@jacyanthis Profile picture

Jacy Reese Anthis

@jacyanthis

Humanity is learning to coexist with a new class of beings. I research the rise of these “digital minds.” HCI/ML/soc/stats @SentienceInst @Stanford @UChicago

Similar User
Mercy For Animals photo

@MercyForAnimals

Barn Sanctuary photo

@BarnSanctuary

FARM Animal Rights Movement photo

@FARMUSA

Farm Sanctuary photo

@FarmSanctuary

We Animals photo

@WeAnimals

Gene Baur photo

@genebaur

Animal Charity Evaluators (ACE) photo

@AnimalCharityEv

Animals Australia photo

@AnimalsAus

John Oberg photo

@JohnOberg

The Humane League photo

@TheHumaneLeague

Rachel Atcheson photo

@r_atcheson

ALDF photo

@ALDF

Shelly Ⓥ photo

@Cshells33Wells

TorontoPigSave photo

@TorontoPigSave

Animal Aid photo

@AnimalAid

Pinned

I discussed digital minds, AI rights, and mesa-optimizers with @AnnieLowrey at @TheAtlantic Humanity's treatment of animals does not bode well for how AIs will treat us or how we will treat sentient AIs. We must move forward with caution and humility. 🧵 theatlantic.com/ideas/archive/…


Jacy Reese Anthis Reposted

Following a historic victory at the polls, what will a second Trump term mean for America's AI future? ti.me/3Cm5cdx


Jacy Reese Anthis Reposted

Can LLMs be induced to deviate from optimal gameplay in a simple game by threats of pain or promises of pleasure? And does the probability of deviating depend on the intensity of the promised pleasure/pain? According to our new paper, released today (arxiv.org/abs/2411.02432), the…

Tweet Image 1

Jacy Reese Anthis Reposted

New paper: The possibility of AI welfare and moral patienthood—that is, of AI systems with their own interests and moral significance—is no longer a sci-fi issue. It's a very real possibility in the near term. And we need to start taking it seriously.

Tweet Image 1

Jacy Reese Anthis Reposted

Is encouraging LLMs to reason through a task always beneficial?🤔 NO🛑- inspired by when verbal thinking makes humans worse at tasks, we predict when CoT impairs LLMs & find 3 types of failure cases. In one OpenAI o1 preview accuracy drops 36.3% compared to GPT-4o zero-shot!😱

Tweet Image 1

Jacy Reese Anthis Reposted

I wrote about a 14 year old boy, Sewell Setzer, who died by suicide earlier this year after becoming emotionally attached to an AI chatbot on CharacterAI. His mother is suing the company, alleging they put young users in danger. nytimes.com/2024/10/23/tec…


Jacy Reese Anthis Reposted

Excited for our proposed ICLR 2025 workshop on human-AI co-evolution! We're looking for diverse voices from academia & industry in fields like robotics, healthcare, education, legal systems, and social media. Interested in presenting or attending? tinyurl.com/3z25jwvc

Tweet Image 1

Jacy Reese Anthis Reposted

If you are in a regulated industry, you are required to install something like Crowdstrike on all your machines. If you use Crowdstrike, your auditor checks a single line and moves on. If you use anything else, your auditor opens up an expensive new chapter of his book.


You've heard of "AI red teaming" frontier LLMs, but what is it? Does it work? Who benefits? Questions for our #CSCW2024 workshop! The team includes RT leads OpenAI (@_lamaahmad) and Microsoft (@ram_ssk). Cite paper: arxiv.org/abs/2407.07786 Apply to join: bit.ly/airedteam

Tweet Image 1

Jacy Reese Anthis Reposted

When talking abt personal data people share w/ @OpenAI & privacy implications, I get the 'come on! people don't share that w/ ChatGPT!🫷' In our @COLM_conf paper, we study disclosures, and find many concerning⚠️ cases of sensitive information sharing: tinyurl.com/ChatGPT-person…

Tweet Image 1

Jacy Reese Anthis Reposted

New paper out! Very excited that we’re able to share STAR: SocioTechnical Approach to Red Teaming Language Models. We've made some methodological advancements focusing on human red teaming for ethical and social harms. 🧵Check out arxiv.org/abs/2406.11757

Tweet Image 1

Jacy Reese Anthis Reposted

ARC is cool, and I look forward to the results, but I expect near-term solutions will be: 1) within the span of existing techniques and/or 2) clearly overfit to ARC, thus showing this to be yet another benchmark that seems more general before it's solved than after

I'm partnering with @mikeknoop to launch ARC Prize: a $1,000,000 competition to create an AI that can adapt to novelty and solve simple reasoning problems. Let's get back on track towards AGI. Website: arcprize.org ARC Prize on @kaggle: kaggle.com/competitions/a…



Jacy Reese Anthis Reposted

Excited to share a unifying formalism for the main problem I’ve tackled since starting my PhD! 🎉 Current AI Alignment techniques ignore the fact that human preferences/values can change. What would it take to account for this? 🤔 A thread 🧵⬇️

Tweet Image 1

Jacy Reese Anthis Reposted

We're sharing progress toward understanding the neural activity of language models. We improved methods for training sparse autoencoders at scale, disentangling GPT-4’s internal representations into 16 million features—which often appear to correspond to understandable concepts.…


Jacy Reese Anthis Reposted

Wrote up some thoughts on a growing problem I see with HCI conference submissions: the influx of what can only be called LLM wrapper papers, and what we might do about it. Here is "LLM Wrapper Papers are Hurting HCI Research": ianarawjo.medium.com/llm-wrapper-pa…


Jacy Reese Anthis Reposted

How to get ⚔️Chatbot Arena⚔️ model rankings with 2000× less time (5 minutes) and 5000× less cost ($0.6)? Maybe simply mix the classic benchmarks. 🚀 Introducing MixEval, a new 🥇gold-standard🥇 LLM evaluation paradigm standing on the shoulder of giants (classic benchmarks).…

Tweet Image 1

Jacy Reese Anthis Reposted

We're making @metaculus open source: Soon you'll be able to audit, critique, and build on our code. We want to live in a better epistemic environment, and we can get there faster working together: metaculus.com/notebooks/2507…


Loading...

Something went wrong.


Something went wrong.