wave and particle behaviours and entanglement
Wave computation like in article - just waves
Digital computation - just impulses or currents in conductors.
Sorare is a fantasy sports gaming experience and marketplace featuring officially licensed digital player cards (minted as NFTs).
Tech Stack: PostgreSQL / Rails / GraphQL / React
We're hiring a replacement for a colleague who's leaving for a world tour.
If you want to read the objective offer you can find it below, but here is the subjective view of what makes me enjoy my job as a sofware engineer there:
- very competent and kind coworkers
- small team that moves fast and decides quickly
- passionate user community
- top work conditions
- bonus: as a sports fan this is heaven
Don't hesitate to reach to me if you have any question:hugo [at] sorare.com.
Read more: https://sorare.com/careers?jid=c81d9cc6-4fcc-472a-a927-e6941...