So they integrate Gemini to summarize open web pages and consolidate all your open tabs into summaries. (Open lot's of pages, then summarize them all.) You can search your history with natural language and type Gemini queries directly into the address bar.
This will give them a cognitive profile of you: reading comprehension, decision-making patterns, knowledge gaps, etc.
This sounds an awful lot like Microsoft's Recall, only implemented in the browser.
Granted, there have been a lot of times I have trouble finding a website in my history, open tabs or even bookmarks, so I could potentially see how that might be advantageous as long as I was in a situation where I had a second browser for "non-work" related tasks, or this was strictly prohibited in in-private mode.
>4. Find webpages you previously visited
>For those frustrating instances when you want to jump back into a past project but don’t want to scroll through your history to find an important website you previously visited, soon you’ll be able to use Gemini in Chrome to recall it for you. Once launched, you can try prompts like “what was the website that I saw the walnut desk on last week?” or “what was that blog I read on back to school shopping?”
As for their "agentic browsing assistant", I don't have much trouble adding stuff to my shopping cart or other minor tasks. I'm still waiting on that 'Google Duplex' [1] feature they announced years ago that claimed Google would make phone calls for me to make appointments and etc. Make a doctor's appointment? Dispute a charge on my bill? That's what I want.
Sometimes I wish companies would stop forcing AI features down our throats and putting them just everywhere. At least I hope I can properly disable all of this. I don't need an AI agent scanning everything behind my back.
Very interested to see how well the agentic features work compared to ChatGPT’s cloud version. At the very least, I imagine bot detection/prevention (I.e., CAPTCHA) will be less of issue with Google’s strategy since the browser’s fingerprint will differ Chrome user to Chrome user.
Some of these features would be nice to have, but I'm not sure I even want my browser to have these capabilities unless there is a mechanism to keep it all local. This is a monumental change to the amount of user data Google can collect via Chrome.
Not a comment re: AI in Chrome. I did click into the video to hopefully get an understanding of the features,. but sadly did not. Sorta got an idea, but the jump cuts, super-quick overviews, trying to identify everyone speaking, etc. E.g. ...something about tabs... ...used to take 20 minutes, now seconds... ...working with OTHER Google products (WTH)... Hey Google, make a simpler, more accessible video that is "just the facts mam". Dozens of jump cuts in a few minutes is disconcerting, and may prove dangerous to some.
I don't know the future of browsers given the trends in AI, but it seems fine to add an opt in ability to browsers to allow an LLM to access the current (or a set) of tabs. If it works it would reduce the amount of copy-paste, which seems like a good thing.
It's hardly a killer feature. I'm still going to use chatgpt (and gemini) a tremendous amount.
When the webpage AI argues with the browser AI, which argues with the OS AI which argues with the on-chip mainboard, CPU and GPU AIs, while the monitor AIs frantically try to make notes and the smarthome AI watches all of it and can only shake its metaphorical head.
A lot of these new AI features can be helpful as long as the users' control the data. Is there an alternate world where Google might become the good guy here? In this world, as might be expected, the company making the worlds best spyware, wants to expand its spying.
It is astonishing that the word “privacy” appears zero times in this announcement. There have been repeated controversies over exactly how Google sees just the URL I visit. Now they want to see the entire contents of multiple browser tabs?
It is astonishing that the word “privacy” appears zero times in this announcement. There have been repeated controversies over exactly how Google sees just the URL I visit. Now they want to see the entire contents of multiple browser tabs?
And yet my healthcare company's IT department insists everyone only use Chrome on work computers.
I've been bringing up the HIPAA implications of that for years, but people just look at me like I asked the dog to do algebra. All the IT department cares about is that Chrome is free.
Can I block it as a site host? (Please don't respond about how I shouldn't want to and isn't it just like some other usecase that I'd obviously want not to block)
Taking a quick test spin. Seems to be enabled (on mac) by the existing "Gemini in Chrome" extension but requires a Chrome update. Which has a global shortcut (across all mac apps) of "Ctrl+g" (this can be changed). The additional AI features seem to come from the tab content's integration into the Gemini console (previously (I believe) the Gemini extension was merely a thin console to the Gemini service)).
The direct tab integration works by first showing the Gemini console (ctrl-g or Gemini icon in the mac system tray) where there is then a 'Share current page' icon below the text view.
This adds a blue border around the chrome window indicating the current page can be shared with chrome. It's not clear, but I believe the page content is only shared once a prompt is made intending to use the page's content. However, the share-enablement remains enabled for all tabs (all tab windows will have the blue border) until turned off in the Gemini consol. Again, it's not clear if just merely browsing these tabs will automatically share that content with Google.
The Gemini integration does not perform actions (can't ask it to do stuff directly with the site's content (navigate, click buttons, etc).
But direct summarization works well (try it on an HN comment page or news article).
Overall, I like this feature as long as I understand what and when things are being shared and ability to turn off easily.
This seems huge to me. As in: the initial release of Google Chrome huge. I don’t know if they can really pull off things they showed but if they do I’m sure this will be a massive success. Which is pretty sad considering the privacy implications for this. Imagine how much more data they will have on everyone. Scary.
51 comments
[ 3.0 ms ] story [ 54.1 ms ] threadThis will give them a cognitive profile of you: reading comprehension, decision-making patterns, knowledge gaps, etc.
Scary.
Granted, there have been a lot of times I have trouble finding a website in my history, open tabs or even bookmarks, so I could potentially see how that might be advantageous as long as I was in a situation where I had a second browser for "non-work" related tasks, or this was strictly prohibited in in-private mode.
>4. Find webpages you previously visited
>For those frustrating instances when you want to jump back into a past project but don’t want to scroll through your history to find an important website you previously visited, soon you’ll be able to use Gemini in Chrome to recall it for you. Once launched, you can try prompts like “what was the website that I saw the walnut desk on last week?” or “what was that blog I read on back to school shopping?”
As for their "agentic browsing assistant", I don't have much trouble adding stuff to my shopping cart or other minor tasks. I'm still waiting on that 'Google Duplex' [1] feature they announced years ago that claimed Google would make phone calls for me to make appointments and etc. Make a doctor's appointment? Dispute a charge on my bill? That's what I want.
[1] https://youtu.be/D5VN56jQMWM
I don't know the future of browsers given the trends in AI, but it seems fine to add an opt in ability to browsers to allow an LLM to access the current (or a set) of tabs. If it works it would reduce the amount of copy-paste, which seems like a good thing.
It's hardly a killer feature. I'm still going to use chatgpt (and gemini) a tremendous amount.
I really want the 2008 Google where everything they made was welcomed and not hated on sight.
Agentic browser? This. is. what. I. want.
Asking the browser about "that specific thing I might have seen last week?" Sign me the f up!
I'm not being sarcastic, I really wish I could have all of this and not having to worry about antagonistic companies and governments.
Fantastic, Googles AI will be fighting to stop the scams Googles advertising promotes to me...
"We need to push for the right to opt out of AI features."
You've had access to Firefox for over 20 years.
And yet my healthcare company's IT department insists everyone only use Chrome on work computers.
I've been bringing up the HIPAA implications of that for years, but people just look at me like I asked the dog to do algebra. All the IT department cares about is that Chrome is free.
The direct tab integration works by first showing the Gemini console (ctrl-g or Gemini icon in the mac system tray) where there is then a 'Share current page' icon below the text view.
This adds a blue border around the chrome window indicating the current page can be shared with chrome. It's not clear, but I believe the page content is only shared once a prompt is made intending to use the page's content. However, the share-enablement remains enabled for all tabs (all tab windows will have the blue border) until turned off in the Gemini consol. Again, it's not clear if just merely browsing these tabs will automatically share that content with Google.
The Gemini integration does not perform actions (can't ask it to do stuff directly with the site's content (navigate, click buttons, etc).
But direct summarization works well (try it on an HN comment page or news article).
Overall, I like this feature as long as I understand what and when things are being shared and ability to turn off easily.