3D dissolving human head made with cube shaped particles. 3D dissolving human head made with cube shaped particles. Getty
I’d be happy if by the time I retire, we have [artificial intelligence] systems that are as smart as a cat,” Yann LeCun, Meta‘s chief AI scientist, Turing Award winner and one of the founding fathers of deep learning, tells Newsweek as part of an ongoing series of conversations about the future of AI, “and that retirement is coming fast, by the way, so I don’t have much time.”
LeCun sees the extraordinary promise of AI on the horizon. But so far we haven’t seen this degree of success. While venture capital and corporate investment pours billions of dollars into AI dream factories promising revolutionary transformations—whether it’s curing cancer or finally taming the email inbox—a stark reality persists: Most artificial intelligence initiatives collapse under their own ambitions.
Visitors look at Tesla’s humanoid robot Optimus at its exhibition booth during the World Artificial Intelligence Conference (WAIC) in Shanghai on July 5, 2024. Visitors look at Tesla’s humanoid robot Optimus at its exhibition booth during the World Artificial Intelligence Conference (WAIC) in Shanghai on July 5, 2024. STR/AFP via Getty
The gulf between technological marvel and practical utility resembles a paradise island ringed by shipwrecks—the quest for supreme omniscience has left the tech landscape littered with sophisticated failures. In the pursuit of self-driving cars, Apple spent over $10 billion developing its autonomous car before abandoning the project entirely. GM burned close to $10 billion on its Cruise robotaxi unit before shutting it down in December 2024. Five years ago, Elon Musk said: “We’re headed toward a situation where AI is vastly smarter than humans and I think that time frame is less than five years from now.” But so far, we’re holding our own.
Against this backdrop of inflated expectations and deflating results, a more nuanced understanding has emerged from those like LeCun, who’ve spent decades wrestling with the actual mechanics of intelligent systems. To cut through the industry’s hype and identify what’s reliable, Newsweek has gathered a remarkable constellation of experts through its AI Impact interview series.
The urgency driving these conversations extends beyond the tech titans racing to build machine consciousness. Executives across all industries currently confront a complex calculus. What AI can actually accomplish today remains murky—pattern recognition and language processing reveal stunning breakthroughs, yet in practice, the limitations are glaring. More uncertain is if today’s astounding capabilities will continue to advance at such a mind-boggling pace. How much better will it get? And most uncertain of all: When will the AI revolution that changes everything actually arrive—is it coming in the 2030s, which OpenAI‘s Sam Altman predicts will be “wildly different from any time that’s come before”? Or is it already here? How do you invest wisely in a technology evolving faster than anyone can track, where the wrong bet means competitive extinction, yet the right approach remains maddeningly unclear?
AI promises to revolutionize how businesses operate—from automating back-office functions to optimizing supply chains and analyzing vast troves of data for strategic insights. Companies that master AI integration could gain insurmountable competitive advantages, while those that don’t risk obsolescence. However, the RAND Corporation found that more than 80 percent of AI projects fail—twice the rate of failure for information technology projects without AI. The comfortable option of caution has vanished; in a fast-changing landscape, the future demands decisions today.
From these wide-ranging dialogues, six essential lessons emerge from prognosticators and practitioners who have spent decades building, studying and deploying complex systems in the real world.
AI chief for Facebook owner Meta Yann LeCun poses during the AI Action Summit in Saclay on February 6, 2025. AI chief for Facebook owner Meta Yann LeCun poses during the AI Action Summit in Saclay on February 6, 2025. JOEL SAGET/AFP via Getty
Lesson 1: Humans Must Be in Control
Of all the dreams about artificial intelligence, none seduces Silicon Valley luminaries more completely than the vision of a human-less future where machines operate without oversight. Altman believes the technology he’s building will very soon do “95 percent of what marketers use agencies, strategists and creative professionals for today—easily, nearly instantly and at almost no cost be handled by the AI…. Images, videos, campaign ideas? No problem.” In a private meeting with lawmakers, Altman warns that “upwards of 70 percent of jobs could be eliminated by AI.”
These aren’t idle speculations. Enormous sums of money have been marshaled in this quest for human-free automation, yet results often fall short of promises. Robotics pioneer Rodney Brooks, former head of the MIT Artificial Intelligence Lab and a founder of iRobot, knows from decades of building real-world applications from frontier technologies, that to be widely adopted, even the most clever tools must leave room for humans. “People only accept new technologies when they don’t lose their sense of control,” he says in Newsweek‘s AI Impact interview series.
Brooks illustrates this principle by pointing to hospital delivery robots designed to transport dirty dishes and linens. He says he often sees these potentially labor-saving machines “turned off and pushed to the side” because medical staff, rushing through corridors doing life-saving work, encounter the robots blocking their path with no way to tell them to get out of the way. So after a while, the machines end up disabled and shunted aside.
The irony is that even while touting an automated future, the limits of AI often mean that humans are very much in the loop. When Elon Musk showed off his humanoid robot Optimus at a press event in 2024, the robots were remote controlled by humans. Before Cruise suspended operations, its “driverless” vehicles required remote human assistance every four to five miles.
“A 17-year-old can learn to drive a car in about 20 hours, even less, sometimes, largely without causing any accident,” LeCun tells Newsweek. “We have millions of hours of training data of people driving cars around, and we still don’t have self-driving cars,” he says. “So that means, in terms of understanding the world, we’re missing something really, really big.”
Despite having invented much of the underlying technology behind today’s large language models, LeCun argues they are fundamentally insufficient for achieving the autonomous capabilities that drive much of Silicon Valley’s AI hype. “If the path that my colleagues and I are on at [Facebook AI Research] and NYU…if we can make this work within three to five years, we’ll have a much better paradigm for systems that are controllable in the sense that you can give them goals, and they will, you know, by construction, the only thing they can do is accomplish those goals.”
Rodney Brooks speaks with Connie Loizos at the TechCrunch Sessions: Robotics at Kresge Auditorium on July 17, 2017 in Cambridge, Massachusetts. TechCrunch Sessions: Robotics is a single-day event designed to facilitate in-depth conversation and networking… Rodney Brooks speaks with Connie Loizos at the TechCrunch Sessions: Robotics at Kresge Auditorium on July 17, 2017 in Cambridge, Massachusetts. TechCrunch Sessions: Robotics is a single-day event designed to facilitate in-depth conversation and networking with the technologists, researchers and students of the robotics community as well as the founders and investors and was attended by more than 700 people.
Paul Marotta/Getty for TechCrunch
Lesson 2: Augment, Don’t Automate
“Right now,” Stanford neuroscientist David Eagleman tells Newsweek of the most successful AI deployments, “it’s all about co-piloting.” For individuals, he explains, “we can synergize with it and speed things up enormously.” This partnership model consistently outperforms automation attempts across multiple industries. Even among AI’s biggest proponents, augmentation appears to be winning the day. Despite AI writing 30 percent of the company’s code, Microsoft CEO Satya Nadella continues hiring engineers to focus on distinctly human qualities like “bringing clarity” to ambiguous situations. Google CEO Sundar Pichai treats AI as “an accelerator” that can eliminate tedious tasks rather than replacing human workers entirely.
An operator replaces bottles of alcoholic spirits above “Toni”, an automated cocktail maker made by “Makr Shakr”, ahead of a press event to promote the “AI: More than Human” exhibition at the Barbican Centre on… An operator replaces bottles of alcoholic spirits above “Toni”, an automated cocktail maker made by “Makr Shakr”, ahead of a press event to promote the “AI: More than Human” exhibition at the Barbican Centre on August 07, 2019 in London, England. Six bartenders were asked to mix certain cocktails in a bid to beat the robotic system in a taste-test challenge. The “AI: More than Human” exhibition takes a look at creative and scientific developments in artificial intelligence, and explores the evolution of the relationship between humans and technology.
Leon Neal/Getty
LLMs excel at generating options but cannot determine which one’s better—a determination that requires a world model including human values, contextual understanding and experiential wisdom—a capability that no model currently possesses. The technology can produce impressive outputs, but it lacks grounding to assess their appropriateness, quality or real-world implications. The solution, Eagleman suggests, lies in designing “AI systems to check on other AI systems” and creating “translators to dumb things down for us so that we can understand what is going on.”
The economic evidence decisively supports this collaborative approach. A 2023 study by Stanford economist Erik Brynjolfsson has shown why augmentation works: AI assistance delivered 14 percent productivity increases to customer service workers and 34 percent improvement for novice workers when used as a support tool rather than replacement technology.
Klarna CEO Sebastian Siemiatkowski learned this after receiving widespread attention for declaring “AI can already do all of the jobs that we, as humans, do” while replacing 700 customer service contractors with AI systems in February of last year. But soon after, he discovered that Klarna customers were being handed off in one-third of cases to human agents when the AI couldn’t resolve complex issues. Within months, Siemiatkowski acknowledged the AI resulted in “lower quality” customer experiences, prompting a switch to an augmented approach. The company has hired humans again and now uses AI to handle routine queries while the human agents tackle the most complex customer cases.
In May, Siemiatkowski said that cutting labor costs had “been a too predominant evaluation factor” because “what you end up having is lower quality.” He added that “investing in the quality of the human support is the way of the future for us.”
Daron Acemoglu, an MIT economist who won the Nobel Prize in 2024, has spent decades studying technology’s impact on workers and economic growth. He now warns that Silicon Valley has been following “the wrong direction for AI. We’re using it too much for automation and not enough for providing expertise and information to workers.”
Lesson 3: Pick Tasks AI Is Good At
Large language models can write poetry, summarize research papers and generate code with startling fluency. What took trillions of tokens, billions of parameters, petabytes of data and acres of GPU servers to discover was that the written word contains far more predictable patterns than anyone expected. “It’s astonishing how well that generates language,” admits Brooks. “I don’t think most people 10 years ago could have believed that would work so well.” The fundamental surprise: “What LLMs have shown us is we can emulate language with that thoughtless part.”
That’s a reference to Nobel Prize-winning psychologist Daniel Kahneman‘s Thinking, Fast and Slow, which divides cognition into automatic System 1 responses and effortful System 2 deliberation. LLMs function like System 1 processors, excelling at language tasks like writing, editing and translation—but are likely to fail at things that require System 2 deliberation, like abstract reasoning, creative problem-solving, and adapting to novel situations. The key question becomes: What types of problems can be solved with System 1-like processing alone?
For decades, Kahneman and Gary Klein, a psychologist who researches naturalistic decision-making, had a running disagreement about whether human intuition could be trusted. Klein championed expert fast decision-makers like firefighters and nurses. Kahneman emphasized systematic biases making intuition unreliable. In 2009, their dispute produced surprising agreement in a joint paper they wrote, “Conditions for Intuitive Expertise.” Klein was partially vindicated: Practiced experts do develop reliable intuition skills, but only when two conditions are met: “an environment that is sufficiently regular as to be predictable” and “an opportunity to learn the regularities by prolonged practice and feedback.”
It’s easy to extend these requirements to AI: regular patterns plus large datasets. For example, in January 2025, the Mayo Clinic reported that a model it had built to analyze pathology slides to diagnose cancer was not performing as well as human doctors. Despite 1.2 million tissue samples from 490,000 cases, the model did not have enough examples of each of the thousands of possible disorders that pathologists identify. For rare conditions, “you’ll find 20 samples over 10 years,” one of the researchers told MIT Technology Review—insufficient for pattern recognition.
A Googler walks back to Google’s South Lake Union office in Seattle, Washington. A Googler walks back to Google’s South Lake Union office in Seattle, Washington. iStock/Getty
“Even defining what regular is is not trivial,” notes Dana-Farber Cancer Institute CEO Ben Ebert to Newsweek. He points to the 2017 release of an algorithm that its developers claimed could detect pneumonia “at a level exceeding practicing radiologists.” But there was a problem, Ebert explains, “if you took a chest X-ray from a different hospital [than where it as developed], it completely didn’t work.” The problem is that it had learned patterns specific to that hospital, not general disease patterns. “The thing with AI is that because of how it was trained, it doesn’t realize that there’s a systematic bias.”
While there’s no precise way to measure regularity or to ever say how much data is enough, the Kahneman-Klein framework provides some clear directional guidelines. For instance, the head of a law firm would be able to recognize that AI would be better applied on drafting contracts— standardized formats with plenty of training examples—but struggle coming up with novel legal arguments. AI succeeds where human expertise can develop and struggles where even experienced professionals must rely on intuition alone.
Lesson 4: Use AI to Generate Possibilities, Not Answers
Asked for his assessment of LLMs’ reliability, Brooks is blunt: “They’re bullsh****** until we can ground them in reality.” His colorful language captures a truth: LLMs excel at persuasive speech “without regard to the truth,” fitting American philosopher Harry Frankfurt’s definition of bullsh**.
When LLMs hallucinate, the consequences can be spectacular. In 2023, investors spotted an incorrect claim in a Google Bard promotional video about the James Webb Space Telescope, wiping $100 billion from the company’s market value in a single trading session. That same year, attorney Steven Schwartz discovered the danger of relying on ChatGPT when a federal judge spotted six fake court cases Schwartz cited which the AI had invented, earning the lawyer national notoriety and $5,000 in court fines.
This imprecision stems from fundamental LLM architecture rather than fixable bugs. Researchers at Apple published a study in June, “The Illusion of Thinking,” which found that advanced reasoning models “face complete accuracy collapse beyond certain complexities,” even when provided with explicit problem-solving instructions. Yet this weakness becomes a strength when marketing teams need to generate dozens of concepts instantly or strategic planners want to discover unconsidered possibilities—and even in precision-critical fields like medicine.
When Eli Van Allen, the chief of the division of population sciences at Dana-Farber Cancer Institute, was a medical resident, he and classmates watched House episodes over lunch, competing to beat Hugh Laurie at identifying conditions. The trick wasn’t coming up with obvious diagnoses, but rather recalling faint possibilities.
Van Allen sees similar value in AI’s diagnostic brainstorming, which surfaces possibilities human doctors might miss. Where physicians remember “diagnosis 965” but overlook “diagnosis 9652,” AI can “pull down all 10,000 possibilities instantly” and help clinicians ensure “that tree to be the right tree and not prune too many limbs early on.”
Teaching machines to stop hallucinating is fighting against the grain—the real trick is to teach humans how to harness AI’s creativity, transforming its most dangerous flaw into a valuable feature.
Dreams of humans being entirely replaced by tech aren’t yet realistic. Klarna needs humans for customer service. Sebastian Siemiatkowski attends the official launch of the Klarna Pop-Up on June 04, 2019 in London, England. Dreams of humans being entirely replaced by tech aren’t yet realistic. Klarna needs humans for customer service. Sebastian Siemiatkowski attends the official launch of the Klarna Pop-Up on June 04, 2019 in London, England. David M. Benett/Dave Benett/Getty for Klarna
Lesson 5: Solve Human Problems
Venture capitalists have poured billions into AI companies convinced that extraordinary technology will inevitably find extraordinary uses. Their build-first mentality has created a graveyard of startups that died searching for problems to solve, dazzling users with technical sophistication while leaving them wondering what they’re supposed to do with it.
Apple’s Genmoji lets users create custom emojis from prompts like “a taco riding a skateboard.” But user reactions that started as “incredibly fun, creative and a great way to add more expression” soon became “the magic wore off quickly.” Suno AI can produce blues tracks that sound like they came straight from a Delta juke joint—but why? Google’s NotebookLM generates podcasts from any text filled with realistic human-like vocal quirks—authentic pauses, laughter, casual banter—that impress. Yet Cornell law professor Michael C. Dorf discovered after feeding it his writing that the results “sounded like a conversation among people who read my columns, lacked legal training, were reasonably smart and got about half of what I was saying but didn’t really follow a number of key points.” Does anyone need their meeting presentations turned into podcasts?
This autonomous test car is controlled remotely by humans. Researcher Miao Wang poses with an iPhone equipped with a remote interface to drive the “Spirit of Berlin” at Berlin’s Templehof airport November 2, 2009. The… This autonomous test car is controlled remotely by humans. Researcher Miao Wang poses with an iPhone equipped with a remote interface to drive the “Spirit of Berlin” at Berlin’s Templehof airport November 2, 2009. The Spirit of Berlin is a project of the Artificial Intelligence Group, directed by Rojas, at Berlin’s Freie Universitaet.
JOHN MACDOUGALL/AFP via Getty
Tech giants may have resources to indulge viral novelties like OpenAI’s Studio Ghibli image generators, but few outside Silicon Valley do. “The gritty people who run the multitrillion-dollar logistics of the world are not going to be spending billions of dollars based on glitziness. They’re going to be based on return on investment,” Brooks says. Lasting value comes from “understanding who your customers are and where their pain points are and how you are uniquely qualified to fix one of those things for them.” Netflix‘s recommendation engine helps people find something to watch when they’re faced with an overwhelming number of choices. GitHub’s Copilot reduces the tedium of writing boilerplate functions by autocompleting repetitive code patterns. However remarkable AI’s capabilities may be, successful deployments start with clearly defined human problems—not the reverse.
Lesson 6: Embrace Creative Partnership
When the pandemic isolated legendary production designer Rick Carter from his usual creative collaborators—directors such as Steven Spielberg, James Cameron and J.J. Abrams—he made a startling discovery while experimenting with AI video tools like Midjourney. “I can prompt it and even make mistakes, and it comes back with things that…I’m just going to call it an adjunct to what I am thinking,'” Carter tells Newsweek in the AI Impact interview series. Carter discovered something crucial: AI works best as a creative conversation partner rather than a creative generator. “It starts to interface with how I’m seeing things, and it stimulates me to move further in that direction.”
BEVERLY HILLS, CA – JANUARY 31: Rick Carter arrives at the Art Directors Guild 20th Annual Excellence In Production Awards held at The Beverly Hilton Hotel on January 31, 2016 in Beverly Hills, California. (Photo… BEVERLY HILLS, CA – JANUARY 31: Rick Carter arrives at the Art Directors Guild 20th Annual Excellence In Production Awards held at The Beverly Hilton Hotel on January 31, 2016 in Beverly Hills, California. (Photo by /FilmMagic)
Michael Tran/FilmMagic
His experience echoes decades of collaborating on films like Jurassic Park, Avatar and Star Wars: The Rise of Skywalker. “You’re being prompted, as a production designer, by the director,” he explains. “And then there’s a dialogue. It’s back-and-forth.” Carter says of Spielberg: “Steven makes a point of not knowing what he’s going to do.” He once asked Spielberg why he did it this way and the director replied, “Well, if I know what I’m going to do, then it’s like having a job at Denny’s, and I’m just servicing an order.”
Academic research demonstrates how this back-and-forth creative dialogue amplifies human capabilities. German researchers Jennifer Haase and Sebastian Pokutta found that true “co-creativity” is “a fusion of human creativity with advanced AI capabilities, where both entities contribute significantly to a shared creative product.” In a study published in Nature, DeepMind and Oxford researchers found that AI-mathematician partnerships achieve “surprising results by leveraging the respective strengths” when AI serves as “a test bed for intuition”—quickly verifying which hunches about mathematical connections “may be worth pursuing and, if so, guidance as to how they may be related.” This collaborative approach led to breakthroughs including “one of the first connections between the algebraic and geometric structure of knots” and progress on a 40-year-old unsolved problem in representation theory. As Carter discovered, AI may lack its own creative heart, but it can amplify yours—if you engage it as a partner in dialogue rather than a generator of finished ideas.
Agriculture technology with 3d rendering robot assistant in indoor farm or glasshouse. Agriculture technology with 3d rendering robot assistant in indoor farm or glasshouse. iStock/Getty
Extending Human Judgment
When OpenAI chose the name ChatGPT, the acronym carried deeper significance. While officially “Generative Pre-trained Transformer,” the letters also evoke the concept of a “General Purpose Technology”—an economic term reserved for innovations capable of transforming entire civilizations. Writing, metalworking, electricity: These foundational advances reshaped the very structure of human society. Few doubt AI belongs in this pantheon, yet like those earlier revolutions, its ultimate applications remain tantalizingly unclear.
This uncertainty is not unprecedented. If you somehow managed to transport an electrical generator to the 1850s, few would have any idea what it is or what to do with it, even though at that point scientists had studied electricity for centuries. Electric lighting, motors, telecommunications—those were all still faint visions of a far distant future.
The gap between technological capability and practical deployment has always challenged human imagination, but the patterns emerging from our AI Impact conversations reveal the true promise of AI lies not in replacing human judgment but in extending it. Where automation dreams crash against real-world complexity, augmentation thrives by preserving what humans excel at while amplifying capabilities through machine partnership.
These principles—maintain human control, foster collaboration over replacement, target domains with sufficient regularity for learning, generate possibilities rather than answers, solve genuine human problems and encourage creative dialogue—cut through promotional fog to reveal a pragmatic yet transformative path forward, not toward a human-free utopia, but toward sophisticated new partnerships between mind and machine.
What can AI really do now? What can AI really do now? Illustration by Thomas Kuhlenbeck/Ikon Images
Correction 6/25/2025, 3:48 pm: Corrected Eli Van Allen’s title
Russia allegedly field-testing deadly next-gen AI drone powered by Nvidia Jetson Orin — Ukrainian military official says Shahed MS001 is a ‘digital predator’ that identifies targets on its own
Ukrainian Major General Vladyslav (Владислав Клочков) Klochkov says Russia is field-testing a deadly new drone that can use AI and thermal vision to think on its own, identifying targets without coordinates and bypassing most air defense systems. According to the senior military figure, inside you will find the Nvidia Jetson Orin, which has enabled the MS001 to become “an autonomous combat platform that sees, analyzes, decides, and strikes without external commands.”
Digital predator dynamically weighs targets
With the Jetson Orin as its brain, the upgraded MS001 drone doesn’t just follow prescribed coordinates, like some hyper-accurate doodle bug. It actually thinks. “It identifies targets, selects the highest-value one, adjusts its trajectory, and adapts to changes — even in the face of GPS jamming or target maneuvers,” says Klochkov. “This is not a loitering munition. It is a digital predator.”
Even worse, the MS001 is allegedly operating in coordinated drone groups, persisting in its maximum destructive purpose despite the best efforts of Ukraine’s electronic warfare and other anti-drone systems.
Frustrated with warfare tech development speeds
Klochkov signs off his post by informing his LinkedIn followers that “We are not only fighting Russia. We are fighting inertia.” What he appears to wish for is an acceleration of Ukraine’s own assault drone capabilities. The Major General seems particularly disappointed in the Ukrainian system of procurement rounds, slowing field-testing and deployment of improved responses to new Shahed drone generations.
Shahed drones are originally an Iranian design but have gained great notoriety due to their sustained use by the Russian army to attack Ukrainian targets. The MS001 is substantially upgraded in the ‘smarts’ department thanks to Western/allies technologies.
Klochkov says the MS001 is powered by the following key technologies:
Nvidia Jetson Orin — machine learning, video processing, object recognition
Thermal imager — operates at night and in low visibility
Nasir GPS with CRPA antenna — spoof-resistant navigation
FPGA chips — onboard adaptive logic
Radio modem — for telemetry and swarm communication
Cute AI dev board with deadly potential (Image credit: Nvidia)
Western tech sanctions are supposed to neuter this kind of military threat from nations like Russia and Iran. This news indicates that such trade barriers are leaky, at best, and probably not taken seriously enough.
Get Tom’s Hardware’s best news and in-depth reviews, straight to your inbox.
Not the first Russia-deployed drone discovered using Nvidia AI
This isn’t the first Russian drone system that is thought to have adopted Nvidia’s Jetson Orin as a key component.
A month ago, Ukraine’s Defense Express site said that a new “smart suicide attack unmanned aerial vehicle with artificial intelligence,” dubbed the V2U, was powered by Nvidia’s little AI computer.
While the Shahed MS001s use an Iranian design, the V2U looks like it is more reliant on Chinese tech, including the Chinese-made Leetop A603 carrier board.
Follow Tom’s Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.
On today’s show, Andy simulates the Packers 2025 season utilizing artificial intelligence. Find out the results on today’s all-new Pack-A-Day Podcast! #Packers #GreenBayPackers #ai To become a member of the Pack-A-Day Podcast, click here: https://www.youtube.com/channel/UCSGx5Pq0zA_7O726M3JEptA/join Don’t forget to subscribe!!! Twitter/BlueSky: @andyhermannfl If you’d like to support my channel, please donate to: PayPal: https://paypal.me/andyhermannfl Venmo: @Andrew_Herman Email: [email protected] Discord: https://t.co/iVVltoB2Hg
WHO Director-General’s remarks at the XVII BRICS Leaders’ Summit, session on Strengthening Multilateralism, Economic-Financial Affairs, and Artificial Intelligence – 6 July 2025
Excellencies, Heads of State, Heads of Government,
Heads of delegation,
Dear colleagues and friends,
Thank you, President Lula, and Brazil’s BRICS Presidency for your commitment to equity, solidarity, and multilateralism.
My intervention will focus on three key issues: challenges to multilateralism, cuts to Official Development Assistance, and the role of AI and other digital tools.
First, we are facing significant challenges to multilateralism.
However, there was good news at the World Health Assembly in May.
WHO’s Member States demonstrated their commitment to international solidarity through the adoption of the Pandemic Agreement. South Africa co-chaired the negotiations, and I would like to thank South Africa.
It is time to finalize the next steps.
We ask the BRICS to complete the annex on Pathogen Access and Benefit Sharing so that the Agreement is ready for ratification at next year’s World Health Assembly. Brazil is co-chairing the committee, and I thank Brazil for their leadership.
Second, are cuts to Official Development Assistance.
Compounding the chronic domestic underinvestment and aid dependency in developing countries, drastic cuts to foreign aid have disrupted health services, costing lives and pushing millions into poverty.
The recent Financing for Development conference in Sevilla made progress in key areas, particularly in addressing the debt trap that prevents vital investments in health and education.
Going forward, it is critical for countries to mobilize domestic resources and foster self-reliance to support primary healthcare as the foundation of universal health coverage.
Because health is not a cost to contain, it’s an investment in people and prosperity.
Third, is AI and other digital tools.
Planning for the future of health requires us to embrace a digital future, including the use of artificial intelligence. The future of health is digital.
AI has the potential to predict disease outbreaks, improve diagnosis, expand access, and enable local production.
AI can serve as a powerful tool for equity.
However, it is crucial to ensure that AI is used safely, ethically, and equitably.
We encourage governments, especially BRICS, to invest in AI and digital health, including governance and national digital public infrastructure, to modernize health systems while addressing ethical, safety, and equity issues.
WHO will be by your side every step of the way, providing guidance, norms, and standards.
Excellencies, only by working together through multilateralism can we build a healthier, safer, and fairer world for all.
