What will the Internet look like in the decades to come? Will it remain a global, open and interconnected network, or will political, technological and economic forces transform it into something very different?
These questions were at the heart of a conversation between Vint Cerf, Internet pioneer and Google evangelist, and Jerome Glenn, Co-founder and CEO of The Millennium Project, during World Futures Day on March 1, 2024.
Cerf’s extraordinary experience in designing the architecture of the Internet explored not only where the Internet is going, but also how society can prepare for a future increasingly shaped by artificial intelligence, quantum computing, connected devices, and new forms of digital communication.
Looking back at fifty years of Internet history, Cerf offered a fascinating perspective on where the network may be heading—and on the technological and social challenges that could shape its future.
From a network of computers to a network of everything
Cerf began by putting the outstanding technological transformation of recent decades into perspective. The Internet of today is almost unrecognizable compared with the network he and Bob Kahn (1974) began designing in the 1970s. For Cerf, the Internet itself should be understood as the underlying mechanism that moves packets of information from one place to another. He compareed it to a road system: the Internet defines the rules of the road, but it does not determine what kinds of vehicles will use those roads, what they will carry, or what applications will eventually emerge.
The World Wide Web, for example, created by Tim Berners-Lee in the 1990s, is not the Internet itself, but a set of technologies and protocols that operates over the Internet. HTTP, the HyperText Transfer Protocol Protocol, emerged around the beginning of the 1990s through the work of Tim Berners-Lee and his colleagues at CERN. The subsequent development of graphical browsers helped transform the Web into a much more accessible and visually rich medium.
Mosaic, developed by Marc Andreessen and Eric Bina at the National Center for Supercomputing Applications, changed the way people experienced the Internet. Instead of a predominantly text-oriented environment, the Web began to look more like a magazine, combining formatted text and images and, eventually, audio and video. People could look at the HTML behind a website, copy it, study how it worked, modify it and build something new. There was no established course for becoming a webmaster in those early days; developers often learned by examining what other people had created. Over time, standards for the Web evolved through the World Wide Web Consortium (W3C), just as Internet standards evolved through the Internet Engineering Task Force (IETF).
Cerf also recalled how quickly this innovation became a commercial phenomenon. When entrepreneur Jim Clark recognized the potential of Mosaic, he helped establish Netscape Communications, bringing Andreessen, Bina and others into the new company. Founded in 1994, Netscape went public in 1995, becoming one of the symbols of the early Internet boom.
For Cerf, this history illustrates one of the defining characteristics of the Internet: its creators established an open architecture without knowing what applications would eventually emerge on top of it. The Web was one of the most consequential examples—and it transformed the Internet from a technological infrastructure into a mass communication and commercial platform. By 2024, an estimated 5.5 billion people were online, while another 2.6 billion remained offline. The amount of data that networks can carry has also increased by many orders of magnitude compared with the early Internet, from the tens of kilobits per second of early network backbones to hundreds of gigabits per second on individual high-capacity optical channels.
Cerf pointed to an emerging world in which almost any object could potentially become an Internet-connected device, from smartphones and sensors to appliances, vehicles and medical equipment. Wi-Fi and wearable devices are extending connectivity into everyday environments, while the Internet of Things is bringing more and more physical objects into the network. The Internet is therefore becoming not simply a network of people and computers, but a network of things, environments and increasingly biological and medical systems.
One of the most striking ideas emerging from Cerf’s discussion is that software may become one of humanity’s most inexhaustible resources. He pointed to countries investing heavily in digital technologies and suggested that, unlike physical resources such as oil, software is limited primarily by our ability to think how to imagine and build it. This opens an enormous field of possibilities. Healthcare is already becoming increasingly digital, with hospitals using sensors and connected systems to monitor patients and detect unexpected changes. Similar developments are taking place across many sectors. The Internet’s application space, Cerf argued, will continue to evolve. And the unpredictability of its evolution is built into the original design of the network.
Artificial intelligence: extraordinary opportunities, extraordinary risks
No discussion about the future of the Internet can ignore artificial intelligence. Cerf said that he continues to be both excited and worried about artificial intelligence and machine learning. He also stressed that the rapid development of AI makes it important to understand much more deeply how these systems work and how they can go wrong. In his view, sharing knowledge about their capabilities and limitations will be essential as societies begin to develop appropriate forms of AI governance.
He sees enormous potential in machine learning. One example is the use of artificial intelligence to understand how proteins fold. Systems capable of processing hundreds of millions of proteins create knowledge on a scale that would be impossible through manual research. Another example comes from Google’s data centers, where machine learning has been used to optimize cooling systems and significantly reduce energy consumption. Large language models (LLMs) introduce another dimension. They can translate languages, recognize and generate speech, and produce text, images and sound.
But their limitations are equally important. Cerf illustrated them with a personal experiment: he asked a Large language model (LLM) to write his own obituary. The system produced a convincing-looking text, but it attributed achievements to him that belonged to other people and even invented family members. The problem, Cerf explained, lies partly in how these systems are trained. LLMs learn statistical relationships among enormous quantities of text. They can generate highly plausible answers without necessarily understanding the context in the human sense. This is one source of what are commonly called AI “hallucinations.” Moreover, LLMs do not always perform arithmetic reliably. One possible solution is to connect them through APIs to specialized programs capable of carrying out tasks such as arithmetic and algebraic computation. Although recent advancements in reasoning models have mitigated these errors, Cerf’s warning remains relevant today.
Yet Cerf did not dismiss the technology. On the contrary, he argued that LLMs are more than what some people call “statistical parrots.” He believes there is something deeper happening within the statistical models built from the enormous amounts of data they ingest, including a nascent ability to perform some kinds of reasoning. Cerf also emphasized that the growing power of artificial intelligence creates an equally important question about governance. He noted that companies and other organizations are working to understand how appropriate rules for artificial intelligence might be developed, but he also stressed that the field is still evolving and that there are no simple answers. In his view, society needs to understand much more deeply how these systems work and how they behave, while sharing knowledge about their capabilities and limitations.
The challenge is not simply that artificial intelligence can produce harmful or misleading outputs. Powerful digital tools can be used constructively or abusively. Cerf compared this problem to image-editing technologies such as Photoshop: the technology itself is not necessarily the problem; the problem arises when it is deliberately used to misrepresent reality. This makes questions of authenticity, provenance and trust increasingly important. How can we establish where a digital object came from? How can we determine whether it has been altered?
Cerf also argued that understanding AI’s risks requires more than technical expertise. Technology companies may need to work more closely with anthropologists, sociologists, and neuroscientists to anticipate the social and economic consequences of technologies that can be used in unexpected ways. The challenge may therefore be a failure of imagination: we need to consider not only what technology can do, but also how people and societies may use it.
This also raises broader questions about privacy, anonymity and accountability. Cerf suggested that accountability systems may become increasingly important, without necessarily eliminating online anonymity. Instead, mechanisms like an Internet license could allow serious abuse to be traced and responsibility established when necessary. Strong authentication may also become more important as digital systems are increasingly used for financial transactions, public communication, and actions carried out on behalf of others.
Cerf’s concern is therefore not simply that AI can generate false information, but that increasingly powerful digital tools can make it harder to establish what is authentic and where a piece of information came from. Digital signatures, cryptographic hashing and other forms of provenance could become increasingly important in maintaining trust in a digital environment.
A fragmented Internet?
Technology is not the only force shaping the future of the Internet. Politics may be equally important with “data.” Cerf is particularly concerned about the Internet fragmentation as a consequence of “data sovereignty.” Governments have legitimate reasons to regulate data, protect privacy and establish rules for the use of personal information, such as the European Union’s GDPR. But, in Cerf’s view, regulation can become counterproductive if it undermines the fundamental principle on which the Internet was built: connectivity, the ability of one party somewhere on the network to communicate with another party somewhere else. The Internet’s architecture was deliberately designed without assuming that governments, borders or specific applications would determine who could communicate with whom. Cerf therefore warns against policies that could create national or regional “pieces” of the Internet separated by political boundaries.
To address security and privacy problems, the network needs stronger mechanisms for protecting communications, including encryption and technologies designed to make connections more secure and resilient. One example is the widespread adoption of HTTPS, which adds encryption through TLS to Web communications. Cerf also pointed to QUIC, a newer transport protocol originally developed at Google and subsequently standardized by the IETF. QUIC incorporates cryptographic security and can provide features such as connection migration and multiple concurrent streams, helping networks recover more effectively from changing conditions and avoiding some of the limitations of traditional TCP connections.
Connectivity, in Cerf’s words, is fundamental. It drives commerce, scientific research, the Web and the discovery and exchange of knowledge. Countries should therefore resist policies that impose a kind of “Westphalian” model on the Internet, treating it as a collection of nationally bounded networks. The Internet was designed precisely to allow communication across such boundaries. Its continued value depends on maintaining that flexibility while addressing legitimate concerns about security, privacy and data protection.
The Internet leaves Earth
Perhaps the most futuristic aspect of Cerf’s vision is that the Internet may eventually extend beyond our planet. Lower-orbit satellite communications like Starlink are expanding connectivity, while efforts are also underway to increase the resilience of terrestrial networks, including subsea optical-fiber systems. Greater redundancy and switching capabilities could allow traffic to be rerouted when a cable is damaged or cut, making the global network more resilient.
Work on the architecture of an Interplanetary Internet has been ongoing since the late 1990s, and experimental networking demonstrations have already taken place around Mars since 2004 and on the International Space Station since 2010. As humanity returns to the Moon, these technologies could become increasingly important. NASA’s lunar networking efforts and ESA’s Moonlight programme are developing communications and navigation infrastructure to support a sustained human and robotic presence around the Moon. As we eventually send people farther into space, an Internet designed for interplanetary communication could become essential infrastructure. This challenge is enormous because we cannot go faster than the speed of light, which feels slow on an astronomical scale: it takes eight minutes for light to get from the Sun to us, and it can take up to twenty minutes to get a message to Mars. An interplanetary Internet therefore cannot simply reproduce the terrestrial Internet; it needs to tolerate long delays, intermittent connections, and extreme conditions.
But Cerf’s vision extends beyond establishing communication between nearby planets. With sufficient computing capability and reliable power sources, humanity could eventually send robotic systems much farther into deep space—systems capable of operating autonomously and potentially returning information about what they have discovered. The challenge therefore is not simply to build better communication networks. It is to build machines and infrastructure capable of surviving for extremely long periods of time without human intervention. That means highly resilient designs, extensive redundancy, and energy sources capable of operating far from conventional infrastructure. In deep space, where solar power may become increasingly difficult to rely on, nuclear power could play an important role in sustaining long-duration missions.
The idea of sending robotic explorers to other stars is already being considered. Cerf referred to work on a spacecraft concept designed to reach Alpha Centauri within roughly a century. With current propulsion systems, such a journey would take tens of thousands of years. Reaching another star within a humanly meaningful timeframe would therefore require propulsion systems capable of reaching a significant fraction of the speed of light—perhaps 10 or 20 percent. The technological challenges are immense. But the underlying idea is clear: the future Internet could become part of a much larger infrastructure for autonomous exploration, extending human knowledge far beyond Earth.
Could there be a quantum Internet?
The quantum Internet would not simply be a faster version of today’s Internet. It would be a fundamentally different kind of network, designed to connect quantum systems and distribute quantum states across physical distance. Google, IBM, and others are building quantum computers that use entanglement among qubits to perform concurrent calculations. However, quantum states are fragile and dissipate when energy is injected to manipulate gates. To keep computations stable, 100 to 1,000 physical qubits must represent a single logical qubit. Because building a massive single machine with millions of qubits is extremely difficult, the approach is to connect smaller quantum machines using optical fiber networks. By keeping one entangled photon at the source and sending the other through a fiber to a receiving machine, the two systems link together. This requires specialized quantum repeaters to move photons without losing their delicate entanglement.
Only a limited number of quantum algorithms currently exist. Early applications may act like 1950s analog computers—modeling nonlinear processes to get low-resolution baseline answers (perhaps 2 or 3 bits of accuracy), which conventional computers then refine. Yet, much like the historic introduction of VisiCalc in 1979—which replaced the manual, week-long paper calculations Cerf did for the Apollo program—quantum computers will allow us to solve ultra-hard optimization problems in real time, exploring solution spaces impossible for today’s supercomputers. Cerf’s broader point is that the importance of quantum computing may ultimately lie not simply in doing existing calculations faster, but in making entirely new classes of problems computationally tractable.
Quantum computing also introduces a significant risk to current cryptographic systems. A sufficiently powerful, fault-tolerant quantum computer running Shor’s algorithm could break widely used public-key systems such as RSA by efficiently factoring the large numbers on which their security depends. This is why post-quantum cryptography is already becoming a priority: new cryptographic methods are being developed to remain secure even in a future in which large-scale quantum computers become available. The transition is particularly important because encrypted information intercepted today could potentially be stored and decrypted in the future.
Could there be life beyond Earth?
The expansion of computational and robotic capabilities also raises another question: could we discover life elsewhere in the universe? We have now identified a very significant number of exoplanets, demonstrating that it is normal for stars to have planets. Some may even have conditions that could be compatible with life as we know it.
But this leads to a deeper question: what about life as we do not know it? What if another life form were not based on carbon or did not share the biological characteristics we associate with life on Earth? If we encountered such a system, how would we recognize it as life?
The same problem becomes even more difficult when we consider intelligent life, Cerf said. How would we know that another intelligent life form was present if its intelligence operated according to principles fundamentally different from our own?
These questions lie at the heart of the search for extraterrestrial intelligence, or SETI. The challenge is not simply to search for another civilization, but to determine what evidence of intelligence might look like when we do not yet know what kind of intelligence we are looking for.
Preparing for a future we cannot predict
Cerf expressed his conviction that, “with sufficient computing power, we might indeed end up with a kind of artificial life.” Computers could potentially model and emulate some of the ways in which real biological systems function simply by providing them with artificial inputs. Beyond single cells, sufficient computing capability allows for the creation of digital twins of cities, biological systems, and other complex systems, to model their behavior, test scenarios and explore interactions that would be difficult to study directly in the real world, —as we are already starting to do today.
The conversation between Cerf and Glenn also highlighted a fundamental question of futures research: how can we prepare for a future that is constantly being reshaped by technology?
Cerf emphasized that technological transformation is not a new phenomenon. He recalled the extraordinary changes that took place between the end of the nineteenth century and the beginning of the twentieth: radioactivity, wireless communication, aviation, Einstein’s revolution in physics, radio broadcasting and many other breakthroughs transformed society within a remarkably short period. We are living through a similar period today. Artificial intelligence, biotechnology, quantum computing, digital medicine and space technologies are reshaping the boundaries of what is possible. The challenge is not to predict one single future, but to prepare for multiple possible futures.
Cerf’s perspective is ultimately optimistic. Humanity has repeatedly adapted to dramatic technological and social changes. But his optimism is not unconditional. When asked about the possibility of surviving major challenges such as climate change, Cerf gave a strikingly candid answer: he said he was not sure that humanity would survive it. Climate change, technological risks and political fragmentation are real threats, and technological progress alone does not guarantee that humanity will successfully navigate them.
The future of the Internet will therefore depend not only on faster networks, smarter machines or quantum technologies. It will depend on the choices societies make about openness, connectivity, security, privacy, and cooperation. The Internet was designed as a flexible system whose creators could not know what applications would eventually emerge. More than fifty years later, that flexibility remains its greatest strength.
And perhaps the most important lesson from the conversation between Jerome Glenn and Vint Cerf during World Futures Day 2024 is precisely this: the future of the Internet cannot be predicted with certainty, but it can be shaped by the principles we choose to preserve. As the Internet expands from Earth to space, from computers to billions of connected devices, and from human communication to quantum systems, and perhaps eventually to autonomous robotic exploration, artificial forms of life and new forms of intelligence, the question will not simply be how far technology can go. It will be whether we can ensure that the networks we build continue to connect us rather than divide us.
This web article was supported by the Prince Mohammad Bin Fahd Foundation Research Grants for Humanitarian Development 2025 under the project titled “World Futures Day as a Cross-Cultural Participatory Foresight method for Addressing Global Challenges”.




is there a video?