Technology

Not only that, they admitted that it is only "most" users that understand it's not accurate.

What about the rest? Just fuck em I guess? Let them eat poisonous mushrooms and wire their life savings away to a Nigerian prince.

And by making their user's initial impression one that might be wrong, they effectively poison people's thought processes as they sift through the rest of the actual results. Is that a better experience for the users?

And just to be absolutely clear, my understanding is that giving your users the best experience possible is what's going to result in retaining the most market share. Search is free, users will happily use a different search engine if a particular one doesn't work or annoys them.

So here's my conclusion: this method of using AI in search is not only frustrating for users, but also bad for Google. For the sake of their users, they should stop. For their own sake, they should stop.

Luddite!! Don’t you understand that number go up??

I work with a bunch of non tech yokels. They 100% for sure do not understand that AI answers must be checked and verified.

I can't wait till all this shit gets pay walled and 95% of people won't be using it. I don't really see Google being willing to just eat the losses on ai search queries forever.

The court caught that too:

The court also seemed to take a dig at Google for expecting users not to “blindly trust” AI overviews, noting that the AI tool’s utility “would be significantly diminished if the ‘AI overview’ were generally regarded as unreliable and if every single displayed link required independent verification.”

This is the value proposition of llms in general. They are great if you don't care about quality. They second quality matters their time-saving value drops off to near 0.

i saw some sites have disclaimers saying ai outputs are for entertainment purposes only.

in line of goog's defense: "everybody (most users) knows that"

It could hypothetically help you direct your search by surfacing useful keywords or relevant events or names or something like it. But since they didn't make it do that, it's not really reliable for anything but an energy expensive way to remind yourself of things you already know (what was the command for X again)

most likely is false?

I think the idea is that the info is probably true, but has high enough likelihood of being false that you better check anyway, if it's something that matters. There's a whole topic in machine learning called "probable approximate correctness" that tries to make that notion precise. Les Valiant's book of a similar title introduced the concept and looks very good. I have it but haven't read it yet.

This isn't new since ChatGPT and friends dropped. For years before that, Google search results did limited interpretation of natural language requests, not just keyword match frequency. The SEO arms race drove a different kind of AI in search fetching for at least a decade before natural language chatbot tech hit the scene.

I don't know how much is intentional enshittification to make AI results look better vs how much is simple neglect of the SEO arms race vs maybe it's genuinely getting harder to deliver good simple search results with LLMs acting as SEO agents?

What I do know is: "AI Mode" delivers more useful information than the old style page link list does these days. The pages linked from the AI Mode results tend to be relevant and useful more than the top page of page links. Hallucinations are way down from where they were 2+ years ago, even better than "top results" misses used to be, IMO. If you're not getting enough sources in your first AI mode response, ask for more - it delivers.

As was true since the first days of the internet: trust nothing. This is random junk people stick on the web for their own purposes, you have been warned.