@odedrechavi
Epigenetic Inheritance, Neuroscience & anything biology-related https://www.odedrechavilab.com/ q.e.d: https://www.qedscience.com Organizer of “The Woodstock of Biology” TED: https://shorturl.at/myFTY Huberman Lab Podcast: https://youtu.be/CDUetQMKM6g
"AI is here to stay...From writing to reviewing...It is essential to engage with it [&] develop use cases that benefit the community...The clear winning option is use of AI review by authors prior to submission.” link.springer.com/article/10.1... via @tlemberger.bsky.social @odedrechavi.bsky.social
The Russian lit version:
Chekhov wrote: "I am afraid of Tolstoy's death. If he were to die, there would be a big, empty place in my life. To begin with, because I have never loved any man as much as him."
Chekhov died first and after his death Tolstoy wrote: "I never knew he loved me so much."
Thank you!!
Hard pick! Lots of things interest me as well, maybe @leslievosshall.bsky.social :)
Alejandro you can start tomorrow
If you could do a sabbatical in any lab to recharge your energy and get excited again about a totally different field in biology (different from what you are studying), where would you go?
“I made just a few small changes to the text”
does not mean what you think it means.
when the editors do the job and the reviewers do their job it can function, but that's not always the case... I don't think we should necessarily compare AI to the ideal reviewers and editors, we should start by comparing it to the average reviewer...
There's no doubt that AI slop exist (plenty of it), but that doesn't mean AI can't be used to get good feedback.
It also warns of some dangers of AI review, which I think can be mitigated if AI review is used not as a final judge, but as a way to flag problems and highlight strengths to help humans make better decisions (with more information to base those decisions on).
An excellent piece by @jvoigts.bsky.social, which also includes a cool simulator that takes into account “Baseline Review Probability” and “Proportion of Bad Actors” (unfair reviewers), demonstrating how a small number of bad reviewers can screw up peer review.
We're measuing everything, not only comparisions Vs. human and the authors' feelings. More papers on such benchmarking (quantitative) soon!
when it's your turn to give lab meeting and you don't have new results to show
Hi! It certainly handles references, and will in the future also factor in their relative vaildiy (a big challenge that we're working on).
👇
Read this @emboreports.org paper here=> link.springer.com/article/10.1...
Yes! An important first step, and there's plenty more to do (and to measure!)
It was a very cool experiment! The author-centric perspective makes a lot of sense.
An important next step is to benchmark multiple AI review platforms. Maybe this can also pave the way for benchmarking human peer review across journals and defining "quality features" of the peer review process.
It looks at the figures too. That's harder, but we are constantly improving at it (and it's one of our main efforts going further).
Read the paper here: link.springer.com/article/10.1...
A pleasure doing this with the great Thomas Lemberger
@tlemberger.bsky.social and Niv Samuel Mastboim
@nivsm.bsky.social
At the same time, we are building an alternative model that puts agency directly in scientists’ hands.
Researchers should be the ones deciding when their work is ready to be shared. We are building the infrastructure to support that.
q.e.d. is not working in isolation; we are collaborating with leading pro-scientists organizations, including EMBO (and other journals), Review Commons, and OpenRxiv (
@openrxiv.bsky.social @biorxivpreprint.bsky.social), and are working closely with researchers across fields.
The conclusion was clear: Scientists want AI feedback to strengthen their work IN ADVANCE, under their control. Not as a gatekeeper, but as a tool for constructive input.
That is exactly what we are building at q.e.d! We are on the authors' side.
The AI analysis was compared to the combined feedback of multiple human reviewers, not to a single report, and had no access to those reviews.
We then asked authors how they evaluate the strengths and limitations of both approaches, and how they would actually want to use AI.
Here’s what we did: Authors whose manuscripts had already received journal-agnostic review at Review Commons were provided with an independent AI review generated by @qedscience.bsky.social
It’s finally out! Together with @embopress.org and
@reviewcommons.org, we conducted a structured side-by-side comparison of human peer review and our AI scientific review (see thread 👇👇👇🔥).
It’s published. That’s all that matters
CDG is in the top 3 worst airports
PI before approving a draft they didn’t read
"It's more of a comment than a question"
