
Geoffrey Hinton Says AI Needs Maternal Instincts. Here's What It Takes
Hinton doesn't think the solution to this threat is to force AI to submit to humans. Instead, he suggests building "maternal instincts" into AI models so that they genuinely care about us humans and therefore won't want to wipe us out.
Following my Forbes article on how Alan Turing, known as the father of AI, laid the foundation of AI by building a digital 'child-machine', Hinton's call for 'super-intelligent caring AI mothers' raises an intriguing question:
Is the entire AI industry made up of fathers and godfathers trying to build an AI family of artificial children and mothers? And what would it take for them to succeed?
1. Geoffrey Hinton As God, Not Godfather
In an interview on CNN, a reporter asks Hinton how easy it is from a technological perspective to build motherly instincts — 'is there an example of that being done at all?' Hinton's answer provides a rare insight into how he and AI developers like him understand themselves and the work they do:
'The only real example we have is evolution. Evolution obviously made a pretty good job with mothers.' Hinton says that the reason we lack technological examples is because AI developers haven't tried to build artificial maternal instincts – yet:
'They have been focusing on making these things more intelligent. But intelligence is just one part of a being. We need to make them have empathy towards us. And we don't know how to do that yet. But evolution managed and we should be able to do it too.'
So that's the idea: If evolution can do it, so can we.
But what does this idea imply? And who are the 'we' who can accomplish the same as evolution? Hinton is known as the godfather of AI, but the achievement he talks about seems to require a god rather than a godfather. Who else would understand themselves and the work they do as on par with evolution?
2. AI Shaped By Pain, Confusion, And Worry
In addition to a god, rather than a godfather, it takes some work to get from the lying AI we know today to the caring 'AI mother' Hinton calls for. What does this work look like? And where do we start?
While making AI intelligent has required computer scientists like Hinton to study and simulate neural networks, giving AI maternal instincts requires studying something entirely different. Maternal instincts are neither innate nor learned (neither part of the child brain/program nor the educational process, Turing set out to simulate).
They arise during pregnancy, are embodied at birth, and evolve throughout motherhood. From the outside, maternal instincts may be most associated with care and compassion. But from the inside, the development of these instincts through pregnancy, birth, and parenting is just as much about pain, confusion, and worry.
And that's just it. Maternal instincts develop from the inside. From a place nobody – not even the mother herself – is able to define and describe. How does Hinton and an industry of fathers and godfathers plan to study and simulate that?
3. Humans Stop Being Human
In a LinkedIn post commenting on Hinton's CNN interview, another AI godfather, Meta's Yann LeCunn, says that 'simple, low-level hardwired guardrail objectives' like "don't run over people" and 'don't flail your arm if there are people around, particularly if you are holding a kitchen knife,' would be the AI equivalent of instinct or drive in animals and humans.
This suggests that the fathers and godfathers of AI may have a chance to succeed in building an AI family of artificial children and mothers. All it takes is for children and mothers to stop being influenced by pregnancy, birth, and upbringing and start following simple, low-level hardwired guardrail objectives. That would allow the AI industry to say they were able to do what evolution did: create AI mothers who have the same maternal instinct as human mothers. Not because AI starts behaving like human mothers. But because human mothers start behaving like AI.
Unfortunately – or fortunately, depending on how you look at it – that is unlikely to happen. Just as Geoffrey Hinton says we need AI mothers rather than AI assistants because 'an assistant is someone you can fire, you can't fire your mother,' the pain, confusion and worry that come with pregnancy, birth, and upbringing is not something we can get rid of. So maybe that's what the AI fathers and godfathers should be focusing on: not how to create artificial children, mothers and maternal instincts, but how to use their paternal instincts to build AI that gives way to humans, for better or worse?

Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles


Tom's Guide
5 minutes ago
- Tom's Guide
iPhone 17 release date rumors — this could be the full schedule
Summer's winding down, and that means Apple's big annual keynote is just around the corner, promising exciting new announcements about the iPhone 17 line-up, new Apple Watches, and more. While Apple hasn't revealed the rollout's full schedule yet, the rhythm of iPhone launches feels as predictable as the turning leaves. If history serves, we have a pretty good guess when you'll be able to get your hands on a shiny new iPhone 17. Rumors suggest Apple will unveil its major fall product line-ups on September 9. If history serves, we expect the event to take place at the company's Cupertino campus, where the iPhone 17 series will be revealed. Notably, Apple's rumored to shake up the iPhone lineup this time around by replacing the Plus model with the super-thin iPhone 17 Air. In addition to the iPhone 17 and iPhone 17 Pro, we expect Apple to debut the Apple Watch Series 11 and possibly the Apple Watch Ultra 3. A new version of the AirPods Pro, an updated Apple TV 4K, and the long-awaited smart home hub HomePod 3 could also make an appearance. Apple event announcement date 8 a.m. PT / 11 a.m. ET on August 26 iPhone 17 launch event September 9 iPhone 17, Air, and Pro available for pre-order September 12 iOS 26 release date September 15/16 iPhone 17, Air, and Pro on sale September 19 Apple's yet to confirm the keynote's official date, but trusted voices like Bloomberg's Mark Gurman and Forbes suggest we should hear more on that by Tuesday, August 26. That's also when invites should start rolling out via email to select members of the press. Looking at the time tables of previous iPhone generations, we're likely to see Apple open pre-orders for the iPhone 17 a few days later. Apple tends to roll out pre-orders on the Friday after the initial announcement, with sales going live the Friday after that. So if history serves, the iPhone 17 could go on sale on Friday, September 19. As for iOS 26, the latest update expected to ship with the iPhone 17, that's easy to pin down as well. Last year, Apple pushed out iOS 18 on September 16, exactly one week after its annual keynote and four days before the iPhone 16 series went on sale. If Apple follows the same pattern, we can expect iOS 26 to arrive on either Monday, September 15, or Tuesday, September 16. We don't have long to wait until we learn more. All signs point to Apple's iPhone 17 launch event being just a few short weeks away. Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button. Get instant access to breaking news, the hottest reviews, great deals and helpful tips.


TechCrunch
5 minutes ago
- TechCrunch
AI-powered stuffed animals are coming for your kids
In Brief Do A.I. chatbots packaged inside cute-looking plushies offer a viable alternative to screen time for kids? That's how the companies selling these A.I.-powered kiddie companions are marketing them, but The New York Times' Amanda Hess has some reservations. She recounts a demonstration in which Grem, one of the offerings from startup Curio, tried to bond with her. (Curio also sells a plushie named Grok, with no apparent connection to the Elon Musk-owned chatbot.) Hess writes that this is when she knew, 'I would not be introducing Grem to my own children.' As she talked to the chatbot, she became convinced it was 'less an upgrade to the lifeless teddy bear' and instead 'more like a replacement for me.' She also argues that while these talking toys might keep kids away from a tablet or TV screen, what they're really communicating is that 'the natural endpoint for [children's] curiosity lies inside their phones.' Hess reports that she did, eventually, let her kids play with Grem — but only after she'd removed and hidden the voice box. They still talked to it and played games with it; then they were ready for some TV.


Forbes
6 minutes ago
- Forbes
Are Browsers Key To An Agentic AI Future? Opera, Perplexity Think So
AI-powered conversational search engine Perplexity is in the news for offering to buy Google's Chrome browser for $34.5 billion. But in December of 2024, Perplexity considered buying The Browser Co. And just months ago, Perplexity reportedly offered to buy Brave, the privacy-focused browser, for about $1 billion. Why does Perplexity want a web browser so badly? Possibly because a browser just might be key to our agentic AI future. I recently interviewed Opera senior product leader Henrik Lexow on my TechFirst podcast. Opera, the 30-year-old browser company that pioneered tabbed browsing, pop-up blocking, and ad blockers, has about 300 million active users globally. This year, Opera was the first to bring AI agents right into our browsers in a project called Opera Neon. 'The agentic browser … is that sort of the new operating system?' Lexow asked during the podcast. 'It's a big question.' Regardless of where the browser goes, Opera's pretty sure about the future of the internet itself. That's a huge shift, by the way. An agentic internet would be a massive and fundamental change from a user-driven internet to an agent-driven version. In a user-driven internet, you search, you see results, you make selections, you click links, you fill out forms, you book flights, and you buy products. In a sense, you are the agent. In an agentic internet, you tell something – maybe your agentic browser – to do those things for you. Except you don't say search, look, select, and buy; you say buy me more of the underwear I got six months ago. The agent then consults its memory, forms a plan, takes multiple steps, and handles it all: from which underwear you bought, and where you bought them, to finding the same ones online (and maybe checking around for better prices), to adding them to cart, to checking out … and reporting back to you with the results. The agent – in this case potentially an agentic browser – is therefore essentially a personal assistant, a force multiplier. But will an agentic browser be the main way we engage with agents? Perplexity seems to think it's pretty important, given the company's persistent and repeated but so far unfruitful attempt to buy a browser. Opera certainly thinks so, if only because Opera has a browser, and a very innovative one at that. Opera launched AI in a browser back in 2023 in a project called ARIA. ARIA enabled contextual interactions within web pages in a GPT-based chat interface. Over time, that's evolved to a tripartite strategy under the Opera Neon brand: The reality is that for many of us, most of our work happens in a browser. I'm writing this story in a browser. I recorded the interview in a browser. I've researched Opera and Perplexity in a browser. I made episode art for the podcast in a browser (thanks, Canva). Opera's thesis is that having agents embedded where you work makes them vastly more useful: they have access to your history, to your work, to your sites and apps. Important note: the Neon agentic browser's AI lives locally on your hardware, making it your agent, not Opera's, and not your employer's. This should boost your privacy, which is critical if you're going to give an agent access to very personal information including, likely, your credit card. Of course, this is just one vision of the future. Apple with Siri, as justly maligned as it is, would have another vision. Google, with Gemini and its own vast fleet of Android-enabled phones, would have another. Microsoft's Copilot is another. And OpenAI, which has ChatGPT apps for mobile devices as well as full computers, might have yet another vision of how we'll integrate AI into our lives and work. So whether the browser will be the locus of our agentic AI future or not is yet to be determined. Remember the old proverb: every problem looks like a nail to the person who only has a hammer. But it seems like a fairly good bet to me.