Instead: "Please describe how you used LLMs in preparing your responses, with links to transcripts if applicable."
Posts by Jacob Peacock
2. Requesting transcripts. Linking to a ChatGPT or Claude transcript is great—if you used that tool. Your best applicants are using extensions, skills, context and agents that won't meaningful appear in a transcript. And do you really want to read reams of transcript anyway?
Instead try: "The creative content of your responses must reflect your original writing, not that of an LLM. Answers closely resembling LLM outputs may be penalized."
I see two LLM policy mistakes in hiring processes all the time:
1. Total ban on LLMs. Using an LLM for spellchecking, searching your own past writing for reference, or answering factual experience questions should absolutely be permissible. 🧵
Final week to apply to join the Animal Charity Evaluators Board! Deadline March 1st.
animalcharityevaluators.org/about/our-te...
@animalcharityev.bsky.social is looking for new board members! Join me and a talented group of leaders to guide ACE's work to identify highly cost-effective animal charities. Apply by March 1st.
animalcharityevaluators.org/about/our-te...
Made my first podcast appearance on The Vegan Report, hosted by Rayane Laddi. Seth Green and I talk AI, MAHA, nudges, social desirability bias, and more: rss.com/podcasts/veg...
Lastly, here's the new citation: Peacock, J. R. (2026). Price-, taste-, and convenience-competitive plant-based meat analogues would not currently replace the majority of meat consumption: A narrative review. Appetite, 216, 108301. doi.org/10.1016/j.appet.2025.108301
The new section reviews recent literature which "suggested that decreased plant-based meat, butter and milk analogue prices might cause increased consumption of the corresponding animal product" and moving plant-based meat analogues to the meat aisle didn't meaningfully effect meat consumption
So, what's new?
✅ Shorter
✅ More and updated citations
✅ Methods section
✅ Error bars in figures
✅ New section, "Studies of price and convenience manipulations individually"
My paper "Price-, taste-, and convenience-competitive plant-based meat analogues would not currently replace the majority of meat consumption: A narrative review" is now published in the peer-reviewed journal Appetite! Available here: shorturl.at/nMk2R
A bit last minute, but if you're free 4-6p tonight, come watch me judge submissions for Summer of Math Exposition, a sorta-annual competition held by famed math youtuber @3blue1brown.com, at twitch.tv/enstucky More info on the competition here: www.3blue1brown.com/blog/some1
with @jessica-hope.bsky.social, @setgree.bsky.social, and Maya Mathur
Does adding plant-based meat to the @chipotle menu cause less meat purchasing? Our preregistered RCT (N=4,431) using a simulated food order suggests not. Neither adding sofritas PBM nor an additional hypothetical chicken-like PBM meaningfully reduced meat consumption:
Full study: osf.io/z6rn2_v1
I share your frustration. In live action media, I assume humanoid aliens with additive features reflect limitations of practical effects. But animated works especially tend to have nonsense biology and especially biomechanics. And CGI gives us human clone troopers with impossibly narrow helmets...
High-impact opportunity at Greener By Default for someone with great Python and data analysis skills www.linkedin.com/jobs/view/42...
Fresh out of college, I designed binary classifiers and sentiment analysis tools for a marketing start-up. ~10 years later, off-the-shelf LLMs would easily replace and exceed what I did. Wild to read how much LLMs disrupted natural language processing www.quantamagazine.org/when-chatgpt...
Check out these two brand-new animal advocacy Substacks by Rethink Priorities and Seth Green:
rpstrategicanimalinsights.substack.com
regressiontothemeat.substack.com
One PB option brings that down to 84%, two options 78% and a third option just 75%. This definitely surprised me...I thought 3 options would have a larger effect.
We have a similar study pre-registered here osf.io/z9486%E2%80%... update your priors and start forecasting our results!
This metric moderately favors PB meat burger, although note wide CIs.
One last notable result: simulating counterfactuals suggested adding PB options quickly reaches diminishing returns. With no PB options, 92% of participants chose beef (the rest wouldn't buy).
Third col shows the market share of beef in a simulated two-item market with one of the 3 alternatives; the PB meat burger was slightly worse here. But, in col 4, the effect of reducing the alternative price by 1% on market share of beef (cross-price elasticity).
(Ranges are 95% CIs; I used point estimates where CIs weren't reported.) Estimated market shares of the four products competing at equal prices told slightly different stories in Studies 1 & 2, but the takeaway is that the alternatives probably performed about the same.
(and enables a "counterfactual estimation" as if some options weren't available, although I'm somewhat skeptical of this approach).
Results updated me a bit: while I thought the falafel or veggie burger might come out slightly ahead, it wasn't decisive across 4 different measures:
After a choose-one (discrete choice) question, participants were asked a clever follow-up: suppose that option was out of stock, what would you get instead? This seems like a cheap way to get participants to rank options and thus more sensitivity from a single study
These span the gamut from analog to semi-analog to non-analog plant-based alternatives to meat. After a choose-one (discrete choice) question, participants were asked a clever follow-up: suppose that option was out of stock, what would you get instead?
Catching up on some new plant-based meat literature with "Substitution patterns and price response for plant-based meat alternatives." www.pnas.org/doi/10.1073/... In two studies, participants hypothetically chose from 4 burger patties: beef, plant-based meat, veggie or falafel.
Thanks, enjoyed this!