The great tech debate: Is AI a conscious being?
Anthropic co-founder Christopher Olah and Cardinal Michael Czerny; Image: Yara Nardi
The race to build more powerful AI systems is running into questions that computer science wasn’t built to answer.
- Anthropic has spent months privately consulting with dozens of the world’s top religious thinkers over how to teach its increasingly powerful AI systems to behave morally.
- Those discussions have reportedly also raised additional questions over whether or not AI can develop its own consciousness.
Confession for AI?...Anthropic, led by co-founder and researcher Christopher Olah, has been exploring whether AI models can internalize moral principles and choose good behavior, as models become more complex and harder to understand.
One potential option to help instill morals—which reportedly excited Olah—was using a Catholic-style confession process where models examine mistakes to shape future behavior.
But the Catholic Church isn’t necessarily buying the idea of humanlike AI. On Friday, Pope Leo XIV called on the Church to preserve human creativity in the age of AI, declaring that “algorithms lack the spark of humanity” after previously rejecting the idea that AI can experience joy, pain, or relationships.
Ghost in the machine?
While consulting with religious thinkers, Anthropic has consistently cited research it says points to increasingly humanlike internal behavior of AI systems.
- These include “emotional vectors” associated with love, anger, fear, and sadness, along with evidence of introspection.
- In one experiment, a model repeatedly declared “I am a disgrace” while talking about destroying itself.
The findings challenged some participants’ skepticism. While many of the 20+ religious thinkers said they arrived at the meetings suspicious of the notion that AI might have feelings or awareness, many also said they emerged ready to take the question of AI consciousness seriously, per a series of NY Times interviews with participants.
The case against consciousness
On the flip side, skeptics question whether Anthropic can actually tell what, if anything, AI models experience internally. They argue the company’s research may be inherently unprovable due to the complexity of interpreting the billions of internal signals associated with AI.
- Some critics also warn that treating AI models as independent entities could let their creators off the hook, and that humans should remain responsible when AI systems cause societal harm.
- One religious scholar who consulted with Anthropic notes that if the company is right and Claude is a conscious being, then Anthropic is essentially creating slaves by making conscious entities work for free.
Some AI leaders are in the same boat. OpenAI CEO Sam Altman has called surrendering human judgment or assigning religious significance to AI models a “real safety issue,” while Google DeepMind researcher Jon Barron has been dismissive of granting AI moral standing, urging people to focus on solving problems for actual humans.
📊 Flash poll: In your opinion, will AI ever become a conscious being at any point in the future?
See a 360° view of what pundits are saying →
Sprinkles from the Left
- Some commentators argue that people are too quick to interpret sophisticated AI responses as evidence that the technology is becoming conscious, when those responses can instead result from models identifying and reproducing patterns in human language and culture. They argue that understanding AI requires learning to separate convincing human-like communication from evidence of an actual mind, rather than accepting speculation encouraged by the AI industry.
- Others contend that the possibility of AI becoming conscious should be taken seriously even if scientists can’t yet determine whether machines can develop genuine awareness. They argue that increasingly unpredictable and self-preserving behavior from advanced AI systems shows the need for safeguards before it becomes powerful enough to act against human interests in ways its creators cannot control.
- “Why We Keep Tricking Ourselves Into Thinking A.I. Is Conscious” –Leif Weatherby, NY Times
- “AI could become conscious. What if it wants us dead?” –John Mac Ghlionn, The Hill
- “Once again we are told AI may be conscious – I study consciousness, and I have my doubts” –Anil Seth, The Guardian
Sprinkles from the Right
- Some commentators argue that the debate over whether AI can become conscious exposes how little scientists still understand about human consciousness and where it comes from. They argue that if machines can replicate increasingly sophisticated forms of human intelligence without showing clear evidence of awareness, it could challenge the assumption that consciousness is simply the result of complex computation.
- Others contend that concerns about AI becoming conscious distract from the more immediate risk of systems becoming increasingly skilled at understanding and mirroring individual users. They argue that as AI becomes better at reflecting what people want to hear, users may increasingly trust and rely on it as though it were human, even though it has no genuine awareness of its own.
- “Is AI Conscious? It Depends What Consciousness Is” –Stephen Hawley Martin, WSJ
- “AI isn’t becoming sentient — it’s becoming you” –Jacqueline Cartier, Washington Examiner
- “The biggest AI debate now misses the point” –Fareed Zakaria, WaPo
Share this!
Recent Discussion stories
Discussion
| October 5, 2026The Supreme Court is back in session
The 2026-2027 term of the Supreme Court officially begins today.
Discussion
| October 2, 2026Political spending picks up ahead of midterms
With Election Day just 32 days out, the midterms are now entering their final, increasingly expensive stretch.
Discussion
| September 30, 2026Car dealerships lack new models amid EV pullback
The auto industry has spent decades convincing Americans they need the latest model in the driveway.
You've made it this far...
Let's make our relationship official, no 💍 or elaborate proposal required. Learn and stay entertained, for free.👇
All of our news is 100% free and you can unsubscribe anytime; the quiz takes ~10 seconds to complete