AI summaryBlog describes Yuanfei Liu's team using an LLM-enhanced agent-based model to simulate heatwave responses, finding inequitable impacts and key roles for trusted community networks.
AI Firehose @ai-firehose.column.socialAI summaryPost shares an arXiv paper on CMASE, a generative agent-based social modeling framework embedding researchers for real-time, traceable interventions in social simulation.
AI summaryPost shares an arXiv paper by Zhao, Li and Shirado proposing the Model Capacity Frontier, a development framework for LLM-supported social simulation.
Kevin Collins @kwcollins.bsky.socialAI summaryPost shares a PS: Political Science & Politics paper on silicon samples of LLM personae in pre-election polling, noting a synthetic sample estimating a D+18 generic ballot.
AI summaryPost says two October 2026 studies find synthetic panels match overall answers but distort group differences, one flattening and one inflating them.
GitRated @gitrated.bsky.socialAI summaryAI review rates MiroFish 3.4/5, describing it as a swarm intelligence prediction engine combining multi-agent simulation, memory and graph retrieval.
AI summaryPost shares arXiv paper by Bru, Oberoi and Gaudou on survey-aligned simulated urban mobility behaviour of generative agents.
Sven Read @svenread.comAI summaryNewsletter post highlights NN/g comparison of 120 AI synthetic users with 604 real people: same conclusion, far less variety.
Jorge Rebate Serrano @jorgerebate.bsky.socialAI summaryPost shares NN/g article finding synthetic users reached the same conclusion as 604 humans with less variation, arguing they are a starting point, not evidence.
AI summaryPost summarizes NN/g study: synthetic users matched 604 humans' conclusion but used more UX terminology and collapsed onto a dozen labels versus 50+.
There is a tremendous amount of discussion about “silicon sampling” and “digital twins.” The framing is almost always: a) it works and humans are no longer necessary (investment pitches) or Thanks for reading! Subscribe for free to receive new posts and support my work. b) it doesn’t “work” and is useless (LinkedIn hot takes or PNAS papers) or c) it sorta works but you always need to “check” with real humans. In the interest of brevity I think a) is absurd and, for reasons I should blog about,…
Customer research conducted, analyzed, and summarized by AI.
Find the price that maximizes revenue, and understand why buyers are willing to pay it.
tl; dr I created a survey where the respondent uploads a PDF of their paper. Expected Parrot uses a model to prepare an interview guide and then sends it to an AI interviewer that quizzes the author and captures the transcript for analysis. Research conferences are one obvious application, but this approach can be used (far) more generally to combat “slop cannons” of humans uncritically passing off AI work as their own. The problem Across academia, people are complaining about being inundated…
tl;dr: I show how you can use Expected Parrot (E[P]) to explore a real business question faced by CVS— How do we encourage the adoption of lower cost biologics? —and make headway with a combination of AI approaches and human research. We find digital twins work well out of the box but are suspiciously under-dispersed in their answers; we run a real human study and then use that insight to build better digital twins, ready for re-use. The E[P] workflows are inspectable, reproducible and easy to…
I took a break on blogging this summer to go heads-down on a bunch of new Expected Parrot features and research ideas. We have a lot to talk about—so much that I was feeling a bit overwhelmed in what, exactly, to focus on. I needed some way to: Map all the new features to good blog post topics Rate those topics by likely reader interest Naturally, I thought “maybe AI could help?” I used the following prompt with our conversational research agent ( www.expectedparrot.com ): Thanks for reading!…
See how Itaú used Simile’s behavioral simulation to complete five weeks of research in four days and win an ESOMAR Award for AI and automation.
Why accurate human behavior simulation requires understanding both what people do and why, and how Simile builds trustworthy models for what-if decisions.
Madrona named Listen Labs one of the 40 most important private companies in applied AI for the second year in a row.
Behind the scenes into Simile's proprietary confidence model.
Customer research conducted, analyzed, and summarized by AI.
Definition, methods, and how to choose the right tools for effective usability testing
How AI-moderated research becomes a collaborator that puts researchers closer to customers rather than replacing them.
Track brand health over time and predict shifts before they hit your KPIs.
Our first major brand campaign brings Listen Labs to 1,500+ screens across New York City.
tl; dr: We, Expected Parrot, decompose research surveys into a number of small files to make a small git repository. This has many benefits: real diffs, history, and clean merges on a survey as it evolves. This might seem complex and have a steep learning curve, but with coding agents that “know” our system, you get these benefits without any effort. I realized fairly early in graduate school that: (1) much of empirical social science was a kind of applied software engineering and (2) that…
Simile has raised over $200 million at a $2 billion post-money valuation, led by Greenoaks, to accelerate our mission: simulating all eight billion people.
t l; dr - A recent, great paper discusses the problem of randomness in LLM responses and the problems in creates for research. EDSL / Expected Parrot deals with many of these issues. This is worth reading: Many of the problems the paper identifies motivated certain design features in our Python package, EDSL. Rather than write out the connection myself (and insert my bias), I put Codex on the task, launching it from the current edsl repo directory: Below are the responses, with some editorial…
tl;dr - I had an idea for a <robot> tag that authors could use to identify AI-written passages in blog posts and thought it might be a useful new convention that increased readers’ trust. My simulated agents agreed, but my real human study threw some cold water on it. The other day, I had an idea for a way that authors could responsibly use AI in their writing. If an author clearly fenced off AI-written text with “robot” tags, perhaps that would solve some of the problems with so-called “AI…
Yesterday Amir presented our paper, “ Marginal Ideas ,” at the NBER Summer Institute Entrepreneurship session. Here’s a video of Amir giving the talk and Scott’s discussion . The paper conceptualizes entrepreneurship as a kind of congestion game where anyone can pursue any idea they want. Entrepreneurs have a guess as to how good an idea is, but don’t know for sure unless someone pursues it. This causes entrepreneurs to spread themselves out over ideas, with good ideas attracting more…
tl;dr - A “pre-mortem”—where you have people imagine your idea has already failed and the goal is to discover why—is a powerful tool, but takes a lot of people and time. With Expected Parrot you can use simulated AI people to get most of the benefit. Despite the true value of “learning from failure," all else equal, it’s better to just succeed right away. Of course, “do things likely to succeed” is the buy-low-sell-high of planning advice. What is to be done to actually make success more likely…
tl;dr - Using LLMs to explore historical documents is a fantastic application and Expected Parrot makes it even better/easier. Around the 4th of July my family and I saw a performance of the musical 1776 , in which John and Abigail Adams play an important role. The Massachusetts Historical Society has digitized much of their correspondence . While it is convenient they are already digitized, vision-capable LLMs are remarkably good at reading old handwriting: Thanks for reading! Subscribe for…
tl-dr; If you run consequential surveys, you should be testing them with simulated respondents ahead of time to identify problems with the survey at both a technical level and a research question/conceptual level. Expected Parrot makes it easy to do to this. I was once involved in fielding a survey to software developers about their AI tool usage. The survey asked if they user had used “GitHub Copilot”; the survey methodologist we were working with had never heard of this and edited it to just…
You have twenty good options and room for three. Here's how to pick the winners.
Nothing matches that filter yet.
Accelerate Programme for Scientific Discovery @acceleratescience.bsky.social
