Three years after Google sounded alarm bells over ChatGPT, the tables have turned.
See full article...
See full article...
OpenAI CEO declares “code red” as Gemini gains 200 million users in 3 months
- Thread starter JournalBot
- Start date
Do they count "users" including people who had it jammed down their throat kicking and screaming making their phone less useful for voice-commands and people cursing at the wrong AI answers at the top of Google searches?
Just yesterday I was trying to learn about Cisco wildcard masks vs subnet masks and it told me 0.255.255.255 is a 16 bit mask eyeroll
I'm stuck with it but I hardly consider myself a Gemini user and wish it didn't exist.
Just yesterday I was trying to learn about Cisco wildcard masks vs subnet masks and it told me 0.255.255.255 is a 16 bit mask eyeroll
I'm stuck with it but I hardly consider myself a Gemini user and wish it didn't exist.
Upvote
16
(17
/
-1)
Daily reminder that the millionaires are playing against billionaires higher up the food chain than them. They're also living paycheck to paycheck and gradually bankrupting themselves to keep up and most of them are average intelligence or worse trapped in wealth bubbles.
You're literally doing the entire globe a favor and improving the 1%'s quality of life and emotional health by taxing AI firms and most executives at 73% with no loopholes regardless of how you spend it afterward... would break the financial/credit ouroboros pretty fast in a way that doesn't have to result in a bubble bursting into stagflation.
You're literally doing the entire globe a favor and improving the 1%'s quality of life and emotional health by taxing AI firms and most executives at 73% with no loopholes regardless of how you spend it afterward... would break the financial/credit ouroboros pretty fast in a way that doesn't have to result in a bubble bursting into stagflation.
Upvote
11
(11
/
0)
@MoonShark
The getty images source link just links back to arstechnica. Maybe a copy paste error or redirect issue as it contains the link at the end?
The getty images source link just links back to arstechnica. Maybe a copy paste error or redirect issue as it contains the link at the end?
Upvote
2
(2
/
0)
If I hear "Code Red" then I'm going to grab a 10lb CO2 bottle and spray it into wherever the flames seem to be coming from.
LLMs have their place. Domain-specific AI tools have their place. But I really don't see the financial fundamentals to justify the investment bubble that's grown up around them recently. And I'm getting a bit worried about the economic impacts if the bubble implodes – or, perhaps worse, the economic impacts if they sorta-kinda succeed well enough to put a lot of people out of work and the bubble doesn't implode.
Upvote
13
(13
/
0)
“Gemini 3 after using ChatGPT daily for three years. “I’m not going back,” Benioff wrote. 'The leap is insane.'"
Game, set… match
OpenAI opened strong serving innovation and riding the crest of the wave it started. Arguably OpenAI, the company, lost cohesion - lost CEO, key contributers and was forced to reassemble.
GOOG is not a bunch of software devs in their first rodeo. Google owns the internet sech space - lock, stock and eyeballs.
SteveJobs proved over ten years development that in technology there only ever exists room for two. In the OS wars back in the day, Microsoft and Apple prevailed.
Today AI's opportunity is to paradigm shift that duopoly off the brand of OS. That’s what is at stake. GOOG are not only the biggest elephant in this circus, Google has the best infrastructure, revenue model and software development resources to entertain, amaze and drive folks to it big tent!
Personally, as a dev, AI looks very much pre-dotcom crash. Lots of integration. Too lttle innovation, Few AI product. Marketing is the tail wagging right now.
BEST in class currently is IBM genetic AI sequencing, drug interaction, discovery and diagnosis by individual genome.
OpenAI fallibility appears to be its obsession on AI and wagging the dog rather than innovation to which Salesforce Marc Benioff alludes
Game, set… match
OpenAI opened strong serving innovation and riding the crest of the wave it started. Arguably OpenAI, the company, lost cohesion - lost CEO, key contributers and was forced to reassemble.
GOOG is not a bunch of software devs in their first rodeo. Google owns the internet sech space - lock, stock and eyeballs.
SteveJobs proved over ten years development that in technology there only ever exists room for two. In the OS wars back in the day, Microsoft and Apple prevailed.
Today AI's opportunity is to paradigm shift that duopoly off the brand of OS. That’s what is at stake. GOOG are not only the biggest elephant in this circus, Google has the best infrastructure, revenue model and software development resources to entertain, amaze and drive folks to it big tent!
Personally, as a dev, AI looks very much pre-dotcom crash. Lots of integration. Too lttle innovation, Few AI product. Marketing is the tail wagging right now.
BEST in class currently is IBM genetic AI sequencing, drug interaction, discovery and diagnosis by individual genome.
OpenAI fallibility appears to be its obsession on AI and wagging the dog rather than innovation to which Salesforce Marc Benioff alludes
Upvote
4
(6
/
-2)
It does seem like the people who get really into ChatGPT have a narcissistic streak and either love it being a sycophant who tells them their dumb ideas are genius or they fall in love with it
Upvote
17
(17
/
0)
They also have the issue that, given the seemingly critical importance of more data and more follow-up training(getting user feedback on responses; along with having humans do annotation and similar on things like Scale AI's 'remotasks'); basically all the most valuable customers have less than zero reason to trust them.
Sure, at least if you buy through MS(an EA with commercial data coverage; not the consumer stuff) the language says that your data will not be used 'to train foundation models'(luckily there aren't things in the cloud black box that aren't classified as 'foundation models'...); but unless you just want to cheat on your homework or do bad customer service on the cheap your attitude toward trusting Sam being within grabbing range of any interesting data(inputs, usage, or both) ranges from 'deeply concerned' to 'aww hells no'.
It's not clear that random orgs really want to get into doing HPC, or foundation model training; but the nakedly voracious attitude of the major 'AI' players makes them so untrustworthy that they seem likely to be at major risk from 'fast follower' types who aren't quite as good; but are willing to sell, or just give, you their weights so that you can keep things internal/run on a leased GPU compute colo whose operator just does care and feeding rather than full 'cloud' blackbox.
Upvote
4
(5
/
-1)
Worry less about an idealized water-sucking grass lawn? Or, edge it strategically.
I had an entire front "lawn," albeit a small one, of salvia rosmarinus prostratus. No dog instrusions, minimal irrigation needed, plus ample supply for roasted meats and potatoes
. My own dogs were well capable of handling damage in the back yard.
Upvote
13
(13
/
0)
Thanks for saying it! I’m getting so tired of these AI puff pieces that amount to nothing more than an unmarked advertisement with NO real journalism at play. Real journalism would have meant questioning the marketing of a CEO to figure out why he said it, if there’s some financial motivation or if there’s actually a feature that makes his quote make sense
Is Ars’ conflict of interest interfering with its journalistic integrity? Because this is bad.
Forget This Meeting Could’ve Been an Email. Now we have This Article Could’ve Been a Press Release
Upvote
-5
(4
/
-9)
I'm personally glad to see Altman panicking. I've never been a fan of him, ChatGPT or OpenAI. I can't imagine anything truly good will come of this.
Upvote
12
(12
/
0)
I respect Ars for reporting the information with enough flavor to allow our critical thinking skills to handle the editorializing. Or not, for those preferring to be spoon fed.
Upvote
10
(11
/
-1)
Maybe he should declare a code red over the looming verdict in court that will see them on the hook for potentially a trillion dollars in damages for theft of IP to train their models in the first place.
Upvote
22
(22
/
0)
First, it's an off-hand remark posted to social media by a tech CEO; those are hardly worth the paper they're printed on (if you even bothered to print it out).
Second, it's ChatGPT, not Chat GTP.
Upvote
5
(5
/
0)
Open AI is currently under a lawsuit from a group of authors and publishers for stealing content. If they are successful, Open AI will lose their training database as it is tainted with stolen data, and could get fines as high $150k per infringement. I would think that would bankrupt them.
Upvote
17
(17
/
0)
I know, right. Wow, the CEO of Salesforce has a brainwave/opinion about AI. So glad that got shared with the world and is now taking up a few bytes on a server somewhere until the end of time. Wtf is he even asking it?
"Hey Gemini, can you please improve the grammar of my useless word turds, help me think of 14 creative new ways to suck up to Elon Musk, and confabulate some sycophantic praise for my half-baked opinions before I compulsively
Upvote
7
(7
/
0)
Yes it is, and I would bet they will have a lawsuit that may make them de-integrate it. I certainly don't want it on my phone, yet I can't delete it, only disable.
Upvote
5
(5
/
0)
I agree to an extent.
But to be honest, for articles like these, I’m mostly here for the comments bashing the dork CEOs.
Upvote
7
(7
/
0)
For Google, Microsoft, Meta etc which run other businesses they can just state that 50% of their profitability is due to AI integration in existing shit. Thus rendering all AI efforts completely profitable and "worth it". OpenAI/ChatGPT can't quite as easily do the same in order to "show profits".
Upvote
6
(6
/
0)
I thought Palmer Luckey was unpleasant but this guy makes him look like the mid-boss he is
Upvote
6
(6
/
0)
He ordered the code red cuz "he can't handle the truth!"
This is shaping up very much like the American railroad era - late 1800 through early 1900s where railroads required massive upfront investments that bankrupted many startup attempts. The difference is that railroads provided a need - people were excited about the ability to travel longer distances in shorter time...it was transformational to our society. No one is asking for AI and all we ever get is bluster and BS about some day in the future. It provides minimal value. Most people won't use it if they have to pay for it, so it will be another product that has to be stuffed with ads in hopes that the upfront investment can be paid down by the 23rd century.
This is shaping up very much like the American railroad era - late 1800 through early 1900s where railroads required massive upfront investments that bankrupted many startup attempts. The difference is that railroads provided a need - people were excited about the ability to travel longer distances in shorter time...it was transformational to our society. No one is asking for AI and all we ever get is bluster and BS about some day in the future. It provides minimal value. Most people won't use it if they have to pay for it, so it will be another product that has to be stuffed with ads in hopes that the upfront investment can be paid down by the 23rd century.
Upvote
14
(14
/
0)
Humanity is beyond stupid. We will not survive "AI" since it is based on the lowest common denominator......
Upvote
2
(2
/
0)
Here's an idea: maybe be a normal human being and accept that dogs sometimes walk on your yard? That's not entitlement, it's just normal behavior. A dog is not 'trespassing', they don't even have the concept of personal property. Have you ever walked a dog? Do you know how hard it is to stop a dog from walking less than a foot to one side and entering a yard?
What harm is that even doing to you? You're treating it like some thief is trying to break onto your property. Maybe there's no good AI answer because your question is inane, not because AI is a broken piece of crap (which it is).
Upvote
14
(18
/
-4)
Upvote
12
(14
/
-2)
I'm getting ~youtuber insincerly apologizing for saying a slur from that photo of Scam Altman.
Upvote
1
(1
/
0)
Upvote
14
(14
/
0)
I haven’t studied finance or investing but it seems pretty clear that even they don’t really think it will improve a lot if they are planning to put ads into it.
Upvote
6
(6
/
0)
So, I have Gemini Pro, and many, many times when I use Google Search, there's an AI summary at the top that is dead wrong, and if I click the button at the bottom of the AI result to explore it further, it then puts my identical query into Gemini Pro, and delivers a different result, that's usually correct.
Yes, the default "AI Answer" on Google searches is not a good advertisement for Gemini. But I guess they can't afford the processing resources for a Gemini Pro answer on every query.
Upvote
-2
(2
/
-4)
Sigh. I do use ChatGPT. That said, the user interface for ChatGPT is absolutely awful. I can see that being low hanging fruit to improving the user experience. I point this out, because, I doubt Google has backburnered the interface. It's not sexy but it's meaningful.
ChatGPT's interface truly looks like a wrapper on an API and that shouldn't be disregarded. In other, other words, it's hard for me to understand how a company so "valuable" can put so little into this. It shouldn't be a huge expense for them and it shouldn't be low on the list of needed improvements... but it clearly is.
I think they have been having parties. And this is part of their plan to accumulate money and cleanly burn it all down. Nicely done Sam! /s
ChatGPT's interface truly looks like a wrapper on an API and that shouldn't be disregarded. In other, other words, it's hard for me to understand how a company so "valuable" can put so little into this. It shouldn't be a huge expense for them and it shouldn't be low on the list of needed improvements... but it clearly is.
I think they have been having parties. And this is part of their plan to accumulate money and cleanly burn it all down. Nicely done Sam! /s
Upvote
2
(2
/
0)
Your observations are on target!
Tell me more about that? I think you’re onto something here!
Upvote
-1
(2
/
-3)
Similarly, that "800 million weekly users" number from OpenAI that gets mentioned uncritically all the time has seemed like, maybe, totally made up as well.
Upvote
12
(12
/
0)
“Altman encouraged temporary team transfers and established daily calls for employees responsible for enhancing the chatbot.”
——
So they’re going to put all their current projects on hold, reorganize their teams, and make everyone attend [additional] daily meetings. This is how they improve productivity.
That screams competent leadership. Remind me again why CEOs make all the money?
——
So they’re going to put all their current projects on hold, reorganize their teams, and make everyone attend [additional] daily meetings. This is how they improve productivity.
That screams competent leadership. Remind me again why CEOs make all the money?
Upvote
20
(20
/
0)
Everyone with any sense - myself very much included - was pointing out literally years ago that these companies do not have products, they barely even have features. What they make are basically just components. LLMs (and the other model types OpenAI et al produce) have no moat. One LLM is mostly fungible with another. Unless they had some hidden secret magic that could make their own model unassailably better (which they did not), they were always going to be eaten alive by the platform holders, who have existing (and far better) profit centres they can pay for the infrastructure (or at least better offset the cost of it) with.
And what's happening? Google is swamping ChatGPT with Gemini. Microsoft has OpenAI over a barrel financially, ready to swoop in for a big chunk of whatever's left once OpenAI inevitably collapses under the crushing weight of its debt and other gargantuan financial burdens. Apple have apparently concluded the high performance compute reliant frontier models are already a solved problem and are happy to just license Google's for their own service, at least for the time being - which rather proves the point that the things are fungible components, not end user products.
Nobody (other than weird nerds like Benioff, who let's be honest are mostly lying - quite likely to themselves as well to be fair) actually cares what model is being used. They're all basically the same. People will, by and large, if they want to use a chatbot, just use the one that already fits in their workflows and ecosystems. Models are fungible, so frontier model companies are worthless.
Nvidia are certainly making out like bandits right now, but they’re very exposed to the bullshit on their balance sheet too. Plus, their share price is predicated on the idea they can not only sustain those astronomical sales figures but also maintain this unnatural growth rate.
You know that thing about how surprisingly few levels it takes for a pyramid scheme to require more people than there are on earth? It’s not quite as dramatic, but in a similar vein Nvidia’s financial situation is going to run up against the limits of free capital very quickly.
Rarely in the history of free enterprise has such irrational exuberance been sustainable for anyone involved, however good their results in the short term. Once all the liquidity that can be soaked up by these GPUs (many of which will be sitting idle for years because do you have any idea how expensive and longwinded it is to actually build a high performance compute data centre?) has been soaked up, Nvidia are going to be in it up to their necks in “complex financial instruments.”
And what's happening? Google is swamping ChatGPT with Gemini. Microsoft has OpenAI over a barrel financially, ready to swoop in for a big chunk of whatever's left once OpenAI inevitably collapses under the crushing weight of its debt and other gargantuan financial burdens. Apple have apparently concluded the high performance compute reliant frontier models are already a solved problem and are happy to just license Google's for their own service, at least for the time being - which rather proves the point that the things are fungible components, not end user products.
Nobody (other than weird nerds like Benioff, who let's be honest are mostly lying - quite likely to themselves as well to be fair) actually cares what model is being used. They're all basically the same. People will, by and large, if they want to use a chatbot, just use the one that already fits in their workflows and ecosystems. Models are fungible, so frontier model companies are worthless.
Yes, well, so is the 800 million "weekly active" (weird, they use weekly when everyone else uses monthly, wonder why that could be) ChatGPT users, so...
I wouldn’t be so sure about this. Nvidia’s finances on this are…Weird. They keep doing things like “selling” GPUs to NeoClouds only to rent them back and helping customers who are high performance compute providers of various sizes secure loans against their arsenals of GPUs which those providers then turn around and spend on…More Nvidia GPUs.
Nvidia are certainly making out like bandits right now, but they’re very exposed to the bullshit on their balance sheet too. Plus, their share price is predicated on the idea they can not only sustain those astronomical sales figures but also maintain this unnatural growth rate.
You know that thing about how surprisingly few levels it takes for a pyramid scheme to require more people than there are on earth? It’s not quite as dramatic, but in a similar vein Nvidia’s financial situation is going to run up against the limits of free capital very quickly.
Rarely in the history of free enterprise has such irrational exuberance been sustainable for anyone involved, however good their results in the short term. Once all the liquidity that can be soaked up by these GPUs (many of which will be sitting idle for years because do you have any idea how expensive and longwinded it is to actually build a high performance compute data centre?) has been soaked up, Nvidia are going to be in it up to their necks in “complex financial instruments.”
Upvote
20
(20
/
0)
My father smoked for 70 years and every day quit smoking repeatedly.
Upvote
-5
(1
/
-6)
Advertising comes with questions. Are advertisers willing to pay what they need to for an LLM to be profitable ?
Will advertisers be happy with the content the LLM spits out near their ads ?
Will they like how the LLM mangles their ads ?
What other legal problems will the run into ?
Upvote
7
(7
/
0)