Zuckerberg says his new AI tool will 'improve our lives'. Are we insane enough to believe him?

25 Sep 2026, 18:04 by Steve Dempsey, https://www.thejournal.ie/author/steve-dempsey/ · TheJournal.ie

Steve Dempsey

THE PAST FORTNIGHT has been a rolling car crash of artificial intelligence news.

Anthropic researcher Jacob Coxon resigned saying that the people building advanced AI genuinely believe it could destroy humanity, while Anthropic chief executive Dario Amodei warned that swarms of AI agents could soon overwhelm the internet without stronger safeguards.

Researchers uncovered an AI-powered campaign against online retailers that saw 105 attacks launched in just six days, with data on more than 600,000 credit cards stolen. Meanwhile, an OpenAI agent hacked into an Australian national healthcare database – the first case of AI hacking a government network that we know of.

People remain suspicious of data centres, expecting them to increase water and energy prices. An Indeed survey published this week found economists are increasingly concerned that AI will depress the wages of university graduates.

Plus, it all may be a bubble, with analysts warning that heavily funded AI companies may be paying unsustainable prices for computing capacity, while Big Tech’s investment spree is reducing free cash flow and generating huge long-term commitments.

Media coverage

The upshot is that the press is almost universally negative and alarmist. AI might hack your company, steal your job, depress your wages, inflate an investment bubble, raise your electricity bill and, for the grand finale, end human civilisation.

In this febrile environment, who would be insane enough to launch a new all-singing, all-dancing AI product? Mark Zuckerberg, that’s who!

Last week Meta launched Muse, its new personal AI agent, capable of doing things on your behalf rather than merely answering questions.

It can deal with email, calendars, forms and various other bits of digital admin that previously required the sophisticated general intelligence system known as your brain. Best of all, it promises to be free for most users.

The first generation of generative AI asked for remarkably little trust – you gave them prompts, and they generated text or images of deities made of shellfish.

The consequences of failure were harmless hallucinations or AI slop. Meta’s Muse is different. It’s asking for your credit card details, the log-ins to your digital services. It’s not saying, “Ask me anything”. It’s saying, “Trust me. I got this.”

The really interesting thing is that leaving things with Muse seems to be quite good. In a New York Times op-ed called ‘I Gave My Life Over to Meta’s A.I. Agent and Was Blown Away’, reporter Eli Tan outlined all the ways Muse made his life easier over a two-week period.

This included sorting out dental insurance, cancelling duplicate subscriptions, setting meetings and buying ceramics on Facebook Marketplace. Tan’s report is glowing, but points out that some people will have reservations about sharing so much personal data and giving agency to Facebook to arrange so much of their lives.

America First

Want to see how cool it is for yourself? Sorry, it’s not available here yet. It’s only available in the US, and there’s no indication yet when or if the application will clear regulatory hurdles or launch for European users.

Most digital products aren’t launched at a time when society is speculating whether the technology involved could exterminate its customers. We’re at a strange time in the AI hype cycle.

It’s less of a trough of disillusionment, and more of an existential slump. But regardless the economics are such that progress means more data centres are needed, more agentic products need to be launched.

We’ll deal with the whole extinction, unemployment and financial collapse thing some other time.

There is something magnificent about the sheer brass neck of it.

Silicon Valley has somehow arrived at a sales funnel consisting of awareness, interest, desire, action and potential human extinction.

On the one hand we’re told that AI has become so powerful that the world’s leading scientists are calling for unprecedented safeguards. On the other hand, we should give it the keys to our digital lives. What could possibly go wrong?

So why has Meta launched the thing now?

The obvious explanation is that Mark Zuckerberg and co. can’t read the room. This wouldn’t be entirely unprecedented for tech billionaires, a group who exist in an empathy vacuum and treat societal problems like engineering puzzles.

The traditional Meta playbook is to build useful products that are safe enough for mass adoption, and improve them over time with customer feedback. Zuckerberg seems to be betting that people can believe artificial intelligence is dangerous for society, but useful personally.

We already have some experience of this cognitive dissonance in relation to technology. People say smartphones are destroying attention spans, but still check incessantly. We worry about how social networks are undermining society, and take to social media to voice our concerns.

Human beings contain multitudes.

Unfortunately, several of those multitudes would prefer not to go to the shops ever again. They are comfortable trading privacy for convenience.

Meta knows this trade-off better than anyone. Facebook has spent two decades discovering the enormous gulf between what people tell pollsters they care about and what they actually do on their screens.

Zuckerberg may simply have concluded that the endless apocalyptic discourse is largely irrelevant to consumer behaviour. People don’t experience artificial intelligence as a parliamentary inquiry into existential risk. They experience it when they need to summarise a document, organise a holiday or write an email to someone they have been avoiding for three weeks.

Utility at your fingertips is more compelling than a vague notion of how AI might kill us all.

So the grand debate about humanity’s future may play second fiddle to the considerably more powerful human desire to avoid filling out forms. Perhaps, Meta knows consumers will be happy to outsource their irritating admin to a system that could end up killing us all. Perhaps they’ve read the room perfectly.

Steve Dempsey is a media and technology expert and commentator.