Theology and AI (Efficient or Humanistic Society): On AGI, Superintelligence, the extinction headlines, and where faith belongs in the conversation.

If the news has left you uneasy lately, you are not alone. On September 8, an AI researcher resigned from his lab and warned that the people building this technology believe it could end humanity by the end of the decade. A Nature report says the post drew more than 100 million views in a single day. Millions of ordinary people learned in one week that AI experts openly debate the odds of human extinction.

This post is not meant to add to the panic, and it is not meant to wave the concern away. It is an attempt to separate what is known from what is feared, and then to ask where people of faith should stand.

Where We Left Off: What I Wrote in 2021

Back in 2021, I wrote History and Types of AI, where I explained that AI falls into three categories:

Narrow (or Weak) AI, which performs specific tasks such as translation, voice assistants, and facial recognition.
General (or Strong) AI, which could perform the full range of intellectual tasks a human can, including understanding context and making judgments.
Super AI, which would far exceed human intelligence in nearly every area, from creativity to problem-solving to general wisdom.

At that time, I noted that General AI did not exist outside science fiction and that Super AI was still a prediction. I also pointed out that even if General AI were achieved, such machines would likely not possess sentience, the ability to actually experience and feel. I also mentioned that figures like Elon Musk and Stephen Hawking feared Super AI might lead to the end of the human race.

Five years later, the first category is everywhere, the second is being loudly claimed, and the third has moved from the margins into the headlines. That is why it is worth revisiting.

AGI and Superintelligence: Then and Now

	Narrow AI	General AI (AGI)	Superintelligence (ASI)
What it is	Good at specific tasks	Matches capable humans across many fields	Far beyond the best human minds in nearly everything
My 2021 view	Exists today	Does not exist; unknown if possible	Does not exist; a future prediction
October 2026	Built into daily life	Claimed as close by some labs; disputed by others	Not achieved, but named as an explicit goal by several major labs
The core question	How do we use it well?	Who decides when it has arrived?	Can it be kept under human control?

A simple picture: AGI would be like a brilliant colleague. Superintelligence would be something humanity has never met, and honestly, no one can say with confidence what that looks like.

What Is Actually Happening

In an August interview with TIME, OpenAI CEO Sam Altman said the company expects to have an internal system by the end of 2026 that it calls AGI. Here is the catch: there is no agreed definition of what counts as "achieving" AGI. A company can effectively declare success by its own standard.

Not everyone is convinced. AI researcher Gary Marcus has argued that AGI within five years is extremely unlikely, and that claiming otherwise is marketing while problems like hallucinations and reasoning errors remain unsolved.

Meanwhile, the race toward superintelligence is openly acknowledged. OpenAI, Meta, Safe Superintelligence Inc., and Microsoft have all named it as a goal, even though it has not been achieved.

The Warnings

The loudest alarm came on September 8, when Anthropic researcher Jacob Coxon resigned. Hours later, his colleague Evan Hubinger, who leads alignment science at the same company, said he puts the risk of AI-driven human extinction above 10% within the next decade. Hubinger also said the risk today is low. His worry is that AI systems could eventually improve themselves to the point of becoming an existential risk, and that no one yet has a plan to solve "alignment" for superintelligence. (Alignment simply means making sure a powerful AI actually pursues goals that are good for people.)

He is not alone. Geoffrey Hinton, the Nobel-winning "godfather of AI," has estimated a 10 to 20% chance of AI-caused extinction. The Machine Intelligence Research Institute has gone much further, saying its leadership puts the odds above 90% if there is no aggressive policy response. Around 1,400 AI-sector employees reportedly signed an open letter calling for a slowdown, and Senator Bernie Sanders has proposed a ban on superintelligence.

In 2021 the loudest voices were outsiders like Hawking and Musk. What has changed is that today, some of the warnings are coming from inside the labs.

The Pushback

Serious people see this differently, and a fair post has to say so.

In Nature, RAND researcher Michael Vermeer said extinction forecasts involve so many untestable claims that the debate becomes "more like faith than something scientific." Heidy Khlaaf of the AI Now Institute calls the extinction framing fear-mongering and asks regulators to focus on disinformation, bioweapon risks, and false military intelligence instead. Others, including a widely discussed Jacobin essay, argue that doom talk can double as hype, noting that apocalyptic warnings have not hurt the valuations of the companies making them.

Even some who take the risk seriously say the nearer danger is people, not machines. Computer scientist Stuart Russell has said that in the next year or two, the greater risk is human misuse, such as an AI-enabled terror attack.

A Balanced Reading

Nobody knows the timeline. Predictions range from "this year" to "not in our lifetimes."
The extinction numbers are educated guesses, not measurements. A 10% and a 90% estimate cannot both be right, and neither can be tested.
Everyone has incentives. Companies benefit from hype, and critics and doomsayers benefit from attention too.
Present harms are real. Job disruption, deepfakes, and misuse by bad actors need our attention whatever happens with superintelligence.
The concern is not fringe. When insiders say publicly that they lack a solution for controlling something they are racing to build, wisdom says take it seriously without panicking.

Where Spirituality Belongs

This is where people of faith bring something the headlines lack.

1. We are made in God's image. AI is made in ours.

Genesis 1:27 tells us humanity bears the image of God. AI is a human creation that can imitate our words and reasoning, but thinking is not the same as being a person, and intelligence is not the same as a soul. A machine can model a mind. It does not carry the breath of God.

2. Human ambition has a long story.

At Babel (Genesis 11), people united around technology and a desire to make a name for themselves. Scripture does not condemn building. It warns about building without God. The question to ask of AI is not only "Can we?" but "For whom, and under whose authority?"

3. Do not make AI an idol, and do not make it a demon.

Psalm 115 describes idols as the work of human hands: they have mouths but cannot speak. Some will treat AI as an oracle or savior. Others will treat it as an unstoppable doom. Both give it more power than it deserves.

4. God's sovereignty is not threatened by a server farm.

Colossians 1:17 says that in Christ all things hold together. That is no excuse for carelessness, since Scripture also calls us to wisdom, stewardship, and justice. But we can face this moment with sober eyes instead of fear. "Do not fear" (Isaiah 41:10) was never conditional on a calm news cycle.

5. Wisdom still comes from above.

AI can give us information quickly. It cannot give us repentance, forgiveness, or presence with the suffering. James 1:5 still points us to the One who gives wisdom generously.

A Tool, Not a Throne

I have said before that AI is a remarkable tool, and I still believe that. A tool serves the craftsman, and the craftsman does not serve the tool. In Create First, Consult AI Later, I argued that the order matters: bring your own thinking and prayer first, then let AI help you refine. In Is Using AI Like Stealing?, I wrote that using this technology well is stewardship, not theft. The same principle applies to the bigger questions we face now.

Here is how I would suggest we live in this moment:

Stay informed, not frightened. Read beyond the headline, and notice who is speaking and what they stand to gain.
Hold the builders accountable. Ethics does not stop at our own use of the tool. Support wise regulation, transparency, and the protection of human dignity.
Pray before you scroll. Anxiety feeds on constant headlines. Anchor your mind in Scripture first.
Keep your own voice and calling. Create first, consult AI later, especially in ministry and sacred work.
Remember who holds the future. Intelligence, human or artificial, is a gift that carries responsibility. It is never the Giver.

Whether AGI arrives in two years or never, the most important questions are not technical. Who do we trust? What do we love? Whom do we serve?

We are not anti-technology. We are pro-human, and above all, we are people who trust the God who made humanity in His image and holds all things together.

What are your thoughts? Do the extinction warnings concern you, or do you see them as hype? I would love to hear from you in the comments.

Sources and Further Reading

News in this area is changing quickly. These were the sources used for the 2026 reporting above, and I encourage readers to check the originals:

Theology and AI (Efficient or Humanistic Society)

Tuesday, October 20, 2026

On AGI, Superintelligence, the extinction headlines, and where faith belongs in the conversation.