David Holtz
@daveholtz
Followers
4K
Following
15K
Media
623
Statuses
6K
professor at @columbia
Brooklyn, NY
Joined December 2010
🚨 A new working paper that tackles the big question: "Will prompt engineering still matter in the future?" 🤔 https://t.co/dAFdo0jxm5 This is joint work w/ @eamanjahani, @BenSManning, @joe_z_zhang, @ocolluphid, @malsobay, @CNicolaides and @ssuri. More details below 👇 1/
4
32
111
Very excited to be co-organizing the 2026 edition of BIG.AI@MIT with @sinanaral, @harangju, and the fantastic folks at the @mit_ide. If you've attended the IDE's @CODEConference (which I co-organize), the high-level idea is for https://t.co/WfMCFFiqNI to be a similar conference
0
3
10
Know any strong early-career researchers working on AI + organizations? @TU_Muenchen is hiring for a 2-year Global Postdoc Fellowship with Isabell Welpe's group. I'm on the advisory board for a related research cluster they're running. https://t.co/w0yfzkMF3i
tumcso.com
Join our team - we are hiring!
0
1
1
I had the chance to work with Ed on a project a couple of years ago and can attest that he is great! Any department would be lucky to have him.
Do poverty traps exist? @ed_jee's study of 27 RCTs finds they're real but rare. Though fixed costs make traps common, forward-looking behavior & productivity differences mean they only affect 25% of households. Read more: https://t.co/dIx7XdnJoO
#UChicago #EconJobMarket
0
0
0
It’s not hard to draw the line from “in this field, whether or not plots should have error bars is subjective” to “in this field, it’s okay to write and/or review papers entirely with LLMs.”
@sanxing_chen @pangramlabs My impression of AI reviews is that they tend to ask for extra work: 1. You should have error bars on your plots. 2. You should run with multiple random seeds. 3. You should run on more datasets. These are subjective, maybe true, maybe not, but cause authors lots of extra work.
1
1
2
This is my favorite conference and I'm excited to be co-organizing this year. Please submit your papers.
The NBER Digital Economics and AI conference submission deadline is coming up on December 1. Submit here:
1
10
35
doing this on all my papers now
This paper is really cool, but perhaps my favorite part is that they determine author order via Mario Kart
1
0
3
Launching AI for Public Goods Fast Grants! We'll distribute $150k to advance critical work connecting AI and public goods. đź’° $10k per project đź’° $800 reviewer compensation PUBLIC GOODS := open source, ecosystem services, climate, urban infra, comms, education, science, & more
Announcing AI for Public Goods Fast Grants (AI4PG) - Up to $10K for AI research improving public goods funding. Fast review (2-3 weeks), simple applications (4 pages + 1 budget page), open to any researchers worldwide. Call for reviewers now open! https://t.co/zUTezH1Afc
5
35
156
We're hiring an assistant professor at @Wharton OID in information systems and operations management! Apply here by 11/1 to ensure full consideration: https://t.co/eqPKC5H73P spread the word!
2
13
27
These results suggest that getting a PhD causally worsens mental health, or at least receiving psychiatric medicines. The reversal post PhD degree is particularly convincing. But the up trend among the control group is intriguing. The highly educated are in distress.
72
378
2K
I am once again on the program committee for the Workshop on Platform Analytics (WoPA), which is on April 24-25 at UCSD. This is a great event for folks doing research on digital platforms: relatively small, high-quality presentations, awesome discussants/feedback, and it takes
1
0
3
The DRO division at CBS is hiring this year! We welcome applicants from a wide range of fields, including OM, econ, stats, CS, engineering, and IS. If you're on the market, apply! If you have any questions about the position, please reach out. https://t.co/dAV6J4BIcD
0
3
9
The DRO division at CBS is hiring this year! We welcome applicants from a wide range of fields, including OM, econ, stats, CS, engineering, and IS. If you're on the market, apply! If you have any questions about the position, please reach out. https://t.co/dAV6J4BIcD
0
3
9
“The best prompters weren’t software engineers. They were people who knew how to express ideas clearly in everyday language, not necessarily in code.” — @daveholtz, @mit_ide
mitsloan.mit.edu
Only half of performance gains seen after using a more advanced AI model come from the model itself. The other half come from how users adapted their prompts.
0
7
16
📢 Call for abstracts for the Conference on Digital Experimentation (CODE@MIT) 📨Submit your best work by *Sept 12th*! https://t.co/RX2ySBwKEf Here are three topics we're particularly interested in getting more submissions on this year:
1
7
21
If you switch from spotify to Apple Music you will be stunned by how much your music no longer sounds like it comes from the bottom of an aluminum can
As more artists pull their music from Spotify, Apple Music has launched a new feature that allows listeners to import their playlists over → https://t.co/MHZh3rxbJd
2K
18K
300K
Fake Academia Problem: All professors are doing woke indoctrination! Real Academia Problem: Universities are overly risk averse leading to massive inefficiencies like intense IRB scrutiny for low-risk observational retrospective research and crazy data security rules
2
3
70
why does it feel like every australian survivor player is constantly cutting a wrestling promo
0
0
1
Only half of performance gains seen after using a more advanced AI model come from the model itself. The other half come from how users adapted their prompts. @daveholtz @eamanjahani @mit_ide
mitsloan.mit.edu
Only half of performance gains seen after using a more advanced AI model come from the model itself. The other half come from how users adapted their prompts.
1
2
3