Rise of the ‘answer engine’: AI’s impact triggers fresh questions for internet ad models, power usage and data sharing – News

A lengthy stack of issues and macro trends is shaping the technology industry today, and high on the list is the prospect that the internet engine powering an estimated $16 trillion to $21 trillion global digital economy is about to undergo an enormous transformation.

“The business model of the internet is about to change radically, that’s just a given,” Matthew Prince, co-founder and chief executive of Cloudflare Inc., said during an appearance at EmTech MIT this week, hosted by MIT Technology Review. “The question is, ‘What is it going to change to?’”

Falling traffic and AI scraping

Prince’s question is based on recent data which shows a precipitous drop in referrals and visits over the past year for a broad range of websites, ranging from travel and media to e-commerce and finance. Traffic to CNN’s webpage fell 30% since mid-2024 and HuffPost saw a 40% decline in the same period.

Prince’s explanation for the falling page views is that in the burgeoning AI era, the rules of the game have changed. Google LLC’s introduction of “AI Overviews” for nearly every search has curtailed click-throughs back to content creators.

“Instead of a search engine, it’s an answer engine,” Prince said during the event, held on the campus of MIT in Cambridge, Massachusetts. “That means there isn’t going to be traffic back to those original sources. There isn’t any way for the original content provider to get traffic to sell you something.”

This problem has been further complicated by another new reality: Web searches are increasingly being done by AI bots that are simply scraping information. In July, Cloudflare announced that it would automatically block AI crawlers from scraping the websites under its control, and Amazon.com Inc. made news this week when it sent a “cease and desist” letter to AI startup Perplexity, barring its agents from making purchases on the retail giant’s network.

Ad market for AI providers

There is also the prospect that leading AI providers may soon enter the ad market, enticing leading brands to shift marketing dollars. The signals are becoming louder that OpenAI Group PBC is putting the infrastructure in place to begin selling ads, a revenue stream that the firm’s internal documents have projected could grow to $25 billion annually by 2029.

OpenAI hired an executive in May who previously launched advertising for Facebook’s news feed. Recent interviews with the company’s leadership have included remarks that indicated the company was keeping its options open. There is no similar evidence that Anthropic PBC may be considering soliciting ads for its enterprise-focused platform, but the prospect of a major new revenue stream may be too tempting to resist.

“All of these companies are going to throw huge amounts of money at OpenAI and Anthropic,” said Brian O’Kelley, co-founder and CEO of Scope3. “There’s a lot of companies now looking at how ChatGPT chooses what to share when you ask it a question.”

As the internet undergoes a sea change in its financial model, some leaders in the tech world are taking a harder look at the data engine that has fueled online commerce. Project Liberty, a $500 million initiative backed by U.S. entrepreneur Frank McCourt, has a mission to reimagine web infrastructure in the AI age.

Frank McCourt, founder of Project Liberty and chairman of McCourt Global, spoke about the need for internet users to control their own data.

The group has built an application-agnostic protocol with over 13 million users to provide shared digital rails for an open, interoperable web. The goal, according to McCourt, is to establish common standards for identity and connection that will allow individuals to control the use of their own data.

“These platforms know everything essentially about us, so we can be targeted,” said McCourt. “It’s become this highly manipulated technology. If I can target you with an ad for something I know you want to buy, I can target you with a piece of news I know you will want to digest.”

McCourt’s solution is to avoid platform lock-in and create an architecture that will enable users to manage interaction with AI agents. This will involve fixing technology to install permissioned use of data, developing a policy framework that will support this, and enabling users to benefit from economic value created online.

“Imagine a different paradigm where it’s a data-sharing economy,” McCourt said. “GDP is going up, but are people flourishing? We should be flourishing and in charge of the technology, not slaves to it.”

Sovereign AI fuels datacenter expansion

The struggle for data control is also spilling into the geopolitical arena. Regional AI infrastructure is becoming essential, according to some industry experts, as national governments actively invest in programs and data centers to fuel sovereign AI strategies.

Countries from Southeast Asia to the Middle East are fast-tracking data center buildouts to become self-sufficient in data sovereignty. However, this attempt to keep data within borders may ultimately prove to be a misguided strategy, according to R. David Edelman, former White House adviser on technology policy and director of the MIT Project on Technology.

“There is a little bit of magical thinking happening in governments right now,” Edelman said. “It’s an infrastructure investment plan in national security clothing.”

Data center expansion to support AI is also having an impact on local politics within the U.S. In an appearance during the MIT conference, Carnegie Mellon University’s Costa Samaras noted that two members of the Georgia Public Service Commission lost their seats on Tuesday in what was widely viewed as a referendum on rising energy costs.

Georgia Power had been facing public scrutiny over plans to support data center expansion that would consume 90% of new power over the next six years. A lawsuit recently filed in Kansas City put the brakes on a proposed $12 billion data center. A proposed data center project in Tucson, scheduled to be built over 290 acres, has also run into community opposition over projected water and energy costs.

Samaras believes that the solution lies in advanced nuclear fusion, exploration of new geothermal energy sources, and a national commitment to build a smarter grid. “We are still in innovation infancy around climate and energy technologies,” Samaras said. “If we don’t lean into innovation, we might not make it.”

AI’s rapid adoption has raised a host of questions that the tech industry is struggling to answer. Yet the current path has made clear that change is inevitable and the impact can be either positive if managed properly or quite damaging if not. The outcome may ultimately rely on the willingness of technologists to tackle the hard problems that spawned the success of both the internet and AI today.

“Technology can be designed to operate differently,” McCourt said. “We just have to fix it before we make it more powerful. Changing entrenched technology is very, very hard. This is our moment, folks, to fix how the internet works and to fix how AI works.”

Image: News/Microsoft Designer