stevenbedrick

1.8K posts

stevenbedrick

stevenbedrick

@stevenbedrick

Katılım Mayıs 2008
311 Takip Edilen233 Takipçiler
stevenbedrick
stevenbedrick@stevenbedrick·
Reading this, I found myself thinking of Joseph Weizenbaum's argument that computers are best understood as being fundamentally conservative forces. He wasn't talking about the same things as @LMSacasas was, but I think the thinking resonates more than a little bit.
Charlie Warzel@cwarzel

Last month @LMSacasas suggested that the internet isn’t a time, but a place. Specifically, the past. For Galaxy Brain I spoke to him/expanded on the idea. I think it explains why we are so exhausted by information & why our politics & culture feel so stuck newsletters.theatlantic.com/galaxy-brain/6…

English
1
0
1
0
stevenbedrick
stevenbedrick@stevenbedrick·
Bergstrom's quote in this article was absolutely priceless, and really brightened my day (not what I was expecting in a monkeypox article). The rise of the "an epidemiologist speaking their truth" genre has been one of the few bright points of the last couple of years.
English
0
0
1
0
stevenbedrick
stevenbedrick@stevenbedrick·
@aclmeeting Fabulous, thank you so much! Is there anything under discussion around encouraging/requiring vaccination? I know that's not a simple question by any means, but it seems worth mentioning...
English
0
0
0
0
ACL 2026
ACL 2026@aclmeeting·
😷ACL COVID Policy😷 1/4 It is the policy of ACL conferences that attendees wear high-quality, well-fitting masks while attending indoor functions at ACL events, removing their masks only the minimal amount necessary to eat and drink.
English
3
22
105
0
stevenbedrick
stevenbedrick@stevenbedrick·
This is fabulous! I especially liked their 4th point ("Attribute agency to the human actors building and using the technology, never to the technology itself"). I've made a point lately of using words like "automation" when talking about applications of AI/ML for the same reason.
Tech Policy Press@techpolicypress

The Center on Privacy & Technology at Georgetown Law is removing “artificial intelligence,” “AI,” and “machine learning” from its institutional vocabulary. Read why in this piece from its Executive Director, Emily Tucker: techpolicy.press/artifice-and-i…

English
0
0
1
0
stevenbedrick
stevenbedrick@stevenbedrick·
And +1 for her prominent shout-out to Weizenbaum; how Johnson's article, which was literally about people's reactions to computer-generated text, could have possibly managed to go for 10k words without even mentioning ELIZA is beyond my ability to comprehend.
English
0
0
1
0
stevenbedrick
stevenbedrick@stevenbedrick·
She goes way beyond simply pointing out specific instances of eye-roll-worthy hype- though of course there's plenty of that, fear not! She shows how the hype is driven first principles that are fundamentally mistaken, and how that leads naturally to a misleading narrative frame.
English
1
0
1
0
stevenbedrick
stevenbedrick@stevenbedrick·
Anybody who read last week's NYT Magazine article on large language models, or really any media coverage on AI ever, or who is interested in questions of "framing" in journalism, should make sure to read @emilymbender's latest: @emilymenonbender/on-nyt-magazine-on-ai-resist-the-urge-to-be-impressed-3d92fd9a0edd" target="_blank" rel="nofollow noopener">medium.com/@emilymenonben
English
1
0
1
0
stevenbedrick
stevenbedrick@stevenbedrick·
@emilymbender I work in this general area (ML, for diagnostic purposes, using voice signals), and I found the article almost painful to read for the exact reasons you highlight in your thread. Thanks for taking the time to pour some water on the dumpster-fire of hype!
English
0
1
3
0
stevenbedrick retweetledi
@emilymbender.bsky.social
@emilymbender.bsky.social@emilymbender·
I find this reporting infuriating, so I'm going to use it to create a mini-lesson in detecting #AIhype. If you're interested in following this lesson, please read the article, making note of what you think sounds exciting and what makes you skeptical. nytimes.com/2022/04/05/tec…
English
11
212
643
0
stevenbedrick retweetledi
The Onion
The Onion@TheOnion·
Scientist Really Thought Job Would Be Less Grant Writing And More Glow-In-The-Dark Lizard Making bit.ly/3tQRIjm
The Onion tweet media
English
9
233
2K
0
stevenbedrick
stevenbedrick@stevenbedrick·
@ZachWeiner Autoregressive language model is a very good choice, but honestly I work in this exact field and we usually just call it a "next-{word,character,symbol,token} prediction" model most of the time. Sometimes we'll group it under the umbrella of "sequence prediction."
English
1
0
4
0
Zach Weinersmith
Zach Weinersmith@ZachWeiner·
For a joke - is there a more technical term for next-word-prediction algorithms?
English
29
1
16
0
stevenbedrick retweetledi
Jessica Schleider, PhD
Jessica Schleider, PhD@JSchleiderPhD·
Here is a 🧵 of terms related to money/funding in academia that no one explained to me until I was applying for faculty jobs (or after I'd started), that I pretended to understand until then, and that would’ve been very helpful to learn earlier than I did. 1/a whole bunch
English
45
1K
3.7K
0