When you’re shopping online, you’ll likely find yourself jumping between multiple tabs to read reviews and research prices. It can be cumbersome doing all that back and forth tab switching, and online comparison is something we hear users want help with. In the next few weeks, starting in the U.S., Chrome will introduce Tab compare, a new feature that presents an AI-generated overview of products from across multiple tabs, all in one place.
Imagine you’re looking for a new Bluetooth portable speaker for an upcoming trip, but the product details and reviews are spread across different pages and websites. Soon, Chrome will offer to generate a comparison table by showing a suggestion next to your tabs. By bringing all the essential details — product specs, features, price, ratings — into one tab, you’ll be able to easily compare and make an informed decision without the endless tab switching.
↫ Parisa Tabriz
Is this really what people want from their browser, or am I just completely out of touch? I’m not at all convinced the latter isn’t the case, but this just seems like a filler feature. Is this really what all the AI hype is about? Is this kind of nonsense the end game we’re killing the planet even harder for?
It’s all about the shareholders. Hype drives up stock prices. The game is to inflate the stock prices as much as possible so the big investors can cash out at the last minute, leaving everyone else in the lurch when the bubble implodes. The whole thing is a scam floating on econoreligious fairy dust, completely disconnected from the reality of what consumers want.
That’s not to say machine learning has no use cases (hell, I would benefit from some of them). But stuffing “AI” into everything is just economic shell games. The hype itself is the product, not the machine learning.
I absolutely don’t want my browser offering anything. This should clearly be an extension.
With that out of the way, I am extremely curious to know if you (Thom) ever actually used AI. For example, did you give Claude a go? I’m asking because your takes make me think like you haven’t. BTW, it’s a genuine question, I’m not trying to be mean.
drstorm,
+1
Features like this should be done as extensions. It’s not integral to browser functionality. The browser core should focus on being lean.
I would like to see a hands on AI series. I think it could be popular but Thom might feel like he’s giving into something he objects to.
I love AI as a field of research and technology. To use it to help us buy more of what we dont need to enrich people who dont need it seems a waste of energy. But its just me, i am pro humans.
I’ll be a contrarian…this could be a fantastic use of AI! Going through dozens of tabs pulling out specs, reading hundreds of reviews and compiling spreadsheets is exactly the sort of thing I’d love to have an AI assistant for. It’s one of several great applications and I would be all over this if it were independent, open source, and the AI ran on my own machines. Unfortunately therein lies a huge problem: this is neither the open source nor independent AI future I wanted.
It’s totally irresponsible for society to have monopolistic corporations in positions of such power over users. Vendor locked AI can exacerbate this. My interest in AI is sharply shaped by where it runs and who controls it. It sets a very dangerous precedent for browsers to send so much information to a company like google. Moreover there are significant questions over AI being programmed with ulterior objectives. Google being an advertising company makes this threat particularly concerning. Just as google adjusts search results to increase their own profits, who’s to say google won’t do the same with AI? Even if it doesn’t start out that way, they can get us hooked and then compromise it.
There’s no doubt people will want this, myself included, because it is a huge time saver….but I have serious concerns over control and integrity. I want to have AI that’s aligned to FOSS ideals, but is this just wishful thinking?
So far there is no “SEO” term and thing for bending “Chrome Tabs Summary AI” to the liking of sellers but it will definitely will come if people in masses will start using such features instead of reading specs and reviews.
To much input info might be as bad for humans as for AI… Of course if you really care for quality of output.
jurmcc,
I agree that spamming information sources is bad but I disagree that AI aggregators make the problem worse. The integrity of reviews is a paramount for trust regardless of if you are a human or not. Product review spam affects us all and it is the exact same problem either way, It’s up to ecommerce platforms to weed out the inauthentic data.
AI data aggregators are not responsible for this spam problem, however you can make the case that AI text generation is. AI is able to generate tons of fake reviews at low cost. As captchas become obsolete and AI generation improves in quality, there’s not much that can be trusted. One of the most reliable data points that an ecommerce platform should have is whether the account associated with review actually purchased the product. Some ecommerce sites report this alongside the review and as far as AI aggregators go, they could limit their input to confirmed purchases, which should help. Some vendors may be purchasing their own products, but at least it’s expensive to purchase reviews in this manor.
Even if such a system were perfect, which it is not, and if we could trust google not to optimize the “ai” for their profit margins, which we can not: companies will game such a system to their heart’s content. It’ll be similar, but much worse than current seo.
“Is this kind of nonsense the end game we’re killing the planet even harder for?”
Yes. The answer is yes.
This feels like a great use case of AI to me. I’ve spent Hours researching a particular computer components across multiple tabs and sites to work out which (on balance) has the requisite features.
Certainly in western culture, we want our information to be presented as concicely as possible. And this help that.
In more eastern cultures (eg Japan) the user wants all the information presented in order to form an informed decision.
Both these scenarios are great use cases that can be facilitated with AI.
I can’t wait until this casually starts slipping sponsored items into comparisons.
Kver,
Pretty much the fate of all technology:
Printing press…early spam.
Cable TV… spam on the double.
Telephones, let’s spam them.
Email, let’s spam it.
World wide web, let’s spam it.
Twitter/facebook, let’s spam ’em.
Youtube, let’s spam it.
App stores, spam enabled.
Operating system, spam=on.
AI assistants…exception to the rule? I doubt it.
With the roll-out of manifest v3, google is seeing to it that no more adblocking innovation is possible under chrome. At least in theory though an adblock AI could actually be trained to help block the spam that we’re all being inundated by.
Instead of a nice, mostly efficient filter that blocks content before it’s loaded we’ll have power-guzzling AI chips trying to whack-a-mole crap as it pops in. Then services will of course use AI to try to make ads that bypass other AIs, and we’ll achieve the frickin’ singularity over Tide ads.
Is this the worst reality? Are we already in the ad singularity? Does your mother exist? Do you remember her face? IS HER FACE BOUNTY, THE QUICKER PICKER UPPER?
Kver,
I agree that’s a bad thing, but AI spam generation is already upon us anyway. The tide of AI spam on wordpress blogs including osnews is here. Apart from the obvious spam link, the text that those spammers are submitting is really advanced.
https://www.osnews.com/comments-by/?user=mariah9x
https://www.osnews.com/comments-by/?user=amorapotter
For the purpose of education, I’ll post a quote without the link for obvious reasons.
This was posted in the recent article about the managarm microkernel. Remember how spammers used to post irrelevant text across blogs? That is no longer the case, modern spam passes as human. Other than the very obvious spam link, comment is provocative and on-point.
These spammers are paid to post links, so probably the best way to deal with them is just to block new accounts from posting urls until they can gain some more reputation. Also, maybe detect that they are spammers the same link across thousands of websites.
In terms of an AI generating content that isn’t spamming blogs with commercials, I’m not sure how to feel about that. AI chat bots do produce insightful discussion. AI generated image and videos are already so impressive:
“This AI video generator breaks Hollywood”
https://www.youtube.com/watch?v=N96CcEHlAik
I think too many people have been underestimating AI’s efficacy at replacing skilled humans in the future. More people are likely to face job redundancies as AI becomes more specialized,
Integrating disparate information into a coherent subset based on a query is exactly what LLM AI is good at. However it will be a real shame if it becomes another monopolised lock in for Google. We need someone else to do it better, maybe as FOSS and certainly for a fee. There are no free lunches. Better to pay and get some objectivity than to think it’s free and be manipulated.
Iapx432,
I share your concern that AI will become monopolized by gatekeepers. This is bad for everyone except the gatekeepers. I hope AI can be democratized, but FOSS development can hit financial walls and can even be exploited by corporations using it without contributing back.
Considering how crap the normal Google search is with finding useful information about products, this could actually make it better. Do i want this? No, i want Google to revert the last couple of years development of their search instead.