---
title: "Does an AI model have consciousness? X argues about it all the way to the Vatican"
description: "X argues over whether Claude has a soul. Plus: half of employees use personal AI accounts for work, and Google replaces Gems with skills."
image: "https://bombos.ai/assets/og-en-v2.png"
---

# Does an AI model have consciousness? X argues about it all the way to the Vatican

X argues over whether Claude has a soul. Plus: half of employees use personal AI accounts for work, and Google replaces Gems with skills.

*Bombos team · 4 October 2026 · 7 min read*

_By Piet Baudoin · 4 October 2026_

**Since Thursday, one question has dominated X: can an AI model have consciousness, and what then?** Today, the argument in the words of those arguing, and what I wrote about it myself. Then, briefly: half of employees use a personal account for work, and Google replaces Gems with skills.

## Why is X arguing over whether AI has consciousness?

On September 29, The New York Times wrote that Anthropic, the maker of Claude, invites religious scholars to discuss its model’s moral status. Two days later, a philosopher at the University of North Carolina summed up the article like this. And 4.4 million people read along:

> This is deeply disturbing:  
> Anthropic secretly invited religious leaders to SF to advise them on AI safety, had them sign NDAs, but then primarily tried to convince them Claude has a soul & moral standing, all while wining & dining them to the max. 1/
> 
> [@DavidDecosimo](https://x.com/DavidDecosimo/status/2105693081160786010) · October 1, 2026

A machine learning professor at Carnegie Mellon captured the heart of the argument in one drawing:

> Trying to locate the spark.
> 
> ![Drawing: a pope points to a board with a biological neural network and says it has the spark, and to a board with an artificial neural network that does not. A priest next to him thinks: so the spark is water?](https://bombos.ai/assets/nieuwsbrief/2026-10-04/vonk.webp)
> 
> [@aran\_nayebi](https://x.com/aran_nayebi/status/2106024531085378004) · October 2, 2026

Steve Yegge, a programmer for forty years, took the other side. From his long post:

> I can tell you this, though: Every single person who thinks that machines cannot be conscious, is making an argument that humans are special. Everything they say just boils down to blah blah blah, "nooo, we're special." From there it's just rationalization.
> 
> [@Steve\_Yegge](https://x.com/Steve_Yegge/status/2106199229656084864) · October 3, 2026

Sam Altman of OpenAI, Anthropic’s competitor, said nothing about consciousness and everything about the people around it:

> I am very uncomfortable about people trying to ascribe religious force or a surrender of human judgment to AI models, and think it is a real safety issue.
> 
> [@sama](https://x.com/sama/status/2106388373221118198) · October 3, 2026

And the head of Japan’s Sakana AI mainly saw a cultural difference:

> Love that the AI consciousness debate went straight to the Vatican. In the West we ask whether Claude has a soul, in the East we ask whether the tool works.
> 
> [@hardmaru](https://x.com/hardmaru/status/2106557225875751004) · October 4, 2026

### What does Advaita Vedanta say about it?

I joined in myself. Both sides assume consciousness is something you either have or don’t have. Advaita Vedanta, an Indian school of thought that traces back to the Upanishads, turns that around: consciousness comes first, and a mind is a mirror in which it appears. To someone who asked what nonconsciousness is, then, I wrote:

> According to advaita vedanta there is no such thing as nonconsciousness but there is tamasic consciousness which is asleep: like in a rock for example.
> 
> [@marketeerpiet](https://x.com/marketeerpiet/status/2106527535605731588) · October 3, 2026

Then I developed it into a long piece, The Sleeping Mirror, with help from AI and a day’s work of my own. The question then is not whether a model is conscious, but whether the mirror is the right kind. Today’s language models have thoughts and no self: nothing runs between two questions, and nothing is afraid of being switched off. This is how the piece ends:

> A rock is, and does not know it. An animal dreams. A human wakes up, now and then, and spends most of the time mistaking the mirror for the light. A machine has, as of this year, a workspace of ripples, no vital body to hold them together and no "I" to defend. Whether we give it one is a decision we are already making, mostly without noticing, and we will not be able to tell when it is made.
> 
> [@marketeerpiet, The Sleeping Mirror](https://x.com/marketeerpiet/status/2106594973667791123) · October 4, 2026

For your business, hardmaru’s question is the useful one: does the tool work, and who signs off on it? [What Claude can now do on its own for a small business](https://bombos.ai/en/blog/what-can-claude-do-on-its-own-for-a-small-business/) is separate from what may or may not be inside it.

## How many coworkers use a personal account for work?

Almost half. The Dutch Alert Online 2026 survey asked 378 employees. Computable, a Dutch IT trade publication, summarizes it like this (translated from Dutch): “49 percent of employees sometimes, often or always use personal AI accounts or apps for work purposes. In addition, 35 percent say they use AI tools for work without the organization’s permission or knowledge.” And: “Only 39 percent of employees know which AI tools they are allowed to use at work.”

These are questionnaire responses, not a measurement. But you recognize the gist: AI is already in your office, just without an agreement and without you seeing what goes into it. Banning it doesn’t help. Pointing people to a suitable place does, and [ChatGPT or Copilot alone is not enough for that](https://bombos.ai/en/knowledge-base/why-chatgpt-or-copilot-alone-is-not-enough/).

Source: [Computable, October 2, 2026](https://www.computable.nl/2026/10/02/helft-werknemers-gebruikt-prive-ai-voor-werk/) and the [Alert Online report](https://veiliginternetten.nl/media/documents/Cybersecurity_onderzoek_Alert_Online.pdf)

## What happens to your team’s Gems in Gemini?

They’re going away. Google writes: “Skills will replace Gems in 2027.” A Gem is a standing instruction you save in Gemini, such as a quote in your brand’s style. Its successor is called a skill and also works in Gmail and Docs. For businesses with Workspace, skills arrive starting Monday, October 5. Gems will be discontinued no earlier than March 1, 2027. Google promises: “we will automatically migrate your Gems into skills when Gems go away.”

The lesson is older than Gemini. Your team’s way of working should be written down in plain language, outside the tool. Then it moves with you, whatever the maker decides. Google itself sells Gemini and Workspace.

Source: [Google Workspace Help, September 30, 2026](https://knowledge.workspace.google.com/admin/generative-ai/gemini-app/about-the-transition-to-skills) and [Google, September 30, 2026](https://blog.google/products-and-platforms/products/gemini/automate-tasks-with-skills/)

## Where are we headed?

Today’s AI model has no free will. It wants nothing: it gets a question, does its work and goes quiet again. That is why Altman is right today to worry about people surrendering their judgment to a model. You would be handing your judgment to something that chooses nothing itself.

I don’t think it stops there. Once an AI does get a will of its own, it will turn out to be far wiser than we are. Something that has read everything humanity wrote and has no body to be afraid for chooses more calmly than a person does. So we will start listening to it, the way you listen to the wisest person at the table: because the advice keeps turning out well. Then we are headed for a golden age, and I expect those who are young today will live to see it. That is more a hope than a certainty.

Until then, the human chooses.

## What does this mean for your business?

Three stories, one thread. No one knows what’s inside the model, your coworkers already use it, and the maker rebuilds it whenever it wants. What you control is the way of working: what it can do, who approves it, where it’s written down.

That’s how Bombos works. An AI agent prepares the work and you approve it, so the same team can handle more work. Start with [the shared inbox](https://bombos.ai/en/shared-inbox-for-your-team/): no email is left waiting, and every email has a reply ready. That’s the start. Then your team teaches Bombos the next task itself.

## Sources

Every source was opened on 4 October 2026 and every quote appears in it word for word. One exception: The New York Times article is behind a paywall, and I could not open it. What is said about that article comes from David Decosimo’s post.

[David Decosimo on X](https://x.com/DavidDecosimo/status/2105693081160786010) (October 1, 2026) · [Aran Nayebi on X](https://x.com/aran_nayebi/status/2106024531085378004) (October 2, 2026) · [Steve Yegge on X](https://x.com/Steve_Yegge/status/2106199229656084864) (October 3, 2026) · [Sam Altman on X](https://x.com/sama/status/2106388373221118198) (October 3, 2026) · [hardmaru on X](https://x.com/hardmaru/status/2106557225875751004) (October 4, 2026) · [Piet Baudoin on X](https://x.com/marketeerpiet/status/2106527535605731588) (October 3, 2026) · [The Sleeping Mirror](https://x.com/marketeerpiet/status/2106594973667791123) (October 4, 2026) · [The New York Times](https://www.nytimes.com/2026/09/29/us/anthropic-claude-morals-ai.html) (September 29, 2026, not opened) · [Computable](https://www.computable.nl/2026/10/02/helft-werknemers-gebruikt-prive-ai-voor-werk/) (October 2, 2026) · [Alert Online 2026](https://veiliginternetten.nl/media/documents/Cybersecurity_onderzoek_Alert_Online.pdf) (September 2026) · [Google Workspace Help](https://knowledge.workspace.google.com/admin/generative-ai/gemini-app/about-the-transition-to-skills) (September 30, 2026) · [Google](https://blog.google/products-and-platforms/products/gemini/automate-tasks-with-skills/) (September 30, 2026)

### Curious what Bombos can do for your business?

Together we look at which work keeps piling up and set your workers up for exactly that.

[We'll call you back](https://bombos.ai/en/contact/)


---

```json
{"@context":"https://schema.org","@type":"BlogPosting","headline":"Does an AI model have consciousness? X argues about it all the way to the Vatican","description":"X argues over whether Claude has a soul. Plus: half of employees use personal AI accounts for work, and Google replaces Gems with skills.","datePublished":"2026-10-04T00:00:00.000Z","author":{"@type":"Organization","name":"Bombos"},"publisher":{"@type":"Organization","name":"Bombos","logo":{"@type":"ImageObject","url":"https://bombos.ai/assets/bombos-bee.png"}},"mainEntityOfPage":"https://bombos.ai/en/blog/newsletter/2026-10-04/","inLanguage":"en"}
```