Show newer

I've got a couple things I've like to cover for my Tuesday Deep Dive next week... which of these would you like to see?🤔

(For 1 & 3, I'll be covering potential harms & mitigations, not doing a "how to", just in case that wasn't obvious. 😅)

Hello friends! Here's this week's coffee chat. Today I ended up with more ethics things to discuss than usual (it's not *all* bad though, don't worry).

Hope you can join us! ☕️🤓

("Links please" members on my ko-fi: the links are already up! 💪)

youtube.com/watch?v=VxzLIBBXlv

Hello friends! Here's the link for tomorrow's Coffee Chat. I hope you'll have time to drop by as we talk about what's been happening in ML & NLP this week.

youtube.com/watch?v=VxzLIBBXlv

Had a great stream and a lot of fun! My biggest piece of feedback for @adaobiadibe_@twitter.com would be to add a little disclaimer under the translations that says something like:

"Accuracy not guaranteed. Information might be removed, incorrectly added or paraphrased in the translation."

Show thread

Stream starts in just a couple minutes! Today we're going to be looking at a GPT-3 based "Jargon Translator":

- How easy is it to use?
- How well does it work across different technical domains?
- What works well & what can be improved?

youtube.com/watch?v=M_GS1j0Vvr

For my stream tomorrow, @adaobiadibe_@twitter.com has very generously agreed to let me do a system evaluation of her @explainjargon@twitter.com project. I hope you can join me!

youtube.com/watch?v=M_GS1j0Vvr

Friends, I have a little surprise Friday treat for you! 😁

I talked with @YJernite@twitter.com from @huggingface@twitter.com about responsibly releasing the BLOOM model and the strategies the team considered and ended up using to mitigate potential harms. 👇👇

youtube.com/watch?v=YWWY_RSz0N

RT @libshipwreck@twitter.com

If you think science & technology exist outside of a society’s political culture, I encourage you to read:
- Race After Technlogy by @ruha9@twitter.com
- Programmed Inequality by @histoftech@twitter.com
- Freedom’s Laboratory by @ColdWarScience@twitter.com
- Algorithms of Oppression by @safiyanoble@twitter.com

🐦🔗: twitter.com/libshipwreck/statu

This is a great example of what we briefly chatted about on stream today: the biology of sexes is way, *way* more messy and complicated than the children's book version a lot of us learned in school. 🚻🐁(I know it's a mouse emoji but let's pretend it's a mole.)

Show thread

RT @Naturalcalendar@twitter.com

Hello lovely Nature Noticers.

It is time for today's thread! 🤸‍♀️🤸‍♂️🤸‍♀️

Follow me, and we'll delve deep into the subterranean and subversive world of Talpa europaea, the common European mole.

We'll begin with a bombshell - there are no female moles.

🐦🔗: twitter.com/Naturalcalendar/st

RT @ml_collective@twitter.com

8 researchers from Nigeria got accepted into Deep Learning Indaba @DeepIndaba@twitter.com but couldn't afford the trip. Access to research, mentorship, and networking opportunities like this is vital for early-stage researchers. You can help make their trips possible!
donorbox.org/mlc-support-deep-

🐦🔗: twitter.com/ml_collective/stat

RT @sabelonow@twitter.com

Another reason why African Philosophy is timely for "Decolonial AI" because it is largely a response to Colonization. It frames well the issues that are worsened by AI and provides a modern framing to critique AI's colonial link and effects

🐦🔗: twitter.com/sabelonow/status/1

Stream is starting in just a couple minutes! (And will probably keep going for a couple hours: I swear no matter how much I cut stuff to cover it keeps getting longer.)

Just going to make my coffee now. Hope to see you there! ☕️😁

youtube.com/watch?v=JSZWETbpfp

uhhhhhhh FYI linguists

(for non-linguists: we often use this term to mean made-up but plausible nonwords used in experiments, things like "bloar" or "sreed")

RT @emilyst@twitter.com

good lord, "nonce" means child molester in the UK?

🐦🔗: twitter.com/emilyst/status/154

RT @jesslynnrose@twitter.com

With the current job market, some really great folks are looking for work. Let's help them find good work?

If you work for a non-evil (or just less evil 👀) tech company that's hiring right now, reply with a link and details?

🐦🔗: twitter.com/jesslynnrose/statu

A friendly reminder that "words" are convenient (especially for NLP) but probably not real. 🤷‍♀️

RT @pictureofitself@twitter.com

The Lexicalist Hypothesis, however, has increasingly fallen into disrepute. Quite plainly, it seems that the folk-scientific notion of "word" doesn't really map cleanly onto any meaningful linguistic distinction other than "where the spaces are" 14/

🐦🔗: twitter.com/pictureofitself/st

RT @AIR_lab_MUK@twitter.com

We are excited to be hosting a team from @DalbergTweet@twitter.com on 28th July 2022 who will be presenting about "Building Inclusive Data Ecosystems".
We invite everyone to join us, especially university students. @Makerere@twitter.com @j_nabende@twitter.com @iBaino@twitter.com @MakCoCIS@twitter.com @marconi_lab_mak@twitter.com
See you then!

🐦🔗: twitter.com/AIR_lab_MUK/status

Hello friends, here's the link to stream tomorrow!

We had a very research-heavy one last time but this week we'll focus more on practical/professional information. Hope to see you there! ☕️🤓

youtube.com/watch?v=JSZWETbpfp

RT @AttorneyNora@twitter.com

The gov’t is buying our data from private brokers in violation of our basic rights. This can be our location at a protest or visit to a house of worship, our app usage, & more. The Disinfo Defense League is urging Congress to protect our privacy rights: freepress.net/news/press-relea

🐦🔗: twitter.com/AttorneyNora/statu

This is a such a perfect metaphor

RT @JanelleCShane@twitter.com

AI-generated stuff often has the surface appearance of being something without actually being that thing.

For example: "Rainbow Magiceye stereogram of a unicorn" generated by

🐦🔗: twitter.com/JanelleCShane/stat

Show older
Mastodon

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!