Finally! After all these years dreaming this up, someone put together a social peer review platform that lets you upvote papers and leave comments. This + arxiv ๐. (Ok, the challenge will be to detect fake comments and votes, but at least it's a start.) https://t.co/Kw3kJOMQ04
There is at least one industry research lab where the leadership believes that (super)human-level AI:
- is attainable
- is a scientific research question, not just a question of more compute and more data.
- is not "just around the corner". It will take a while.
- is not an existential risk.
- requires contributions from the entire research community, because no one has a monopoly on good ideas.
- hence requires open source platforms and open research, because no one has a monopoly on good ideas.
- is going to change human condition for the better.
This is very tangential, but even among scientists, debunking doesn't work.
Let's look at some of the empirical work on the topic of replication failures.
First: papers that don't replicate receive more citations than ones that do.
It's a dream to wake up every morning and spend my time working on this problem. I feel very lucky to be a part of @ResearchHub
Big time shout out to @patricklu10 and his team from @quantfive for doing most of the heavy lifting building @ResearchHub's product so far
Go to https://t.co/JpAAl5ufCZ, search your paper and add a project page to it.
Can't wait to see how your readers interact with you after the conference.
Dear #CVPR2023 poster presenters, hope you have a successful experience sharing your papers with all attendees.
Your poster session lasts only a few hours. Isn't it just cool to have a never-ending online poster session?
Has The Machine Learning Review Process Become More Arbitrary As The Field Has Grown? The NeurIPS 2021 Consistency Experiment
If the reviewing were to rerun again, half of the accepted papers would change.
Don't get discouraged at your rejected paper!
https://t.co/BuykV4gKa5
GPT-4 achieves 100% score on MIT problem set.
Some find the result extraordinary, while many others point out the fundamental flaw in the "self-critique" process.
Wondering what the score the paper gets on doublind.
https://t.co/sqzAjbvUwi
https://t.co/8yTYSvSVCN
Introducing #AgentGPT, an attempt at #AutoGPT directly in the browser ๐ค
Give your own AI agent a goal and watch as it thinks, comes up with an execution plan and takes actions. Try for free now at https://t.co/F8Nz4LGC0e
#midjourney text to pic generation is fun. Let's see what logos that MidJourney generated for https://t.co/q6UdPFsGw6. Apparently it cannot generate exact the same word you described but which one do you like?
@MarlosCMachado "Weak Accept: This means that the reviewer
Technically solid: This means that the reviewer found the technical content of the paper to be sound and well-researched." by ChatGPT
@chriswolfvision Based on ChatGPT, One way to get the tweets of AI research folks but not the tweets of AI influencers is to use Twitter's advanced search feature. Here's how you can do it:
1. Go to Twitter's advanced search page: https://t.co/OFE4eELpme
2. In the "Words" field, ...
Iโve seen several tweets criticizing arXiv saying that citing arXiv articles can be considered harmful.
I favor the opposite, pointing to arXiv versions for peer reviewed work as well.
Why? Because authors can amend, fix, and update their work there, which can be important.