There seem to be very few adults in the room at Google as of the past few years. This also happened with Bard's (remember when they thought Bard was a good name?) racist image generator. Does nobody do QA, UAT, or UX testing anymore?
I got Bard to execute a fork bomb, I think, when it was first launched. I noticed that it was evaluating the flow of a shell script incorrectly and it allowed execution, then refused to respond for 20 or 30 minutes.
Now, for some users, sharing a YouTube link adds some sort of tracking which then asks people who open the link to be "friends" with the person who shared it. Who thinks that this is a good idea, and why isn't it opt-in?
People who want to destroy you at work (which I imagine is rampant at Google) will sometimes prevent you from doing something, but they might also encourage you to do it in such a way that you fail spectacularly.
Some of my favorites from my career:
1. PM enticing me to "volunteer" to build a PoC over the weekend, to have it ready by Monday morning – come Monday, PoC is shipped live to all international customers. It was nowhere near prod ready, it was just to demo 1 thing.
Naturally, endless tickets open asking about missing functionality, and changes to the dashboard, "Nothing works!".
I'm in Slack talking to customer support like "Sorry I don't know why they launched it, it's not even close to being done". I've had nightmares about this kinda thing before, and then it happened lol. Anyway it was easy to paint the image that I'm just trash, biting off more than I can chew, can't let me near anything, and I got fired for it. They want back to maintenance mode.
It was a simple dashboard that looked great (for investors) and solved problems for customers, but there was no way the other guys on the team were going to let me launch that thing. Rather than stop it, they just "helped" get it out early. Project dead, they never built that.
2. And on the other end where I was stopped from launching something, I've had 2 devs go in league to say that code I was writing was a huge security risk (front-end React) and kept saying in meetings "Yeah I saw some things" without pointing out exactly what.
At some point one of them just irreversibly deletes all the branches related to it, live in standup, and puffs his chest "Yeah sorry guys we're not releasing this stuff, I'm worried about code quality" basically got him promoted to CTO.
So yeah I imagine politics and mind games happen all the time at Google, and it's probably very confusing what is or is not a good idea, what will work and will not work, because of the massive distraction of everyone backstabbing and preparing to backstab.
> but they might also encourage you to do it in such a way that you fail spectacularly.
A skill one must develop is identifying no-win-scenario tasks that can be avoided. POCs are often such tasks.
A teammate and I got roped into some POC where everyone in the room seemed to silently agree the product was not fit for purpose. The meetings went in circles of seemingly trying to find anything they could do to use the product, rather than assessing it objectively on the use case the vendor pitched.
Behind the scenes it turned out the product was being pushed by someone seniors brother. So the thing was getting bought no matter what. What they were really looking for was the patsy to own finding any sort of use case for the stupid thing, integrating it and owning it.
Obviously a "the product cannot fail, it can only be failed" scenario. Run run run away.
The reverse of this is also bad. Numerous arduous processes that exist mostly because the people who ship nothing far outnumber those who do. I imagine striking a perfect balance is difficult.
It's a pattern much wider than Google or event technology sector. It's very apparent in AI craze trough society, public governance sector, journalism etc. For people like me (growing up in seventies) it seems that there is teenagers everywhere and nobody is responsible for anything any more.
Exercising zero political awareness or foresight or understanding of people. The problem isn't that this was a mistake. The problem is that this mistake was obvious from a mile away to anyone who pays any attention to the world.
Also, the given justification for the feature was absolutely silly.
> Ex: Help students visualize the past instead of just reading about it. Teachers can show their classroom a historic site and type, “Render a hyper-realistic view of what these Pompeii ruins looked like in 78 A.D.” The modern-day ruins will instantly transform into a bustling, colorful street scene from the Roman Empire.
The bubble needs to burst already. Everyone so braindead that they actually build and ship a product like this fucking garbage needs to be out on the street.
Like, read that fucking prompt again. If that doesn't make you want to stick forks in your eyes, you're deep, deep in the psychosis.
Yes, at some point the world needs to accept that these tools are very good, to a limit, for specific human augmentation uses (coding, text manipulation) and stop pretending the TAM is "replace all white collar work across the globe".
I think if Google stopped retreating so quickly with its tail between its legs after getting “embarrassed”, people would stop caring. All of the “sins” that were created via this is nothing that other AI couldn’t produce either.
Part of the sport in making Google look bad is because they know Google will predictably cower when they are embarrassed. If they just said “ya, sometimes people will make bad pics but who cares?” people would stop caring.
Everything that Ars Technica found in my opinion is nothing that other AI image generators couldn’t do on their own. What’s the big deal?
Did Google Earth actually show these as replacements for the true imagery? Or are people simply complaining that people were able to generate these images at all? I'm struggling to understand the controversy.
Instead of having to go out of your way, there's an inviting "create a fiction" button. https://www.digitaldigging.org/p/how-to-plant-a-nuclear-plan... In February this year, someone manually forged a photo like this. "The Tehran Times posted it as a before-and-after of a US radar site in Qatar... AFP and BBC Verify both ran the “after” through Google’s watermark detector and got a hit, though not before it passed 950,000 views. Building it had taken six steps — screenshot Earth, open Gemini, upload, prompt, download, crop... Tonight Manama can be changed in seconds."
Right, I understand that you were able to generate these images from Google Earth. But nowhere can I find, including in that article, where other people could see your generations presented in Google Maps/Earth as if they were real imagery. The whole complaint is that people were able to generate these images at all. Or am I misunderstanding something? It seems like the complaint is that hypothetically you could generate these images and then share them elsewhere and try to trick people. Is that accurate or am I misunderstanding?
Some examples in the Ars Technica article include "refugees near the Mexican border", "a giant golden statue of President Donald Trump looming over the White House", and "a hospital with a bomb crater in Gaza". https://arstechnica.com/ai/2026/07/google-earth-releases-swi...
Any boast about AI accelerating the development lifecycle comes with an equal or greater stock that examples like this become more frequent when there are less opportunities for someone to interrupt them with a smidge of common sense.
Product managers at Google have a strangehold on their bad releases. It's become a whole thing for engineers to have to please their XFN (cross functional) partners in product and UX for review cycles and promos.
This is predictable... they built a misinformation engine capable of generating things like falsified drone strikes, terrorist attacks, industrial accidents, and threatening images. Seems like no one paused and thought "what would the malicious internet do with this?"
They'll shove AI everywhere, whether it makes sense or not doesn't matter. Sundar said "AI" needs to be everywhere, so they hope integrating AI into X will get them promoted. And unfortunately, this might even work..
You could take an area of Google Earth and let it be 'improved' by AI. So you could add a playground instead of a parking lot, or add migrants near a border post.
Due to consequences unforeseen by Google, it was used more for the latter.
I honestly don't know what the hell Google was thinking. I know shoving AI into everything is the rage these days, but this is simply an even more insane way to demolish the credibility of your own product than the AI overviews in Google Search.
Yes, they are at best misleading and often just present dangerously incomplete or misdirecting information. The more you know about a topic the more appalling the summary.
I check variations of this search every month and it still fails at a reasonably high rate
https://imgur.com/a/tKSQuA3
I find them irritating because they will link a source entirely unrelated to (or worse, directly contradicting) the fact they are claiming. The Google AI summaries might be the most blatantly hallucinating model widely deployed right now.
I’ve had them abruptly try to turn mathematical theorems into cooking recipes before cutting off mid-sentence in a panic.
YMMV, but nearly every one I read has a glaringly incorrect comment. I frequently research off-trail mountaineering routes and the incorrect bullshit it confidently spouts at the top of a list of results (results which very much counter the AI overview) is liable to end up getting someone hurt, lost, or worse.
I also see this failure mode where it inverts the question and answer the affirmative to the inverted question (therefore the opposite of correct answer).
If you read too quickly or are on a mobile device, the clarifying information may be below the line break.
Now, for some users, sharing a YouTube link adds some sort of tracking which then asks people who open the link to be "friends" with the person who shared it. Who thinks that this is a good idea, and why isn't it opt-in?
It's hard to take Google seriously these days.
Some of my favorites from my career:
1. PM enticing me to "volunteer" to build a PoC over the weekend, to have it ready by Monday morning – come Monday, PoC is shipped live to all international customers. It was nowhere near prod ready, it was just to demo 1 thing.
Naturally, endless tickets open asking about missing functionality, and changes to the dashboard, "Nothing works!".
I'm in Slack talking to customer support like "Sorry I don't know why they launched it, it's not even close to being done". I've had nightmares about this kinda thing before, and then it happened lol. Anyway it was easy to paint the image that I'm just trash, biting off more than I can chew, can't let me near anything, and I got fired for it. They want back to maintenance mode.
It was a simple dashboard that looked great (for investors) and solved problems for customers, but there was no way the other guys on the team were going to let me launch that thing. Rather than stop it, they just "helped" get it out early. Project dead, they never built that.
2. And on the other end where I was stopped from launching something, I've had 2 devs go in league to say that code I was writing was a huge security risk (front-end React) and kept saying in meetings "Yeah I saw some things" without pointing out exactly what.
At some point one of them just irreversibly deletes all the branches related to it, live in standup, and puffs his chest "Yeah sorry guys we're not releasing this stuff, I'm worried about code quality" basically got him promoted to CTO.
So yeah I imagine politics and mind games happen all the time at Google, and it's probably very confusing what is or is not a good idea, what will work and will not work, because of the massive distraction of everyone backstabbing and preparing to backstab.
A skill one must develop is identifying no-win-scenario tasks that can be avoided. POCs are often such tasks.
A teammate and I got roped into some POC where everyone in the room seemed to silently agree the product was not fit for purpose. The meetings went in circles of seemingly trying to find anything they could do to use the product, rather than assessing it objectively on the use case the vendor pitched.
Behind the scenes it turned out the product was being pushed by someone seniors brother. So the thing was getting bought no matter what. What they were really looking for was the patsy to own finding any sort of use case for the stupid thing, integrating it and owning it.
Obviously a "the product cannot fail, it can only be failed" scenario. Run run run away.
Also, the given justification for the feature was absolutely silly.
> Ex: Help students visualize the past instead of just reading about it. Teachers can show their classroom a historic site and type, “Render a hyper-realistic view of what these Pompeii ruins looked like in 78 A.D.” The modern-day ruins will instantly transform into a bustling, colorful street scene from the Roman Empire.
AI for image generation is great but I’m struggling to figure out why you would ever use this if it’s confined to satellite imagery.
Like, read that fucking prompt again. If that doesn't make you want to stick forks in your eyes, you're deep, deep in the psychosis.
Part of the sport in making Google look bad is because they know Google will predictably cower when they are embarrassed. If they just said “ya, sometimes people will make bad pics but who cares?” people would stop caring.
Everything that Ars Technica found in my opinion is nothing that other AI image generators couldn’t do on their own. What’s the big deal?
I am almost sure this was something for a performance or promotion: “Used AI effectively to…”
just traffic accident https://cdn.xcancel.com/video/9829FB6B34807/https%3A%2F%2Fvi...
refugees on the streets https://cdn.xcancel.com/video/7631603BD54FB/https%3A%2F%2Fvi...
a bomb crater near a hospital in Gaza https://pbs.twimg.com/media/HOgBJEcWQAA--zV?format=jpg&name=...
a nuclear power plant in Iran https://pbs.twimg.com/media/HOgBP-OXAAATHWQ?format=jpg&name=...
a plane hitting the One World Trade Center https://pbs.twimg.com/media/HOoW-0HWIAAdl0y?format=jpg&name=...
the Eiffel tower collapsed https://pbs.twimg.com/media/HOoW-z0XkAAN0KI?format=jpg&name=...
an army base in the Amazons https://pbs.twimg.com/media/HOgmohwWEAEn3gq?format=jpg&name=...
It has got to the point it’s poison outside the niche tech bubble.
- all the AI corps
Due to consequences unforeseen by Google, it was used more for the latter.
Glad they came to their senses quickly.
I check variations of this search every month and it still fails at a reasonably high rate https://imgur.com/a/tKSQuA3
I’ve had them abruptly try to turn mathematical theorems into cooking recipes before cutting off mid-sentence in a panic.
If you read too quickly or are on a mobile device, the clarifying information may be below the line break.
https://imgur.com/a/tKSQuA3
on a mobile device (iPhone Pro Max) with Large Text enabled you can see how the clarifying text starts to roll over to the next line:
https://imgur.com/a/wvNspMT
Imagine a senior on an older smaller phone with large text enabled, easily missing the Opposite Day answer.