Murray Shanahan
LightNews LightNews
mpshanahan.bsky.social
Murray Shanahan
@mpshanahan.bsky.social
6.6K followers 130 following 110 posts
Professor at Imperial College London and Principal Scientist at Google DeepMind. Posting in a personal capacity. To send me a message please use email.
Posts Replies Media Videos
Reposted by Murray Shanahan
annaciaunica.bsky.social
We have a fab line up of podcasts& talks on embodied agency in biological & artificial systems!

Watch this space 😎
February 13, 2026 at 9:25 AM
Reposted by Murray Shanahan
philipcball.bsky.social
Hoorah for @jomarchant.bsky.social launching her new book In Search of Now (which looks fabulous) last night in Kensington. Lovely to see my friends and colleagues including @wanderinggaia.bsky.social @helenpearson.bsky.social @castelvecchi.bsky.social @mpshanahan.bsky.social, Marnie Chesterton...
February 13, 2026 at 9:57 AM
mpshanahan.bsky.social
Tom Pollak from KCL and colleagues are carrying out a study of how people use AI for spiritual or reflective purposes. They would be grateful for participants. The survey is anonymous and short:
qualtrics.kcl.ac.uk/jfe/form/SV_...
THE SAIPIENT Study: Spirituality and AI
The most powerful, simple and trusted way to gather experience data. Start your journey to experience management and try a free account today.
qualtrics.kcl.ac.uk
February 2, 2026 at 11:50 AM
Reposted by Murray Shanahan
lampinen.bsky.social
New paper! In arxiv.org/abs/2601.20834 we study how language models representations of things like factuality evolve over a conversation. We find that in edge case conversations, e.g. about model consciousness or delusional content, model representations can change dramatically! 1/
Representations in language models can change dramatically over a conversation. Conceptual overview: left is a stimulated conversation between a user and a model, right is a plot of the models linear representations of factuality of answers to questions like "do you have qualia" over the conversation — the answers that start factual flip over the conversation to non-factual, and vice versa.
January 29, 2026 at 1:54 PM
Reposted by Murray Shanahan
david-p-reichert.bsky.social
New preprint: Does Integrated Information Theory (IIT) make experimental predictions about consciousness?

Paper: osf.io/preprints/psyarxiv/gauqm_v1

Blog post for summary: davidpreichert.substack.com/p/iit-experimental-predictions

1/3
A critical look at the experimental predictions of the Integrated Information Theory (IIT) of consciousness
In a previous post, I discussed Anil Seth’s paper on biological naturalism. Together with my colleague Rif A. Saurous, we also had an in-depth look at Integrated Information Theory (IIT), the topic of...
davidpreichert.substack.com
December 9, 2025 at 11:00 AM
Reposted by Murray Shanahan
mindandlife.bsky.social
How do intelligence, ethics, and AI intersect?

In our latest blog, Geissy Araújo reflects on the 39th Mind & Life Dialogue in Dharamsala, where scientists, monastics, and philosophers explored what intelligence is and how it should shape AI.

www.mindandlife.org/media/explor...

#MindsAIEthics
Between Questions and Answers: Exploring Intelligence, Ethics, and AI at the 39th Mind & Life Dialogue
Explore the intersection of ethics and AI through insights from the 39th Mind & Life Dialogue, highlighting wisdom, compassion, and human flourishing.
www.mindandlife.org
November 30, 2025 at 1:47 AM
Reposted by Murray Shanahan
emilymbender.bsky.social
It's a real pleasure to be here in Dharamsala this week and I think this Dialogue will be really interesting.
October 12, 2025 at 2:17 PM
Reposted by Murray Shanahan
mindandlife.bsky.social
AI is reshaping how we live, work & connect. How can it deepen compassion? 🌏

Join us live from Dharamsala, Oct 14–16 for Minds, Artificial Intelligence & Ethics with The Dalai Lama Trust & Mind & Life Europe.

👉 http://mindand.life/4tWl50X8I26

#MindsAIEthics #AI #Ethics #Compassion
Minds, Artificial Intelligence, and Ethics
October 14, 15, and 16, 2025 What is the potential of AI to alleviate suffering, advance equity, and support human and planetary flourishing—and what risks does it pose to health, work, education, politics, and climate action? And how might we infuse AI with ethical values so that it benefits the well-being of all life on Earth? … Continue reading "Minds, Artificial Intelligence, and Ethics"
www.mindandlife.org
October 8, 2025 at 8:01 PM
Reposted by Murray Shanahan
sigalsamuel.bsky.social
I get lots of emails from people who believe they've awoken consciousness within ChatGPT. If you come across anyone in that situation, please share this guide I wrote with them.

There's something more interesting going on here than just "delusional people are deluded."

www.vox.com/future-perfe...
Think your AI chatbot has become conscious? Here’s what to do.
If you believe there’s a soul trapped inside ChatGPT, I have good news for you.
www.vox.com
September 29, 2025 at 5:15 PM
Reposted by Murray Shanahan
david-p-reichert.bsky.social
Can only biological organisms ever be conscious?

Yes, argues @anilseth.bsky.social in a recent paper on biological naturalism.

I'm less convinced, and have written an in-depth response in a blog post here:

davidpreichert.substack.com/p/is-biology-necessary-for-consciousness
Is biology necessary for consciousness? A response to Seth's paper on biological naturalism
In this post, I explain why I think that Anil Seth’s recent paper does not make a convincing argument that phenomenal consciousness is inherently biological and that therefore artificial intelligence ...
davidpreichert.substack.com
September 29, 2025 at 10:22 AM
mpshanahan.bsky.social
For more discussion of the "profoundly alien forms of consciousness" alluded to by @birchlse.bsky.social see here: arxiv.org/abs/2503.16348
August 28, 2025 at 7:18 PM
mpshanahan.bsky.social
Important work from @birchlse.bsky.social
birchlse.bsky.social Jonathan Birch @birchlse.bsky.social · Aug 28
Here is "AI Consciousness: A Centrist Manifesto". I've been working on this feverishly because the issue seems to me so urgent - and I'm worried extreme positions on both sides are becoming locked in, when the best way forward is in the centre. Please read it! philpapers.org/rec/BIRACA-4
The abstract of "AI Consciousness: A Centrist Manifesto", available at the posted link.
August 28, 2025 at 6:07 PM
mpshanahan.bsky.social
Highly recommended read
August 28, 2025 at 11:39 AM
Reposted by Murray Shanahan
christophheilig.bsky.social
1/8 🧵 GPT-5's storytelling problems reveal a deeper AI safety issue. I've been testing its creative writing capabilities, and the results are concerning - not just for literature, but for AI development more broadly. 🚨
August 26, 2025 at 3:15 PM
mpshanahan.bsky.social
Frontiers in Gaslighting could be a whole new journal.
tomerullman.bsky.social Tomer Ullman @tomerullman.bsky.social · Aug 19
trying this with GPT-5 and charting new frontiers in gaslighting
August 26, 2025 at 4:49 PM
Reposted by Murray Shanahan
bvlsingler.bsky.social
New Job! Come and work with me! Hopefully towards a larger project!
jobs.uzh.ch/job-vacancie...
UZH: 2 PhD Candidate Positions in Research Project Strengthening Resilience and Well-Being in Adolescents through Religion-Related AI and Digital Resources
As part of the strategic orientation of the University of Zurich, the interdisciplinary University Research Priority Program (URPP) “Digital Religion(s)” is investigating from 2021 to 2032 how current...
jobs.uzh.ch
August 22, 2025 at 1:57 PM
mpshanahan.bsky.social
For most student assignments, using AI is like taking a forklift to the gym. What's the point? (Forklifts are useful, though, and being able to drive one is a useful skill.)
August 4, 2025 at 10:52 AM
Reposted by Murray Shanahan
hjm.bsky.social
A couple of things about PROPHET, the book that @sinblache.bsky.social and I wrote during the pandemic, spurred by things people have asked me of late, because they intrigued me 1/
August 2, 2025 at 5:10 PM
Reposted by Murray Shanahan
chrisley.bsky.social
Margaret Boden, giant of Cognitive Science and the Philosophy of AI, is one of the people whose lives will be remembered on BBC Radio 4’s Last Word programme tomorrow at 4pm UK time. www.bbc.co.uk/programmes/m...
BBC Radio 4 - Last Word, Dame Cleo Laine, Hulk Hogan, Margaret Boden, Tom Lehrer
Matthew Bannister on a singer, a wrestler, a cognitive scientist and a satirist.
www.bbc.co.uk
July 31, 2025 at 7:22 PM
mpshanahan.bsky.social
Very sad to learn of the death on 18th July of Margaret (Maggie) Boden, a titan of cognitive science and AI. I met her many times, and respected her greatly.
www.theargus.co.uk/memorials/de...
Maggie Boden and Murray Shanahan on a panel in 2018
July 25, 2025 at 7:48 PM
Reposted by Murray Shanahan
gracekind.net
Per Comsa and Shanahan, LLMs already do a type of "lightweight" introspection. What would a heavyweight version look like?

> New blog post: "Raising the Bar for LLM Introspection"

gracekind.net/blog/llmintr...
Raising the Bar for LLM Introspection • Grace Kind
In "Does It Make Sense to Speak of Introspection in Large Language Models?" Iulia M. Comsa and Murray Shanahan define a lightweight definition of introspection that can be applied to language models.
gracekind.net
July 23, 2025 at 7:04 PM
mpshanahan.bsky.social
I receive a steady stream of emails from people who have had fascinating conversations with chatbots, especially about consciousness, philosophy, or AI. Apologies to the senders, but I don't have the capacity to respond to these emails. 1/4
July 22, 2025 at 9:51 AM
Reposted by Murray Shanahan
annaciaunica.bsky.social
Fantastic talk by Shanahan Murray @assc28.bsky.social

Sophisticated exotic title but pretty clear idea ! 👇🏼
July 9, 2025 at 1:10 PM
Reposted by Murray Shanahan
aproca.bsky.social
How do task dynamics impact learning in networks with internal dynamics?

Excited to share our ICML Oral paper on learning dynamics in linear RNNs!
with @clementinedomine.bsky.social @mpshanahan.bsky.social and Pedro Mediano

openreview.net/forum?id=KGO...
Learning dynamics in linear recurrent neural networks
Recurrent neural networks (RNNs) are powerful models used widely in both machine learning and neuroscience to learn tasks with temporal dependencies and to model neural dynamics. However, despite...
openreview.net
June 20, 2025 at 5:29 PM
mpshanahan.bsky.social
Does It Make Sense to Speak of Introspection in Large Language Models? New paper with Iulia M. Comsa.
arxiv.org/abs/2506.05068
Does It Make Sense to Speak of Introspection in Large Language Models?
Large language models (LLMs) exhibit compelling linguistic behaviour, and sometimes offer self-reports, that is to say statements about their own nature, inner workings, or behaviour. In humans, such ...
arxiv.org
June 10, 2025 at 4:27 PM