37 comments

[ 2.9 ms ] story [ 77.7 ms ] thread
Hi HN! I'm happy to answer any questions. I'm also open to feedback if there are suggestions.

I was inspired to try this experiment by Nadia Eghbal's grant program: https://medium.com/@nayafia/5-000-no-strings-attached-9e7b95...

This sounds great! Any particular reason only the MIT license is supported and not Apache 2.0? I am aware there are some differences with regard to patent rights, just genuinely curious, because there might be ambitious projects with the goal to incubate with Apache.
I don't have a strong opinion about this, so I changed it to be either MIT or Apache 2.0. Thanks for the feedback!
How about GPL/AGPL? (for staying free longer-term... :)

I have an existing project which could use this help, but is related more to taking human knowledge management to a new level globally, removing some of our current limitations. Is that close enough to AI to apply? ( http://onemodel.org has details)

Is there any reason you haven't set a date to complete the project (an MVP)? I think it is a good idea to set a date so you reduce the scope of the project to something feasible and applicants don't spend a lot of time pursuing a project that doesn't get enough traction. I think 6-8 months would be ok for an AI project.
Good question! I thought about doing that, and maybe I should, but I'm not sure what the right timeframe is, and I didn't want to pre-filter interesting projects that might not fit my preconceptions.

For example, I could imagine funding a project that a student might do in 8 weeks over the summer working full-time, or a project that someone might pursue part-time over the course of a year. My guess is that the first type will dominate the applications (and therefore the grantees), but I don't know yet.

Very cool initiative! The more support that comes to recognizing the open source projects and research that go into so many useful tools -- the better.
How many winners do you intend to select?
Five. If there are a lot more deserving applications, I'll try to find more funding.
Are you open to outside contributors to your project (uncredited)?
Say an applicant pursues something more along the lines of research that might not immediately lead to an implementation - what do you expect of an 'open source' project of this variety?
I'm open-minded about this. If the project has a believable path to a useful open source contribution in the future, then I think it's fundable. This could even include high variance experiments that might end up totally failing.
Could you put like-minded people in touch who submit similar projects?
This is a great idea. Maybe I'll create a slack for everyone who applies.
What if applicant personally doesn't have any previous experience with AI-related technologies. But has a strong affinity towards learning them. Also If he/she has a really good idea.Will it be okay? or the ones with previous experience will be preferred?
Lack of experience in AI is not necessarily disqualifying. I'd like to see some evidence that the person applying is capable of doing the thing they are setting out to do, though.

That could be experience shipping open source software, a relevant math background, a track record of being great at explaining algorithms, or something else I'm not thinking of right now.

Mind doing the same for hardware projects using an open source OS?

A pal I know could use half of that just to set up the legal structure of his outfit.

What I find surprisingly appealing is, the open source heroes that I grew up with, nat friedman in this case, are pushing towards AI / deep-learning. It feels as if "hackers", who previously flocking towards open-source software are gravitating towards AI, stats, machine learning etc.

I am open to scurry along and move to this new place where the cheese is.

http://course.fast.ai/

Super good stuff.

edit: now, without https.

Is this legit? Throws a security error in Chrome.
Hey you're right. But yes, it's legit, without the https it also works.

I'll drop them a line to warn of this, thank you.

I just submitted my application. Would it be possible to add a notification that the application has been successfully submitted? Perhaps that is not possible with the package you are using.
Good idea, tonight maybe I'll write a script to email people a couple times a day.
Are you looking for more academic/business/cloud cases, or would advancements in game AI be an option as well?
Open to all these things; game AI sounds great.
Great initiative! Always good to see more support for open-source and for AI tooling.
@Nat: Any thoughts on a project using AI to do real-time noise reduction, signal identification, and signal extraction in a controlled environment?

To get an idea of what a more complex version would look like, which is closed source, see:

http://cyphercorp.com

Kind of a loaded question, but what do you mean by 'AI' other than developing an algorithm that makes a computer look 'smart'?
Has anybody looked at ThoughtTreasure? Since Nat is willing to accept updates to Open Source Software, this amazing code base would be a good start of AI related stuff. It really could use a better user interface to its databases, for example. Or a tie to WikiData. There are several good things that could be added to it.

https://en.wikipedia.org/wiki/ThoughtTreasure