Launching cf-debate updates: Call for input on 'indicators of consciousness'
This post launches the updates for Co-Sentience Initiative (CoSI) projects, including cf-debate.com.
CoSI exists to help prepare for a future shared with artificial minds. We know very little today about the nature of AI, how society should handle increasingly agentic AI, how we might recognise artificial consciousnesses, and what to do during this critical period of uncertainty and transition.
CoSI believes that our collective intuitions about consciousness will play a major role in the next few years, with the topic moving from academia into mainstream debate. To foster better outcomes, we must hold academic theories to higher standards, raise awareness about key arguments, and improve the quality and diversity of research.
CF-debate.com is the largest repository of formal arguments for and against the possibility of computer consciousness in principle. The debate is far from resolved and far from static, so new arguments (and counter-arguments) are being added continuously, along with other related information. Future posts will share more cf-debate sister projects and the broader CoSI vision.
Now that we’ve got these updates moving, we’ll aim to do them every 1-3 months. We’ve decided to manage the updates via Substack, so it’s easy for you to see old updates, share comments, and unsubscribe at any time. You can also comment on the post or reply to the Substack email alerts to get directly in touch with us.
Three items in this update:
1) New arguments added to the website
2) Let us know what type of content / engagement would be most valuable
3) Call for new content submissions: indicators of consciousness (rewards available)
1) New arguments added
There’s been some great engagement from the research community since launching the site towards the end of 2025, including several contributions from recipients of these updates (thank you!). Alongside over 10k page views and over 500 quiz completions, feedback to improve the arguments on the site has been a real highlight.
Three example additions from community input:
A) Simulation hypothesis
Existing theories of physics cannot rule out that the universe is a simulation. Given this, we cannot rule out that consciousness is simulated. Such simulations are highly likely to support computational functionalism (CF). Any non-computational theory that rules out CF needs to appeal to new physics or place consciousness outside of physics. More details and possible counter-arguments here.
Thanks to Demian Till for crafting this argument.
B) Proper influences
Computational functionalism typically relies on algorithms that behave reliably under both actual and counterfactual circumstances. However, reliability is context sensitive and depends on a threshold that is hard to motivate (how reliable is reliable enough?) — introducing dangerous amounts of subjectivity into the assessment. More details and possible counter-arguments here.
Thanks to Derek Shiller for crafting this argument, which is based on his superb paper about probabilistic bombs in Erkenntnis last year.
C) Physics violations
Fundamental physics teaches us that underlying causal dynamics are purely state-based. If so, then process-based theories of consciousness either aren’t fundamental (they are just proxies for a deeper state-based theory) or consciousness isn’t causally relevant. More details and possible counter-arguments here.
Thanks to Aneil Mallavarapu for crafting this argument! The input has also led us to co-author a rigorous version of the argument, with a paper due for peer review shortly. If anyone would like to see a draft, please let us know — feedback is very welcome.
There are several argument submissions pending review and others not included in the above: more to come in the next update. As always, please get in touch if you have new arguments/counter-arguments to add to the site — or if you’ve spotted a new argument out in the wild. You’re also very welcome if you’d like to help review submissions from others — we do have a backlog!
2) What content / engagement would you like?
This is a new initiative and we want it to be helpful! What would you find most valuable? Pick the top one below (poll live via Substack) or, even better, tell us direct.
You can share your ideas by commenting on the post, DM’ing me (Chris) or others you know at CoSI, or by replying to the Substack email alerts (if you’re a subscriber). You can also submit feedback (anonymously if you wish) via the site.
3) Call for new content! ‘Indicators of consciousness’
Are there particular theories of consciousness you know well? We’re adding a new section to the website later this year and need your help.
The idea is to extract specific indicators which would identify consciousness in a given physical system (a computer, an animal, etc.), assuming that theory is true. Most theories are works-in-progress, so we also want to identify gaps in those indicators or possible ways the indicators could be refined through further research.
In addition to the usual reward scheme ($100 for new arguments; smaller amounts for minor revisions), we have a $1k bounty for the best submission — provided we get at least five that qualify for the usual scheme.
This exercise has already been conducted for a few theories; the leading work is from Patrick Butlin, Rob Long, and colleagues. You can also check out this X thread with more context and a few other examples being developed, as well as this research paper on theory aspects with different indicators.
Scope: We need to work with theories that we can reference formally, e.g. at least one peer reviewed publication that explains the theory.
Post written by Chris, with many thanks to the colleagues, funders, and volunteers that make the Co-Sentience Initiative possible.