This analysis was first published in SvD Näringsliv, in Swedish, on December 6th, 2022. This piece was translated from Swedish by Claude. Some phrasing may differ from a human translation.

A research organisation is behind the biggest thing the tech world has talked about in years: ChatGPT. The new AI software is taking the internet by storm. Could this be the foundation of the next Google?

Ask and you shall receive, as the saying goes. That has rarely been more literally true than with ChatGPT. In just one week, the new AI software has gained over a million users.

Part of the innovation lies in the interface: you can chat directly with the software, asking questions in natural language. The responses also feel more human than the correct-but-mechanical answers we are used to from search engines. It feels like talking to someone who is genuinely trying to help — and who can refine their answers based on your feedback.

Ask it directly what to make for a Swedish Christmas dinner and it responds in an instant, with a confident, reasonable list. The impression is striking.

But the most interesting thing about this phenomenon is not the novelty of the user experience. It is what the underlying technology could form the basis of — and what comes next.

For a company like Google, which built a world-leading position by being the best at finding things on the internet, services like ChatGPT represent a new and unusual challenge. Google has been essentially uncontested in search for a long time. But as so often with new technology, the challenge is coming from the side — in this case from a partly non-profit research organisation, OpenAI — rather than from a direct competitor.

The challenge for Google is not the technology itself. Google employs thousands of people working on machine learning. The issue is scale and the company’s revenue model.

Google handles around 8.5 billion searches every day. Getting a service with a million users to run quickly and reliably is a challenge. Getting Google to work for everyone on Earth is a fundamentally different and vastly harder problem.

But the deeper issue is financial. Google generated nearly 6 trillion kronor in revenue over the past twelve months. The vast majority came from ads linked to search results. Changing how search results — and the accompanying ads — are presented therefore carries enormous risk. The smallest change could result in billions in lost revenue.

Precision is what Google charges for. That dependence points toward incremental changes rather than radical redesigns. The question is not whether Google could technically overhaul its search engine. The question is whether anyone inside Google can justify that kind of risk. Large companies, as Google has now become, tend to favour the safe over the bold in moments like this.

Some healthy scepticism about ChatGPT is also warranted. Stack Overflow — a question-and-answer site for software developers — has temporarily banned answers generated by ChatGPT. The responses are simply wrong too often. And what makes it worse is that the wrong answers tend to look plausible and well-sourced. That undermines the entire point of the site: being able to find a correct answer quickly.

It is early days. But the direction is clear: AI-powered conversational interfaces are coming for search. Whether it is OpenAI, Google, or someone else that ends up defining what that looks like is still very much an open question.