Google Has Changed

I have been amused by many of the reports of Google Gemini’s peculiar behavior (for excellent examples see here, here, here, here, and read Jeff’s piece), but, in my view, the most interesting thing about this episode is what it has shown about where Google now is as a company. All told, Google Gemini has yielded precisely the opposite reaction that Google’s famous search engine did upon its release. In 1998, those of us who were using AltaVista or Ask Jeeves or whatever found Google and said, “wow, that’s incredible!” This week, everyone who has had some experience with other AI products has found Gemini and said, “what total garbage!” (Or worse.)

This doesn’t seem to be an accident. Google has more employees now than it’s ever had, but, despite this — or perhaps because of it? — its products are more mediocre than they have ever been. In this, Google reminds me of the Windows Vista-era Microsoft Corporation, which, having inextricably insinuated itself into every corporate office and government department in the world, seemed content to sit back and print money. When, I wonder, was the last time that Google did anything great? Its search engine was released 26 years ago. Gmail — which was once better than the alternatives, but is now not — was released in 2004. Chrome was released in 2008. Since around then, the company has become sclerotic — content to sit on its laurels.

Given its market position, Google can probably do this for quite a long time. And, in that time, it will be able to ship out awful — even evil — products, which will be adopted by tens of millions of users (and schools) out of sheer inertia. Currently, Google is integrated into all manner of online things in a way that raises the cost of entry for any competitors. Its email service is the default option on most clients; its authentication is used by thousands of sites and APIs; its Docs and Sheets products are effectively the default for online collaboration; its search engine tends to be the first option in one’s browser. Once Gemini is fully bundled in, it will have an in-built audience — even if, relative to other AI products, it’s bad. Given the nature of America’s economy, and the likelihood that AI will be key to the future of tech, Google will not be able to get away with this forever. But can I see it happening for ten, 15, or 20 years? Yes, I can.

Which is a problem, because, in the medium term, LLM AI systems will likely replace search engines as the primary tool that we use to look up information, and, with its 85 percent share of that market, Google will be able to transition its customers from Search to Gemini with an ease that everyone else will envy. You may have noticed that Google Search has become less useful recently. That, in part, is because SEO experts have worked out how to manipulate it to their ends. Have you ever wondered why every recipe for spaghetti carbonara begins with, “My grandfather was born in Napoli in 1923 . . .” and then continues with a five-paragraph story before one can find mention of a single egg? That’s SEO. In theory, AI can get around that by being directly responsive to your queries. You ask it a question, and it tells you the answer without any of the detritus that you get elsewhere. But, of course, for that to be useful, AI has to be both good and uncorrupted. If it’s bad and corrupted, as Google Gemini is, it’s not only useless; it’s dangerous.

There is a reason that many of the same progressives who are always trying to control the news media have started calling for the regulation of AI, and that reason is that they understand this dynamic all too well. With a search engine — even one whose results have been biased — users can at least choose which link to click. With AI tools, one is given only one answer per request. If those answers can be, er, “shaped” — be that by government regulations or by the blinkered ideologues at Gemini — then those who have done the “shaping” will have increased their political power enormously. Yesterday, the programmer and investor Marc Andreessen suggested that this was the plan. “Big Tech AI generates the output it does because it is precisely executing the specific ideological, radical, biased agenda of its creators,” Andreessen wrote. “The apparently bizarre output is 100% intended. It is working as designed.” I suspect that Andreessen is right. And if he’s right, then we have just been told a great deal about what sort of company Google is likely to become a quarter century after its founding.