Google's Gemini Headaches Spur $90B Selloff
forbes.com
forbes.com
If this is rooted too deeply in Google's culture by now, they might take very long to recover. Like decades long.
Has anybody else here seen how the UIs of Google Adsense and Google Ads have evolved during recent years? They are a mess. Another indicator that something is wrong.
I witnessed firsthand, how SMBs tried to put ads on Google Ads in the latest version and simply gave up. Because they couldn't figure out the interface.
Did Google lose their ability to bang out useful software?
GA4 is even worse as well. I often find it just straight up doesn't load correctly.
Reddit and SO are barely recommended.
Product reviews used to go to higher quality blogs, Reddit, Consumer Reports now are now just all content filler sites.
During Trump they decided allowing anything except for MSNBC/ABC/CBS/CNN/etc on the first five pages of anything remotely news related were the only allowable results. This mirrors exactly the Gemini/OpenAI nerfs for PC.
Given 90% of my searches are for questions about what people think about things, code/food/drugs/health/medicine/etc, and topical events, it's become completely useless without adding "reddit" or something after.
I am presuming you are referring to Anthos.
Anthos is very well suited for the needs of large enterprise. It makes Kubernetes less painful for our development teams. It has reduced the learning curve for our teams.
So, their core services are fine; for anything that is not in the 'hot money making' path, good luck, some are great some are bad.
Every single one seemed to hold two contradictory ideas in their head - they were changing the world & their work was nowhere near any actual product or something that produced revenue.
It's like GOOG screened for this kind of person for over a decade, and now they are the majority of their staff.
If you want math 'from first principle' theory types because that's your proxy for competence in unrelated area or perhaps even IQ...
On the other hand, you could go work on a team which is ... not tied to revenue, or even a product. Then your value is whatever you can convince others of, or you can simply be forgotten about as there is no way to tell whether you are doing your job. I've seen people spend 4+ years delivering not a thing in this manner while product focused teams work 80 hours a week.
Which group would you choose to be in?
Anecdotal observation is that extremely successful businesses are often marvels of inefficiency as there is simply no need to make the business efficient. Efficiency is often contrary to employee goals as well.
Making a successful and well run business is very difficult.
This is true. Metrics are really about who owns the narrative and who runs the org. If the latter group get their guy into power then decisions will be made that cause inefficiency, waste and even reduced revenue far longer than is logical.
I've noticed over the last 10-15 years that computer science became the most popular major at my alma mater. It literally used to be the least popular a couple decades ago. A professor even showed us the data. It's also one of the easier majors now by grade distribution. My thought is that computer science has become like a new business/finance degree and that it attracts people who aren't interested in product creation anymore. Maybe this coincides with what you are seeing.
Doesn't mean he could/should run a business.
Many such cases of tech geniuses who have the mental capacity to deliver, and the verbal skills to advocate for their own power.. but not the mindset to actually deliver the things they've advocated for!
Pretty hard to screen for, and can do a lot of damage to an org as they siphon off resources from the teams of doers, squander them, and then take their ball & leave when it stops being fun.
Once we had a large news publication client who had a requirement of optimizing their backend for costs as they were spending something like north of $4000 per month in server costs. This was with Cloudflare on top since the homepage and some sections could never be cached as it had to be real time. To address this issue, we consulted an engineer who had no idea about even the database offerings Google had. He recommended something like a NoSQL solution. The more we grilled him for info on implementation and why he suggested it the more empty talk he gave us. And he wasn’t the only one. Google is really large, so we talked to a handful of them and almost all of them seemed very incompetent.
Frustrated, I walked past one of the areas where people were supposed to be working. And there were a lot of empty seats. And the few that were there were watching some Korean drama on YouTube, etc.
The rest were either sleeping off in couches and/playing snookers downstairs. For an outsider, it felt like a great culture thing to me. But later, after a dozen more visits and same working style observations, it occured to me that this no accountability cozy environment actually didn’t allow them to weed out the bad ones who didn’t even know the product they were supposed to specialize in and nothing good comes out of a company that is built on such levels of incompetence.
Soon after, I quit consulting for Google Cloud products as they also managed to increase prices without justifying the value proposition. I am not alone, many consultants who worked with Google would tell you the same. Their work culture or lack thereof is their worst enemy.
There were so many broken systems there. Once I made it clear to my boss I wanted to quit Google entirely he told me to just stick around for a year as he didn't want to do the paperwork to fire me.
Thankfully I'm back at a startup. These days if I see a lot of YOE at Google on a resume it is a red flag.
Ugh, plus 1 to that. As much as people love hating "the new orange guy", there's a lesson behind the fact that he got rid of 80% of staff in a big company in an afternoon and pretty much nothing adverse happened.
To be clear the revenue collapse might be unrelated to the full but I'd call it adverse.
It is completely unrelated. Revenue collapsed due to the powers that be getting angry at him for ending the censorship regime.
As for outages, the issues were little to none.
The boosting was done deliberatly and paid for by government agencies (of which some ex-employees where on Twitter's payroll).
It's quite irrelevant IMO whether the deliberate censorship for a certain US political flavor was done automatically or manually.
it's amazing how bad information fluffed up by the angry news media gets thoroughly routed in people's mind, then gets quietly defanged and drilled to earth to the point it's a completely innocuous gossip chain.
see currently the Biden impeachment and Russian propaganda laundering.
people really want to believe salacious stories that cut against popular perception to the point they just get to keep riding that wave of mental backflips to suit their increasingly deranged phantasm.
Yes, the lucrative Nazi market of CatTurd2 and friends
I would rather have consistency. Think of all the users that figured out how to buy ads the first time that might have stopped buying ads when the UI changed.
It is really hard to get users to change. The old MS Excel interface to many people a long time to adjust to the ribbon. I know people who have still not figured out the new pivot table UI.
Got access the other day to their latest AI models, and while everyone in the market will let you a) create an API token and b) use it, for Google you've a mess of service accounts, badly named permissions, services that change name all the time, 'just works' authorization that's basically a lock in to their own cloud. After spending a few hours with their docs, I just gave up - not worth the effort for now.
We're now also being asked to revalidate our integrations with them in the name of security (no change on our end, thousands of users using it every day). Another absolutely useless, bureaucratic, Kafkaesque process, with rules changing all the time, and not properly applied.
Google has long lost its way on software...
Separate from recent poor generations, I’d be reluctant to use any Google AI, as if for some reason I’d liked it, the odds are too high it’ll get shut down or replaced by some worse internal competitor.
Disclaimer: I'm still salty about Google Reader, Domains and Play Music.
This stuff should be easy.
The only way both can stop this is by sticking to a product, for better or for worse, and then actually innovate, not sit still once they've wrangled competitors.
The pattern from a business perspective is so repetitive as to be tiring: Get involved with anything that threatens their core revenue source, suffocate or buy out competitors, then barely push the needle forward ensuring that disruption never comes from within.
I believe that if Google weren't so against self-disruption that much of the latest AI innovation would be entirely theirs to own.
I can't think of any apps or services that Apple has launched in the past decade and has since killed, whereas without even trying I can rattle off at least ten such projects by Google.
While tech enthusiasts are a minority, they often drive the adoption of new products and technologies.
The biggest insult really is these Google products often do have a reasonably large and dedicated following, it just doesn’t seem to be enough to move the needle compared to a monster ad business.
Currently at 293.
When Stadia was announced, I was one to push back about the silly "they'll just shut it down" stuff by pointing out that there's a big different between a massively invested product like Stadia and an RSS feed reader or a Could Print service. Then they showed me that investment $ wasn't a reassurance either...
I don't think they'll shut down Gemini, in fact I'd be utterly shocked if they did, so I do think it's memey and silly to avoid Gemini because "they might shut it down," but I would certainly not bet my company on it (by building on top of their APIs).
2023-02-27: $89.87
2024-02-27: $137.17
"Good! Hope it falls further below $97 or $95 then I will just buy some Google stock." [0]
"I can confidently say that this stock drop is a typical over-reaction from the markets." [1]
Now it is recovered and is up 52% up since then. Not as good as Meta's >300% recovery but it wasn't exactly over for Google.
If Microsoft and Meta are better investments, this isn't good for Google irrespective of how much GOOG went up.
Google can produce cool scientific research but it is basically IBM of this era.
Please note: I think it is clear that the Gemini models still need work to be historically representative and accurate, however, I do not think there is any real danger in airing with the side of caution and trying to fight the innate biases in the training data. It seems like this was Google's goal.
I ran this by my 70-year old immigrant parents—who are Democrats and by no means culture-war people—and they found this very cringeworthy (not that they even know what “cringe” is—that’s the closest translation of their reaction in English).
Do they know you're sharing personal info?!
If he is in any way annoyed or harmed by my sharing it, I hope he will let me know.
Could you please provide evidence of this being google goal? From everything I have seen, from the posts of google employees who worked on it, to Geminis own responses, they were being racist against white people. In fact considering google retains so much power over your average person I believe it would be considered systemic racist.
To claim it is being overblown to me sounds like you are trying to say this form of racism is okay.
World Economic Forum pushes it with a list of huge companies that align with their initiatives. They bragged even movie makers do that like how they put LGBT and feminism into as many movies as possible, and hire some directors whose only success was feminist movies. Meta has DEI pages, a page giving money to feminists to push their theories, and their writing on alignment looks like others I just mentioned.
So, yeah, there’s definitely a culture war going on. We had values, policies, etc that worked well. These companies often existed and got successful because of that kind of thinking. Starting in liberal universities (often cultural Marxism), people with different concepts are getting into positions of power. These concepts wrecked many prior countries but their proponents want to force them on everyone else.
Every now and then, you see clear evidence of what they do leak into the media. In this case, some of those racists at Google probably programmed the AI to reduce white males or increase those in favored groups (“diversity”). Then, it started doing the same kinds of things its creators would promote in mass media, but in a dumber, obvious way.
Nobody should depend on AI’s designed by people who may hate or seek to minimize specific groups. God created everyone in His image, told us to love them all, there’s “neither Jew nor Greek” now, and justice is rewarding any who work for that reward. People need to drop this stuff in favor of non-discriminating practices.
to back up my claim, id cite a few of those incel culture stories Google has had over the last decade suggesting they have some of that "libertarian" strain indistinguishable from fascism or "states rights" ideology.
Forbes was pretty famous for being able to manipulate for SEO purposes.
All that's happening here, it seems to me, is that this episode is casting a bright light on the fact that the cultural values of (a) inclusive representation, and (b) avoiding "harmful" representations of protected categories, are mutually contradictory. But nobody's even attempting to acknowledge this possibility except the idiot "go woke go broke" crowd, which is not exactly a good faith discussion.
It can be many problems, but I think it is certainly an AI problem. A big one. It really shows that asking a computer for something is fundamentaly a bad idea. Computers are for, well, computing. They are not for answering history questions or whatever.
No, that is not what it shows at all. Everything you do on a computer is "asking a computer for something." Do you think computers are computing things that no one asked for?
Straight from the horse’s mouth :)
> I said they didn't want to be racist, what they haven't yet figured out is: 1) how to not be racist, while aligning with 2) how to be anti-racist.
> They've also confused "racial" with "racist".
Just because "people" want a thing doesn't mean the outcomes will be acceptable. Google is supposed to be excellent at figuring out this gap, with SMEs and product managers.
And to add to all this, it feels like an undesirable extension of the "ethical AI team" debacles (search for Timnit Gebru and Margaret Mitchell) of a couple of years ago. They haven't solved this internal conundrum yet, and the current example seemed like an easy one to foresee.
Was it easy to mention in meetings? Especially if you weren't from a minority background.
I may be part of that idiot group then
Edit: spelling
So, the engineering team saw previous complaints about image generators providing too "stereotypical" images in some contexts and pushed the generator more towards returning "diverse" results. But Gemini isn't a smart human, it's a blunt machine and so it doesn't know in which contexts "diversity" is appropriate.
I think the underlying issue here is that until these AI tools genuinely are smarter than smart humans (and probably even after), the second they get released they're going to be under massive scrutiny. And no person, let alone some crappy AI is going to give perfect answers to 1,000,000 users asking questions attempting to trip it up.
The solution to this is obvious though - you accept it's not perfect! The problem is that's not an acceptable solution for either Google's brand or Google's shareprice.
I dont think you're crazy, but while "woke" vs "anti-woke" may have been the catalyst, the issue is really that AI can fail pretty spectacularly, and the law of unintended consequences when applied to AI is at best, funny, and at worst, extremely harmful.
Google did exactly what their own perception of cultural attitudes led them to do.
This model clearly isn’t exactly in tune with the actual cultural zeitgeist, otherwise they wouldn’t have this issue.
IMO it doesn’t seem aligned with broad cultural attitudes - it appears in line with a specific cultural attitude (one often called ‘woke’ although I don’t think that’s a particularly useful term).
But the “prevailing cultural attitudes” don’t demand that—as evidenced by the mockery at Gemini’s expense. One of the highest rated shows on TV right now is “Young Sheldon,” where every single character is a white southerner. That’s a spin-off of Big Bang Theory, which reached #1 in its timeslot, in which every character except the token Indian was white. And how many Asians do you see on Univision? It’s a very small portion of the public that’s clamoring for contrived scenes where there’s one white, one Latino, one black person, and one Asian around a dinner table.
https://www.rollingstone.com/tv-movies/tv-movie-features/osc...
Or what I wrote in my other comment:
https://news.ycombinator.com/item?id=39530386
The example you give may have some problem driving it. Likewise, all the shows and channels whose cast are almost all black, Latino, women, LGBT, etc. It’s also possible the roles just went to whoever they’re around the most or showed up. I’ve also been in groups that were mostly white who invited countless non-whites to visit or join. They just never did for who knows what reason.
All that said, it’s anecdotal evidence. What’s more concrete is the huge number of companies, plus awards like Oscars, changing their criteria to require hiring non-whites, non-males, and non-straights. One criteria required the plot be designed to promote that group’s politics which often happens. The Mother being the last one I saw like that and director turned out to do movies to push feminism. Also, who you have sex with used to be irrelevant to hiring. Now, liberals reward LGBT practices with jobs and movie roles.
They’re definitely doing it. They don’t have to but are just backing those like them. Their ideological partners in groups they favor. The way human corruption and sin has always worked, eh?
[1] https://stratechery.com/2024/gemini-and-googles-culture/
I doubt the patched model will fix the underlying issue. They'll just apply exceptions to a few noteworthy historical keywords. They've been vocal about their filtering of image search for years and this is in line with that.