Can AI Search destroy the internet as we know it?

Okay, maybe "destroy" is a bit extreme, but how we find answers online is changing dramatically, and the consequences are not minor.

Remember how Google revolutionised everything with its algorithm and ad model two decades ago? The entire ecosystem of websites and digital advertising we know today was profoundly impacted and defined by Google Search and sponsored links.

This made Google one of the most valuable companies on the planet (US$1.7 trillion in mid-February 2024) and allowed it to dominate almost a third of the total ad spend globally.

New AI search engines are putting this immense value at risk and promising to completely upend the model that has sustained many corners of the internet since the beginning of this century.

These new tools aren't search engines; they're answer machines. They scour the internet at light speed, finding relevant content to synthesise into human-like answers. Their conversational style is both easier and more engaging than sifting through a list of links.

Microsoft's Bing dipped its toe in the water when it launched its ChatGPT-supported version in 2023, but the company was extremely cautious, and Bing ultimately didn't threaten its own search page or challenge Google Search - at least until now.

But the continued evolution of ChatGPT and other players, like Perplexity.AI, has forced Google to enter the game with its very sophisticated Gemini Advance this year after an initial move with Bard.

Google was able to respond with such a robust solution because it has led AI research for years - one of its papers is considered critical for some of the recent leaps in Large Language Models.

The fact that the company only responded to external pressure clearly indicates that it knows how disruptive this technology is to the current search engine model.

The problem for Google Search and many players of the internet ecosystem is that the new AI answer machines use internet content to respond to any question without forcing you to look elsewhere. If you're satisfied with the answers, you can move on and never visit another webpage again.

Of course, we won't trust these new AIs entirely and will still try to confirm information with other sources. But the reality is that even for those who need to get it right and are sceptical (like me), the volume of searches and clicks will decrease dramatically. Apart from subscriptions, nobody knows how to monetise the search.

On an individual level, the initial upside is clear: you no longer have to hunt for the right website and information, saving time and energy (a subscription of US$20 or US$30 a month seems low to get answers for anything you can think of).

But the potential downside is systemic. With traffic highly concentrated on one or a few AI search platforms, many companies and content creators of all sizes and types will have fewer financial, commercial, or personal incentives to continue producing original high-quality content. Most content may start to be produced by AI as a way of reducing costs.

The colourful, chaotic, and almost infinite (even if highly imperfect) internet we have today could become smaller, more homogenous, concentrated and synthetic.

Some new players, like Perplexity AI, are trying to ride this wave without turning everything upside down. They share all the sources of information and links so you can dive deeper and beyond the AI answer. They also would love to make money from advertising.

But if my experience is similar to most, the tendency is to significantly decrease the number of times you visit other websites or make traditional searches.

If this becomes the new norm (and there are good reasons to believe it will), besides less valuable human-produced information and more content creators struggling, we will see the power of these new tools and the companies behind them dwarf anything Google or Meta have achieved in the past couple of decades.

There are no simple ways to guarantee we reap the benefits of these incredible technologies and avoid their potential for destruction.

But if we fail to understand what is happening and wait for the results, we will find ourselves with problems orders of magnitude larger than the trust and fake news crises we face today.

Let me know what you think in the comments.