Status Update
Comments
qi...@gmail.com <qi...@gmail.com> #2
I want to try AI Assistance in Devtools in Google Canary 132.
I followed the instruct to login account, set language to US English, but when it came to “ turned on AI assistance in settings Settings > AI Innovations in DevTools.”, I found I cannot turn on the switches.
Those two switches seemed to be disabled, nothing happened when I click on them.
Did I do something incorrect? Or did I missing any operation?
I need some help, thanks.
al...@google.com <al...@google.com> #3
kh...@gmail.com <kh...@gmail.com> #4
sa...@gmail.com <sa...@gmail.com> #5
ch...@gmail.com <ch...@gmail.com> #6
mi...@gmail.com <mi...@gmail.com> #7
ch...@gmail.com <ch...@gmail.com> #8
al...@universityschool.edu.pe <al...@universityschool.edu.pe>
mr...@google.com <mr...@google.com> #9 Restricted
ka...@gmail.com <ka...@gmail.com> #10
mr...@google.com <mr...@google.com> #11
Thanks everyone for your feedback so far. I deleted some comments unrelated to AI assistance to make sure the conversation stays focused.
#4, yes. Better surfacing met or unmet requirements to use AI assistance is something we are currently working on.
#5: is it getting in your way? Do you want to have it accessible somehow? Would making it less prominent in the UI already work for you?
#7: yes, currently English is the only supported language. We are evaluating if support for other languages is important and I recorded your feedback.
en...@gmail.com <en...@gmail.com> #12
It seems like this is a feature that would be used every once in a while, and the other options in the list like "Reveal in elements panel", "Add attribute", "Edit HTML", etc... should take priority in the order.
an...@gmail.com <an...@gmail.com> #13
I use the "Reveal Element in Panel" context menu item inside the console quite frequently, and have done so for years. It deserves its place near the top of the menu. It's been somewhat jarring to have its placement upended by a vague "Ask AI" menu item which I have no interest in or intention of ever using. Even with the AI "features" disabled and unusable, I still have to see this option in the context menu for some reason. Please remove it!
The AI feature list purports to offer context about about console messages, but I only see this "Ask AI" option when i'm clicking on elements that have been logged in the console, and not other console messages. Like everything branded with "AI" these days, this feature seems half-baked and shoehorned in.
ka...@googlemail.com <ka...@googlemail.com> #14
st...@taopix.com <st...@taopix.com> #15
st...@taopix.com <st...@taopix.com> #16
mr...@google.com <mr...@google.com> #17
Thanks everyone for taking the time to give feedback about AI assistance.
To #12, #13 and #14: we hear your and other voices, asking for an option to fully disable AI assistance and/or reduce it's prominence in the context menus.
To #13: Console insights and the new AI assistance are two separate features. See
To #15: We are already tracking that internally as a feature request! :)
To #16: It would be awesome if you could create a screen recording of what your observing. I'll follow up with you via mail.
fl...@gmail.com <fl...@gmail.com> #18
am...@gmail.com <am...@gmail.com> #19
I would like a definitive way to opt out of all "AI assistance" features in the browser. I don't want to see "Ask AI" anywhere at all, not at the top, nor at the bottom. Not less prominent, I want it gone.
mr...@google.com <mr...@google.com> #20
There is now a
However, adding more context on why you want to disable it would be much appreciated.
am...@gmail.com <am...@gmail.com> #21
Thank you, that worked.
On Linux, create a JSON file (name doesn't matter) in /etc/opt/chrome/policies/managed/
that contains the following entry:
{
"DevToolsGenAiSettings": 2
}
This also disables "Console insights".
I would like to fully disable "AI assistance" because I don't want useless/harmful/immoral/unethical anti-features to pollute frequently used user interfaces.
ch...@gmail.com <ch...@gmail.com> #22
se...@gmail.com <se...@gmail.com> #23
ta...@gmail.com <ta...@gmail.com> #24
ma...@gmail.com <ma...@gmail.com> #25
kn...@gmail.com <kn...@gmail.com> #26
ke...@gmail.com <ke...@gmail.com> #27
mo...@gmail.com <mo...@gmail.com> #28
mr...@google.com <mr...@google.com> #29
#23: Can you please check your Google account is meeting all other requirements listed on
#26: AI assistance for sources is available in Chrome Canary (see
#28: Happy you find AI assistance helpful! Chat history for AI assistance is available in Chrome Canary and will come to Chrome stable in January, see
ee...@gmail.com <ee...@gmail.com> #30
to...@gmail.com <to...@gmail.com> #31
I strongly do not want any AI related functions enabled or displayed in the Dev Tools. I have them disabled as much as possible and the settings but chromium is still cluttering interactions with AI related indicators that increase cognitive load and actually cause me some level of mental distress. AI is a sensitive subject and I for one consider it a 4 letter word.
Please honor our opt-out of AI features and consider this an important use case for all AI related features going forward.
mr...@google.com <mr...@google.com> #32
#30: I'll try to reproduce this and then file a bug internally. Thanks for raising!
#31: While a workaround, there is documented way to remove AI features from the UI as well. See
hi...@gmail.com <hi...@gmail.com> #33
I tried something I feel is really simple and it failed.
Are you able to give me a list of Names and status from the network tab?
I tried a number of variations like "Give me a list of Names and statuses of anything showing in the network tab?"
All results in this error message: "Something unforeseen happened and I can no longer continue. Try your request again and see if that resolves the issue."
fi...@gmail.com <fi...@gmail.com> #34
The context menu is already bloated and not configurable; I have to scroll it when I work on my laptop. In full height, it occupies 2/3 of my 17" screen!
You can't just add new and new items to the menu without regard to its existing contents. That's not how UI/UX is done. Use submenus and, again, please allow to disable the AI item.
qq...@gmail.com <qq...@gmail.com> #35
er...@gmail.com <er...@gmail.com> #36
ka...@gmail.com <ka...@gmail.com> #37
co...@gmail.com <co...@gmail.com> #38
#20: There are lots of reasons not to want AI in the devtools. Here are three:
1) I don't trust AI to be helpful, and I certainly don't trust it more than I trust myself. These are language models, not knowledge models; they intrinsically do not understand truth and falsehood. (They don't even really understand language, either, as the "SolidGoldMagikarp" phenomenon with ChatGPT demonstrated. No one actually went out and modeled language as a concept for these AI; these AI are not the product of people analyzing language, breaking it down into meaningful elements and rules, and expressing that understanding in code. They are the result of companies like Google dumping massive amounts of text into a statistical analysis and pattern matching program; they are autocomplete programs operating at an industrial scale; they aren't any more "aware" or "knowing" than the autocomplete options we get when typing on a smartphone.) At launch, ChatGPT failed at doing basic math with three- or four-digit numbers despite having consumed massive amounts of text about math, because it couldn't comprehend and internalize the concepts it "read" about; Bing's generative AI openly gaslit and fought with users at one point; the generative AI that MDN attempted to introduce (against the strong objections of their own editors) gave inaccurate and low-quality suggestions specifically about web development; and Google's generative AI has historically not understood that rocks and glue-soaked pizza are unsafe for human consumption. Why -- knowing how this technology works, and after seeing all of these failures including from some of the largest companies in tech -- would I ever want to "Ask AI" for answers?
2) The technology is being marketed entirely on hype. Even the label "AI innovations," in the devtools options, feels like a subtle attempt at generating hype by trying to tell us how to feel about the concept: "It's an *innovation!* It's revolutionary! Be amazed!" Compare it to a neutral, practical wording like "AI tools" or "AI-based utilities." Compare the way AI is presented in the devtools to the way every other feature is presented: we aren't offered "Throttling solutions" or "Bracket-matching innovations" or an option to "Leverage auto-expansion of console.trace() messages to improve development synergies;" every other option is presented in a way that focuses on being practical and to-the-point rather than on the authors patting themselves on the back. The placement of "Ask AI" at the very top of the context menu is similarly hype-focused: it deliberately displaces helpful, frequently-used menu items out of the way in order to call attention to itself. It doesn't feel like a design decision made with a genuine intent to be helpful; it feels like an attempt to hijack our attention and muscle memory. This kind of hype and attention-seeking -- particularly in the context of all of the hype-based obfuscation surrounding generative AI as a field -- feels dishonest and provokes contempt.
3) The technology wouldn't be worth it even if it worked. If I need help understanding a CSS property or some SVG attribute, I don't need to send questions to a machine that consumes extravagant amounts of processing power and energy, performing statistical analysis on colossal amounts of scraped text, just to have it summarize online reference materials. I can read those materials myself. They aren't hard to find. The ecological and sociological costs of generative AI, and the ways those costs *will* worsen in the future, are not worth the "benefits" it would produce if it were reliable, and those costs also provoke contempt.
I do not want a thing that I don't trust, and that I feel contempt for, shoehorned into high-priority areas of the user interface, with no option to turn it off. Really, I don't want it present at all. There is no reason why this can't be an on/off toggle within Google Chrome itself, rather than requiring registry edits that literally don't even work.
mo...@gmail.com <mo...@gmail.com> #39
cannot refresh :
Something unforeseen happened and I can no longer continue. Try your request again and see if that resolves the issue.
___-----------------------------------------------------------------
my ai data isn't profiled and saved
& lost on each refresh
___------------------------------------------------------------------------------------
ma...@sdservices.nl <ma...@sdservices.nl> #40
But (and that´s my issue) the AI tool seems to store some previous version of what we've been trying to change.
So if I launch my last updated index.html by uploading new code to the ESP32 as a server, it appears completely as expected: all edits integrated. And now the strange part: refreshing the page (after making some edits in the latest version of the page) grabs an ancient version that the tool had previously elaborated.
Wiped my cookies, no avail. Reset the ESP32, neither.
Is there some way I can avoid this behaviour? It's hard to develop this way.
e2...@gmail.com <e2...@gmail.com> #41
je...@google.com <je...@google.com> #42
z1...@gmail.com <z1...@gmail.com> #43
There is now a documented way to disable AI assistance.
However, adding more context on why you want to disable it would be much appreciated.
Omg, thank you so much; it worked! I wanted to disable it because, while debugging in the console, the AI message constantly interrupts me. When hovering over a log, the pop-up covers a large part of it, making it difficult to properly select the log. It would be great to have an option to enable and disable this pop-up as needed. Currently, even if you turn the AI assistance “off”, it remains active, and the only way to really remove it is through the registry editor.
Description
Drop your feedback on the AI Assistance panel here. Learn more.