Ted Pavlic (he/him/his)

@tedpavlic.bsky.social

ASU Assoc. Prof in SCAI and SOLS. Decision making and behavior in living and artificial autonomous systems. I ❤️ ants, bees, wasps, robots, and optimal foraging theory. mastodon: @tedpavlic@mas.to Twitter/X: @TedPavlic @TEDx: http://youtu.be/9GWXCRetOjk

The effective federal corporate income tax rate: 1950: 50% 1990: 25% 2020: 13% Stop asking "but how we will pay for it?" Restore the corporate tax rate. That is how we will pay for it.

Annual reminder to academic unit directors: If the agenda you've put together for your yearly "retreat" has no items of any value, you should just cancel the retreat or explicitly announce that retreat attendance is 100% optional. Thank you for your attention to this matter.

Nature did a short piece on our preprint about how LLMs will affect the practice of science; my more detailed thread is below.

Scientists using LLMs will ‘do more, less well’, modelling study predicts

Research suggests that the combination of incentives to publish and the use of large language models will lead to more papers, but they will be less refined.

nature.com

Carl T. Bergstrom@carlbergstrom.com · 2w ago

1. We—@eduede.bsky.social, @mjcrockett.bsky.social, Kevin Gross, and I—have a new preprint on the arXiv today, based on ideas that emerged during an @sfiscience.bsky.social workshop in November 2024: The unintended consequences of large language models as a labor-augmenting technology in science.

Gotta be careful with your AI-generated images, kids. Eventually you'll get a job where there is no one else there to check your work -- and your efficiency-maxxing-self may lose that job! "US government map of Africa mislabels every country at global conference" www.reuters.com/world/africa...

US government map of Africa mislabels every country at global conference

A U.S. government map of Africa mislabeled every country during a State Department presentation at ‌a global conference taking place in Brazil this week, causing a stir among attendees who took screen...

reuters.com

US imports of robot vacuums soon banned. Ban includes "devices weighing more than two kilograms that use sensors, network connectivity and software to achieve autonomous navigation, including the avoidance of obstacles." "US bans Chinese and foreign-made humanoid robots" www.dw.com/en/us-bans-c...

US bans Chinese and foreign-made humanoid robots

The Trump administration has banned the import of humanoid and quadruped robots manufactured outside the United States, citing national security concerns. The ban could strain diplomatic tensions with...

dw.com

An 'illuminating' article! 🧪 "All living things emit a faint glow. Could this light be useful?" -- Ultra-weak ‘biophotons’ might be used to diagnose disease, or could even represent a new signalling mechanism in cells. by Jo Marchant (Nature [News Feature], 2026) www.nature.com/articles/d41...

All living things emit a faint glow. Could this light be useful?

Ultra-weak ‘biophotons’ might be used to diagnose disease, or could even represent a new signalling mechanism in cells.

nature.com

For students who look to LLMs for help with their code but run into usage limits, running an open-weight LLM locally doesn't use up that much storage space and may have similar latency as cloud solutions -- and easy to connect to Positron, VSCode, and others. www.xda-developers.com/self-hosting...

I started self-hosting an LLM, and it became my favorite productivity tool

I installed a local LLM for privacy, but the biggest surprise was how much faster it made my daily workflow.

xda-developers.com

Increased productivity = increased opportunity cost, which means lots of downstream effects on how researchers will choose to spend their (highly valuable) time. A similar effect occurred with the advent of the search engine -- people stopped spending time on the web pages they had searched for.

Carl T. Bergstrom@carlbergstrom.com · 2w ago

1. We—@eduede.bsky.social, @mjcrockett.bsky.social, Kevin Gross, and I—have a new preprint on the arXiv today, based on ideas that emerged during an @sfiscience.bsky.social workshop in November 2024: The unintended consequences of large language models as a labor-augmenting technology in science.

Preprint suggesting hallucinations are often more a function of bad context than bad models. The future of LLM-based AI will be better harnesses and establishment of best practices in context management. "AI Agents Do Not Fail Alone:The Context Fails First" arxiv.org/abs/2607.14275

AI Agents Do Not Fail Alone:The Context Fails First

Context engineering has become central to building reliable AI agents, yet it remains largely unmeasured. Agents do not fail in isolation: their behavior is shaped by the instructions, tools, memory, ...

arxiv.org

In high-stress conditions, species with higher growth rates have an advantage - leading to robustness in overall growth despite community deterioration. 🧪 "Predictable shifts in microbial species composition lead to community-wide robustness to environmental stress" www.nature.com/articles/s41...

Predictable shifts in microbial species composition lead to community-wide robustness to environmental stress - Nature Microbiology

Environmental stress shifts microbial composition towards species with high growth rates, providing community-wide protection.

nature.com