Hacker News Reader

Comment by lars_von_pidor | Hacker News Reader

The only reason Google is pushing this AI crap is so that they can shove ads right into people's throats without them being able to use ad blockers (it's easy to block a web script but virtually impossible to block the text itself), effectively doubling their profits overnight.

View on HN

Block the AI overviews with extensions like https://chromewebstore.google.com/detail/hide-google-ai-over... or use a userscript to do the same.

You can block the entire AI response, but not the paid-for product placement in the response separately.

Block the entire AI response. It's not a good thing. It tells you whatever google wants you to see. It's an incredibly powerful brainwashing tool.

The search results without AI also tell you whatever Google wants you to see. The immediate solution is not to block AI summaries, it's to stop using Google entirely.

Not to mention the entire well is "poisoned" now. You can avoid LLM points of entry. You can't go to a random source and expect to avoid generative output.

There is a way to see old results, by adding "before:2023" to the search query.

Great, as long as you don't mind the nexus of all human communication to be frozen in time three years ago.

For many queries it does not matter.

These days, the AI response is often a lot better than the actual search results. Search result quality has dropped drastically the last decade. Sometimes it feels even Altavista had better results than today's Google.

The blog post says ‘These formats will also continue to be clearly labeled as “Sponsored.”’. We will probably be able to block them about as well as we can block sponsored search results.

Alternatively, just change your browser's default search shortcut, and add &udm=14 to the end of the normal google search. It changes the default search results to "web" rather than "All", which removes all the extraneous crap they've added over the years.

Compare https://www.google.com/search?q=test to https://www.google.com/search?q=test&udm=14

The wild thing is how much faster it is to load. I'd almost forgotten how fast Google's default search used to be. Thank you!

Or even better: stop using Google.

Use Kagi instead.

Both just give me a recaptcha. Start page doesn't:

https://www.startpage.com/sp/search?q=test

Which search engines does it use? I couldn't find info, and I'm just curious.

Can't find full list either, but it at least includes Bing and Google: https://support.startpage.com/hc/en-us/articles/452243553384...

How do they send the query on your behalf privately, without incurring insane search api costs?

Wow, great tip! Thank you!

Could also just block the element in ublock?

Semi-seriously: I imagine we'll live to see the day when we run an adblocker that runs a small model to semantically filter out ads in Google search results

Entirely accurate, but what an absolute waste of resources across the board.

Fighting AI with AI?

What a wild future.

I realistically don't see what other future there is. Either disconnect or adopt anti-grievance local AI, lest you be bombarded with ads and slop everywhere.

The first thing you install on your browser is uBlock, is it not?

Don't get me wrong, I hate this timeline, but the monster is out the box and there is no putting it back in.

Sounds like a good fit for a small, on-device model. Can Chrome extensions use the new Prompt API, which has caused a stir because Google pushed it through against opposition of virtually everyone else? (https://developer.chrome.com/docs/ai/prompt-api) Would be hilarious.

You won't be able to tell if it's an ad, if it's just biased (and the ad is not discolsed).

They could always sell ads like "recommend my tool more when user asks for cupcakes in London".

And then, the output would be: "My top 3 recomendations are X, Y, Z".

And maybe only X is the one that paid and Y and Z are organic.

That's almost certainly illegal in many jurisdictions, and they'd definitely not be able to hide that they're doing it indefinitely. A sure way to be massively sued.

Well, won't it be hard to test?

The customers could be paying for "increased visibility" in LLM results, but nothing specific.

Also, the ad platform could simply hint towards a customer, instead of directly mentioning, thus not being a true direct ad.

(NB: customer = the one buying the increased LLM visibility)

For example, a user could ask for "prebuild computers", and the ad platform could state: "the best ones for your use-case are those with AMD CPU, NVidia GPU, 2TB of RAM and a Z960X motherboard", and when you search for it, surprise-surprise, there is only one company in your location selling this specific setup. Or maybe, instead of linking directly to the company, it links to keywords/ideas that link to that company.

You get the gist, by using dynamically generated plain text and having more control over the user journey, you could actually steer the user into a specific direction without ever mentioning it.

All it takes is one determined journalist running experiments, or one whistleblower inside the company exposing these systems.

Plus, you'd have to offer this service somehow, meaning you'd have to describe it to potential customers... just another bunch of people you'd have to swear to perpetual secrecy.

I'm not denying it could (or even will) happen, though. Anything for a buck for these companies.

Being a monopoly is already illegal in many jurisdictions and that doesn't seem to stop Google.

"Cupcakes in London" is not a good example since it's directly asking for advertising. Ads in more information-oriented prompts would be much easier to spot, for example looking for out of context brands and products.

This might come as a surprise to many, but the sole reason Google exist is to make a profit. More profit means more success means more profit, that's why they did create a company in the first place. Mindblowing stuff, that.

In the US at least, I believe the FTC requires ads to be clear and conspicuous when those ads are designed to otherwise blend into the general editorial style. I could see AI being regulated as influencer marketing, but hopefully with more enforcement.

> but virtually impossible to block the text itself

Why do you believe so?

As long as there is a clear indication somewhere on the webpage (in the metadata or in the text itself) that a specific portion of a text is an ad, a browser extension will be able to block it.

And I assume that there are laws mandating that the ads must be clearly marked in order to be distinguishable from the genuine content.

That's only doable if the ads are artificially injected. But what if they are part of the training, system prompt or the search results that are fed to the AI? What if Google Search bumps up their paying advertiser up in the internal search results for Gemini (as they are basically already doing)? The AI will be biased towards the advertisers without literally embedding an ad into the output text.

> if they are part of the training

That would be an intentional poisoning of the models with biased or outright untruthful data.

I believe that many people would be unwilling to use such models.

They won't be if the models are "free", which is the case for AI Mode in Google Search. That's why common people still use Google despite it being an ad-ridden slopfest, it's "free"!

> what if they are part of the training

No way Google is going to bake the ads into training data. Their entire business is built on auctioning off each ad slot in realtime.

The law will not be updated or enforced. Laws don't reflect justice, they reflect the power relations in the society at the time the law was written.

Big tech is paying handsomely for this, and I don't think the populace is going to outbribe them.

It's just gonna say "this whole thing might be a big ad" and they will fight the fines in court for years, lose and book those fines as cost of doing business while laughing all the way to the bank

There's no way EU would let Google display ads without clearly marking them as such. So any ad blocker should be able to continue detect the block or link that's an ad.

Competitors will be very happy though.

Another massive reason is that ChatGPT and similar apps are eating their lunch. Asking a question to ChatGPT actually tends to be pretty convenient compared to the top X results that are just SEO optimized slop.

Reply on news.ycombinator.com