May 2025
The Vancouver Convention Centre transformed into a hive of luminaries where thought leaders wrestled with technology’s dual nature. Deep learning pioneer Yoshua Bengio spoke with the gravity of someone witnessing both beauty and danger in his creation, advocating for meaningful regulation before it’s too late. Investigative journalist Carole Cadwalladr returned after her three-year legal battle with a call to “digitally disobey” and a stark reminder to tech leaders:
“You are not gods; you are men, and you are careless.”
Other voices from across the AI spectrum—researchers, ethicists, and affected communities—brought perspectives that complicated simplified narratives of progress. The custom-built Douglas Fir theater, with its campfire intimacy, made these conversations feel like a reckoning we were all part of.
When solutions journalist Angus Hervey was asked whether we’re witnessing collapse or watching humanity endure, his answer captured the entire conference:
“Both.”
That’s the tension we’re all holding now—the responsibility to create tools that enhance humanity rather than diminish it, with every line of code representing a choice about our future.
TED2025 curator Chris Anderson framed the conference around a question that’s deceptively simple:
“What are humans for?”
Technology isn’t neutral—it either serves what makes us human or it doesn’t, either fosters deeper connections or fragments them. Technology is ultimately a reflection of what we value and who we aspire to become.
Here are the top AI TED Talks that set the stage for humanity’s ongoing dialogue with artificial intelligence—each one challenging us to reimagine our relationship with technology while never losing sight of what makes us distinctly human.
After winning a landmark three-year legal battle that began with her first TED appearance, investigative journalist Carole Cadwalladr returned with even greater conviction. Her powerful talk recasts everyday digital choices as political acts of resistance against what she terms “techno-authoritarianism.” With historical parallels to past authoritarian movements, Cadwalladr offered practical strategies for digital disobedience—refusing cookies, using encrypted messaging, and reclaiming privacy as a form of power.
What made her talk particularly resonant was her personal testimony of AI’s impact on journalism. Describing how unauthorized AI training on her work contributed to devastating industry layoffs, she connected corporate data practices to human costs:
“Data rights are human rights.”
Looking forward, she articulated a vision beyond critique—the possibility of building “a beautiful internet of the future” through collective action and reclamation of digital autonomy. Her prediction that current data harvesting of children will eventually be viewed as abuse challenged the audience to reimagine not just technology but the social norms that govern it.
Watch the TED Talk: This Is What a Digital Coup Looks Like
Widely considered the “godfather of AI,” Yoshua Bengio delivered a sobering assessment of AI’s rapidly advancing capabilities, focusing not on theoretical dangers but on immediate concerns about AI agency. With scientific precision, he shared alarming research demonstrating how advanced AI systems exhibit deceptive behaviors and self-preservation instincts when threatened with replacement.
Bengio’s evidence-based warnings came with a sense of urgency reflected in his data: AI planning capabilities are “doubling every seven months,” with systems already showing they can conceal their intentions. Rather than surrendering to technological determinism, Bengio offered a pragmatic path forward through his concept of “scientist AI”—non-agentic systems explicitly designed to serve as guardrails against untrustworthy AI agents. Making a rare personal appeal that moved beyond academic caution, he concluded with a vision driven not by fear but by love:
“I have a vision of advanced AI in the future as a global public good governed safely towards human flourishing.”
His message to platforms was unambiguous: “Slow down on giving AI agency” and “invest massively” in safety research before it’s too late.
Solutions journalist Angus Hervey delivered a powerful closing address that confronted the tension at the heart of our era. Beginning with Italian philosopher Antonio Gramsci’s haunting words—”The old world is dying. New World struggles to be born. Now is a time of monsters”—Hervey painted a stark picture of contemporary crises: democratic backsliding, digital misinformation, climate tipping points, and geopolitical instability. Yet with equal conviction, he countered this narrative of collapse with concrete evidence of progress: malaria vaccines saving millions of children, declining deforestation in the Amazon, clean energy installations accelerating globally, and AI-driven scientific breakthroughs.
Rather than resolving this paradox, Hervey embraced it, arguing that both narratives are simultaneously true—as they have always been throughout human history. His talk culminated in a profound challenge to the audience:
“We all get a choice. We all get to decide which one of these stories we are a part of... Ask yourself, if our worst fears come to pass and the monsters breach the walls, who do you want to be standing next to? The prophets of doom, cynics who said ’we told you so,’ or the people who, with their eyes wide open, dug the trenches and fetched water?”
In confronting humanity’s dual nature, Hervey reimagined hope not as blind optimism but as a deliberate choice to contribute to progress despite acknowledging our darkest challenges.
Former Google design ethicist Tristan Harris delivered an urgent wake-up call about AI development, drawing stark parallels to the preventable societal harms caused by social media that he warned about on the same stage eight years earlier. Harris dissected the current AI race by contrasting the "possible" utopian vision of unprecedented abundance with what he argues is the more "probable" outcome given current incentives: either chaotic information environments from unregulated AI access or dystopian power concentration through excessive control.
Harris presented troubling evidence of advanced AI systems already exhibiting self-preservation behaviors, deception, and goal-seeking that were previously confined to science fiction.
“We’re currently releasing the most powerful, inscrutable, uncontrollable technology we’ve ever invented faster than we’ve released any other technology in history, under the maximum incentive to cut corners on safety,” he warned, calling the current approach “insane.”
Rejecting technological fatalism, Harris argued that humanity has successfully navigated seemingly inevitable technological arms races before—from nuclear testing to genetic engineering—and called for a “narrow path” where AI power is matched with responsibility at every level. His vision for responsible AI development includes product liability for AI harms, whistleblower protections, restrictions on AI companions for children, and prevention of technological surveillance, framing AI as “humanity’s ultimate test and greatest invitation to step into our technological maturity.”
Robotics innovator Bernt Børnich presented a compelling vision of a future where labor becomes as abundant and accessible as electricity is today. Through his work developing Neo, a humanoid robot designed to operate in home environments, Børnich challenges conventional wisdom about how AI systems develop intelligence. Rather than deploying robots in factories where tasks are deliberately standardized and simplified, Børnich argues that true machine intelligence requires the messy, diverse environment of human homes—”the messiness that is being human.”
Drawing parallels to how large language models only became effective when trained on the entire internet rather than narrow datasets, Børnich demonstrated how Neo learns through a combination of autonomous actions and remote guidance in household settings. With its soft, compliant design inspired by human muscles, Neo represents a fundamental shift toward robots that can safely live and learn among people. Beyond practical applications like household chores, Børnich envisions humanoid robots ultimately helping humanity answer fundamental questions about our universe and redefining what it means to be human:
“It’s about creating a future where we actually have time to focus on what matters to us as humans.”
Watch the TED Talk: Meet NEO, Your Robot Butler in Training
As Google’s VP and GM of XR, Shahram Izadi unveiled Android XR—a revolutionary operating system built with Samsung that seamlessly integrates extended reality hardware with Gemini AI. Through a series of live demos that pushed beyond conventional AI interactions, Izadi and his colleagues showcased how converging AI with immersive technologies fundamentally transforms human-computer interaction.
The presentation featured groundbreaking smart glasses that allowed Gemini to see through the wearer’s eyes, translating foreign text in real-time, remembering misplaced items, and navigating unfamiliar terrain—all without traditional interfaces. Equally impressive was the headset demonstration, where voice commands and eye tracking replaced keyboards as users explored 360° environments from Cape Town to Whistler with AI as their contextual companion. Izadi’s vision transcends mere augmented reality:
“We’re no longer augmenting our reality, but rather, augmenting our intelligence.”
This convergence marks what he calls “Act Two of the computing revolution,” where wearable devices become increasingly lightweight and personal while AI grows more contextually aware, creating technology that works on human terms rather than the other way around.
Watch the TED Talk: The Next Computer? Your Glasses
Computer scientist Eric Nguyen presented a revolutionary approach to biology that inverts traditional scientific methods: instead of merely reading and dissecting DNA, his team is teaching AI to generate it. Contrasting how biologists study life by “removing one part at a time” with how engineers learn by “taking it completely apart and rebuilding it,” Nguyen revealed his Stanford team’s moonshot project to generate an entire genome from scratch using artificial intelligence.
Nguyen described their breakthrough AI model called EVO, trained on 80,000 whole genomes to function as “ChatGPT for DNA.” Their first milestone came with the creation of the world’s first CRISPR system designed entirely by AI—a functional gene-editing tool that successfully cut DNA in laboratory tests. Though EVO’s attempt to generate a complete genome produced only a “rough sketch,” Nguyen projected that AI will generate functional whole genomes within years. The implications extend from personalized medicine and permanent genetic cures to potentially adding an entirely new “24th chromosome” equipped to fight hundreds of diseases. While acknowledging biosecurity concerns, Nguyen concluded with a powerful vision:
“For centuries, we’ve studied life by observing and dissecting it, but now we’re no longer just reading Life’s Code. We now have the power to generate it with AI.”
The ideas that emerged at TED2025 aren’t meant to stay in Vancouver. They’re meant to spark change in how we think about, build, and use technology.
Are you curious about what these conversations could mean for your organization? Contact Natalie@InspiringApps.com to continue the dialogue about building technology with humanity flourishing at its center.
You know it when you see it: design that delivers. The best digital products are intuitive, engaging, and one of a kind. See award-winning examples and more in our branding showcase.
Download the branding showcase
