It's just so... disrespectful.

The reason they do it, of course, is that some subset of the company has AI-related metrics they want to hit, and getting you to click that button is how they do that.

Which is so backwards. Of course, in the first place, a lot of these are free products, so it's not surprising that changes in the product benefit the company instead of you. But there is still supposed to be a modicum of respect: features should be there for you to use if you need them, not trying to get your addition and screaming "use me!". The stupid AI metric they're trying to hit isn't even real or beneficial to them; it's fake wallstreet crap because all the management lives in the same AI bubble and it looks good to say your feature got adopted. But the stupid feature isn't even useful or good. Probably decent documentation would be far better. And you just know the button wiggles on the paid version too, they don't give a shit.

This whole industry is conveniently abstracted away from its users by an app or a website, or whatever, so it manages to get away with treating the users like statistics instead of people. It's so disgusting, that the whole world is being taken over by this. Working at a company that does this should be a source of shame. "Manipulating your users" is a vile bottom-of-the-barrel business model. Fortunately everyone also sucks at it so it's still possible, for the most part, to ignore them (social media feeds being the notable counterexample). But it's just such a shitty world to live in.

If a real person did this stuff to your face--selling you something and then wiggling useless features at you out of the corner of your eye forever--you'd curse them out and try to never do business with them again. That's how it should be when a tech company does it, too.

But all governments, or at least all the ones I deal with on a regular basis, are switching to AI for essentially everything as well. The issue I already see coming more and more is that they just don't have an escape hatch (whereas your bank's helpdesk, which no doubt is doing the same does have an escape hatch to talk to a real human ... usually after waiting 10 hours to one song). Your passport renewal gets denied? "Please try again" and back into the app + AI recognition you go!

I don't really care if someone uses AI for something that it's useful for. But that's not why random bullshit websites are doing it. They're trying to get you use it for some other reason--either a (likely false) hypothesis that "AI feature users = more engaged = more likely to pay" or "AI feature users = an audience that's set up to be advertised to = investors are happy to hear about this" or something.

That, or the <18 and >50 users are all addicted to the AI mode and I have no idea because I'm not in either of those buckets, or something. But it's still disrespectful, even if some people are clicking on it.

I would donate money for this comment to be on billboards. It's exactly how I feel.

Heck, gemeni is obviously so successful and gets used so frequently that android had to hijack the power button (at least on pixels) so when you hold the power button, it activates gemini instead of turns the phone off. Clearly, that is because everyone intends to activate AI instead of turn the damn phone off when they hold the power button. `usage_metric++;` /s

Yet I know most people are also [probably] irritated by stuff like this. But _we_ do it anyway...

Even though I don't use Gemini, I have a permanent Gemini 'FAB' located above the 'Start Chat' FAB in Android / Google Messages and I can't seem to get rid of it.

Material 3 guidelines say "Don't display multiple FABs on a single screen":

https://m3.material.io/components/floating-action-button/gui...

For those like me who didn't know what a FAB is, it stands for "Floating Action Button", which is intended to be the primary action for a page

I was wondering why I don't see what you see, but I realize I use Google Chat (which used to be XMPP, current status I don't know) and not Google Messages. What do you use this app for? RCS messaging?

Ah, the benefit of using the messaging app Google's neglected is the lack of this Gemini button...

Kagi, who up to now have been respectful of their users, have just introduced a floating 'Quick Answers' AI summariser button in the bottom right of their search results page on mobile - precisely where one would place one's thumb to scroll the page.

I've already inadvertently clicked on it once.

Very disappointing and they don't provide any toggle to remove it.

> Very disappointing and they don't provide any toggle to remove it.

There definitely should be an easier way, but here you go:

  .qa-mode-box {
    display: none;
  }

They’re going to remove it

https://kagifeedback.org/d/8729-feedback-on-the-new-quick-an...

Google Sheets now has an intrusive Gemini pop-up that obscures other cells and UI when you try to do things like edit a formula, it's beyond infuriating and I can feel my blood pressure rise just thinking about it.

The way Gemini (which is incredible) has been haphazardly shoved into gsuite products seems devoid of any kind of common sense. Even very basic obvious things via Gemini in Gmail fail. It’s ironic because it’s the one place I’d love a decent model.

Incredible at some things maybe, but for the average user Gemini is what broke Google Assistant (the article touches on that). Most people use Google Assistant to "do things" like play music, interact with smart switches, etc. Rolling out Gemini as a replacement before it could do these things was always going to be viewed as a regression.

People don't care if it can answer questions with more nuance when the smart stuff they paid hundreds or thousands for no longer works.

I use Assistant to add reminders or stuff to my calendar (saying "Add appointment for 22nd of December at 2pm take sleigh out of garage" is a lot faster than clicking around the calendar UI), a few months ago Gemini couldn't do these things, yesterday it offers to do so after I connect it to the Workspace apps (I'm using civilian Google). But asking it to list my reminders gives me a textual listing in its UI instead of loading the Tasks UI...

Getting a response to "Set alarm for 8am" is still a lot quicker in Assistant than in Gemini.

The weird thing is that other AI systems can route commands to different models. Google should be able to send questions to Gemini and commands to Google Assistant.

I used Google Assistant to call my wife whem I was driving.

Previously it just called her, Gemini asks which one of the numbers I have saved should it call, but not as a continued conversation. It comes up with an onscreen list, which I'm not pressing whilst driving, and trying to tell it which one ends with Gemini getting very confused.

Reverted to Google Assistant which still works and will actually call her.

Reminds me of when Google Plus got “integrated” everywhere, but even worse.

If we take the thing nobody asked for and shove it in users faces as often as possible, this will get lots of happy customers and the product will take off!

The number of random buttons in Google products that have been replaced with Gemini features to juice engagement numbers is insane.

Everything from picking an image in Slides now activating image gen to returning to your main search results now being taken up by AI Mode

You wonder if people are really clearing performance reviews off these metrics without anyone digging deeper

I blocked that with uBlock origin the moment I first saw it. So annoying! Someone should make a filter for all of these AI widgets.

We can go for a first release, together. You have these, I have about 8 things filtered out of Atlassian products.

If I remember on Monday, I'll reply with my portion. Don't own nearly enough to go near radiation on my weekend.

Change the domain (before ##) to match your instance:

    your_co.atlassian.net##[data-testid*="issue-smart-request-summary.ui.ai-container"]
    your_co.atlassian.net##[data-testid*="issue.views.issue-base.foundation.status.improve-issue"]
    your_co.atlassian.net##[aria-label="Summarize"]
    your_co.atlassian.net##[data-testid*="atlassian-navigation.ui.conversation-assistant.app-navigation-ai-mate"]
    your_co.atlassian.net##[data-testid*="highlight-actions.ui.popup-target.popup-dialog.action-items.define-button.define-button"]
    your_co.atlassian.net##[data-testid*="context-menu-rovo-chat-button"]
    your_co.atlassian.net##[data-testid*="context-menu-rovo-chat-more-options"]
    your_co.atlassian.net##[data-testid*="atlassian-intelligence-toolbar-button"]
    your_co.atlassian.net##[data-testid*="placeholder-test-id"]

Surely some repeats/inefficiency: barely know what I'm doing :)

Clippy!

Ask Jeeves?

Oh, no wait! Google answers?

Clippy at least had the courtesy of only blocking the bottom right of the window, these ridiculous Gemini pop-ups appear next to whatever cell you have selected and block things like column headers or values that you actually want to look at while editing the formula/value. It's insane to me that this was ever allowed to get deployed without a rational human being calling out the obvious issue.

The Firefox right-click context menu is awful for this.

Incredibly annoying.

I must be blanking, because I can't remember any major shuffles on the (Linux) desktop. Either you've noticed and I haven't, or it's stripped out / disabled by default by the maintainers.

Do you mind elaborating on what I've missed?

I’m using macOS.

The menu item often appears after the rest of the context menu. Since it appears in the middle of the menu it causes a bunch of items to shift and and leads to misclicks.

It would be less of an annoyance if it appeared straightaway but was disabled, and then enabled once it was ready to use, because then it would avoid moving already visible UI elements.

In general I don’t like visible UI elements to change position unless it’s in response to a direct user action (e.g., expanding or collapsing a group). It’s the sort of thing that trashy news website do to ensure they get at least some ad clicks, and it always feels janky and causes frustration.

Thank you. Mine appears fast enough that I hadn't actually noticed - seemingly by the time I release my right-click button. This is on (what I consider) obscenely powerful hardware for a web browser though: a 5600X, 64Gb RAM, and a hardwired connection with geek-level hardware all the way up the chain. Basically the best possible scenario, and it's still noticably slower than the rest of the context menu now that I look for it.

I just tried right clicking on Firefox and got 15 lines of text of which one is "Ask an AI chatbot". Is that it?

Yeah, that’s it, but that item often appears after the rest of the menu for me, causing items below it to shift down one place, leading to misclicks.

I don’t mind that the item exists: I mind that it makes other menu items move after they’ve already appeared.

I just checked and if you mouse over it it has an option "remove ai chatbot" which works immediately, so at least there's that.

I recall that Amazon Q bubble at top right really gets my goat at work.

I just checked on my private account and it does not look so bad. Must be a combination of things leading to the bad behaviour at work.

The underlying problem here is that a click or tap should always refer to whatever _was_ on the screen a few (100?) milliseconds ago. It’s a long-standing bug on all platforms.

That isn't a bug; it's working as intended. I do wonder if that could be mocked up easily in html to see what it feels like though. Sounds like a cool idea

Yeah, my company has migrated from on-prem to the cloud Confluence. In addition to being slow as a snail, the cloud version is littered with so called "AI" landmines. There is an "AI" button almost on top of the scroll bar, there is another one near the edit/save button, then there is a hideous popup which appears on any text selection and of course there is "AI" junk there too. All the while toolbars for doing actual work were stripped, cut or removed completely. Adding rows, resizing, moving etc. is now a fun little point and click game with a built-in 500ms+ lag response. God, I hate llmification of everything.

It's ironic how many PMs want to use Atlassian products, when they're clearly some of the most obtuse options out there. If you care about product, wouldn't something like Linear be an obvious choice? It's snappy! (No affiliation, just liked the product.)