Godfrey Nyoni
FOR most of its history, the internet has been designed around one central assumption: a human is using it.
A person opens a browser, searches for something, clicks a website, reads an article, buys something, sends a message.
But artificial intelligence (AI) is changing that assumption.
AI systems can increasingly search for information, interact with websites, use APIs, compare products and perform tasks on behalf of people.
What happens when the internet is no longer used primarily by humans, but increasingly by software agents acting on their behalf?
There will still be websites, applications, social networks and online services.
But behind the scenes, an increasing amount of internet activity could come from machines communicating with other machines, creating what we might call the AI Web and fundamentally changing how the internet works.
The traditional web was designed around human interaction — a website displays information, a human reads it, clicks a button and completes a transaction.
AI introduces another possibility.
Instead of a human going through a search engine to reach a website, we could increasingly have a human providing an objective to an AI agent, which searches for information, evaluates options and interacts with online systems without the user ever visiting the underlying websites directly.
Instead of asking an AI what the best accounting software options are, a user could eventually say: find an accounting system suitable for my business, compare the options and prepare what I need to choose.
The AI searches websites, compares information, interacts with APIs and prepares a recommendation.
The human provides the goal and the AI performs the digital work.
Today, websites are primarily designed for humans with navigation menus, images, buttons and forms that people understand visually.
AI systems may not need the same interface.
Instead of an AI agent opening a hotel website and reading pages manually, the website could expose structured information availability, price, location, features, booking options for the AI to process directly.
Websites may increasingly need to serve two audiences: humans and machines.
An AI agent could receive a task from a person, search multiple services, collect information, compare results, complete digital actions and return the outcome to the user.
The human may only see the beginning and the end.
The majority of the activity happens between machines, and this creates a major challenge for website owners.
An AI agent could take a purchase request, search product databases, make comparisons and complete a transaction without the user ever visiting the retailer’s website.
A business could have excellent products and receive fewer direct visitors because AI systems are becoming the gateway between customers and websites, changing how businesses think about digital marketing entirely.
For decades, the internet economy has been heavily influenced by clicks businesses competing for search rankings and website traffic.
But AI could reduce the clicks required to obtain information. If an AI answers a question directly, the user may not visit multiple websites.
That is convenient for the user, but creates a difficult question for publishers: if people stop visiting websites, how do websites make money?
Many depend on advertising, subscriptions or traffic-based models.
If AI systems consume information without sending users to the original source, those models could be disrupted.
There is also an attribution problem. Suppose a website publishes an original article and an AI system reads it and provides a summary to a user, does the original publisher receive a visit, revenue, attribution or nothing?
The AI web will need mechanisms ensuring that creators benefit when their information contributes to AI-generated answers, otherwise the internet faces an uncomfortable paradox: the better the original information, the more valuable it becomes to AI, while the publisher receives less direct traffic.
If millions of AI agents begin operating across the internet, a completely new concept emerges: the machine customer.
Some customers may arrive through software agents that evaluate price, availability, quality and delivery terms without caring about beautiful animations or marketing slogans.
Businesses may need to develop two parallel digital strategies: one for human experience and another for machine experience.
The future website may need to communicate with both audiences through structured data, well-designed APIs (Application Programming Interfaces), machine-readable content and reliable authentication.
Businesses may increasingly need to ask: can AI systems understand and interact with our website correctly?
There is a serious problem with this future.
AI systems can make mistakes and an AI agent searching the internet on behalf of a person needs to determine which information is trustworthy, particularly important for financial, medical, government and legal information.
The AI web could also create new cybersecurity risks, as attackers could create malicious services designed to manipulate automated systems.
The internet will need stronger mechanisms for verifying identity, authenticity and permissions.
AI agents will also need digital identity and clearly defined permissions an agent authorised to read information should not automatically have permission to make purchases and automation should not remove meaningful human oversight.
Despite all these changes, humans are unlikely to stop using the internet.
Human curiosity, the desire to watch videos, read stories, explore, create and connect is difficult to automate.
Therefore, the AI web will exist alongside the human web rather than replacing it. Machines will simply become a much larger part of the audience.
This transformation will affect Zimbabwe as well.
A Zimbabwean business could eventually have its information accessed by human customers, search engines, AI assistants, automated procurement systems and financial platforms, making structured and accurate digital information an increasingly important asset.
There is also a significant opportunity.
A large amount of African information for local businesses, services, history, agricultural knowledge and research remains poorly represented online.
African organisations and creators have an opportunity to contribute high-quality local knowledge to the digital ecosystem.
The AI web should not become a system where machines understand the rest of the world but struggle to understand Africa.
For decades, the internet was built around one central user: the human.
That is changing.
People may still browse websites, but increasingly AI agents could browse on their behalf.
Businesses may still build websites, but those websites may need to communicate with machines as effectively as they communicate with humans.
For Zimbabwe and Africa, the businesses, developers, governments and institutions that understand this transition early can begin building digital services that are not only human-friendly but also machine-readable, secure and AI-ready.
The biggest question may no longer be whether humans will stop using the internet. When humans and AI share the internet, who will the web be designed for?
Godfrey Nyoni is a technical consultant at Pique Squid. For feedback, contact www.piquesquid.com/+263786526527
