This reeks of desperation. Almost as if, AGI claims have always been bullshit. Now, they are just scrambling to find a use case.
The only true use case I've found for LLMs is generating acceptable bullshit.
When I needed to let a vendor know that we were not going to renew the contract, I didn't want to have to use my brain power to come up with the business-speak version of "piss off", so I had copilot write the first draft.
It's excellent at bullshit. I'm not sure they can recoup their investment with that. Maybe if they start replacing all the C-suite folks with AI across all industries, it could make a small dent.
They're useful for parsing large documents quickly. I upload a manual for something I'm unfamiliar with and troubleshoot with it.
Doesn't replace me because you still have to know when it's full of shit but it's great for skipping the first 80% of a task.
I like it but I would never pay what it actually takes to run the thing
The only useful thing I've seen come out is my dm being able to put his ideas into fruition quicker and on the fly, everything else has pretty questionable utility from what I've seen
That's great! How much is he paying for that? 'Cause the AI industry needs to recoup a few billion dollars, and every little bit helps.
A few billion dollars? They need to make 2 trillion dollars to make a profit. That's bigger than most economies.
Trust me, I never said it was worth the cost to society lol
I think it has a lot more utility than a lot of people here give it credit for. It's just very easily misused.
Edit: I certainly wouldn't want it to be my 'OS'
LLMs are great at retrieving information while taking a natural language query and synthesizing a natural language response. In essence, it is a glorified search engine, that is better than keyword-based search, but not as great /revolutionary as advertised. So, yes, it has utility. But, people like Altman and most other silicon valley execs had to overhype the damn thing because they only care about pumping their stocks.
Of course, needs to factor in the cost of power, infrastructure and environmental destruction. That depends on the how did LLMs scale with rise in these inputs? But, everyone is more concerned with outspending everyone else, in hopes that they will magically reach Narnia (I mean, AGI), but it might as well be Narnia. The result: a fucking bubble.
If they had kept their and everyone else's expectations in check, this would not have happened. But, hey, line must go up, at any cost.
some places with the datacenters have seen thier electricity bill increased as a result.
Imagine people paying for the 'profit' of the corporations. Socialism for me, capitalism for thee.
Now I get why they can't come out of the closet
Haha they don't have even have stocks! They may be legally barred from going public and nobody will want to but them, in which case they'll be fucked (but Altman will walk away with bags full)
they are know they are losing money, nvidia investing 100billion seems like a stopgap measure.
nvidia is investing 100 billion in OpenAI so that OpenAI can spend it all on GPUs from nvidia. Sounds like a totally healthy industry
Don't dismiss this nonsense.
Someone once told me the browser would be the platform of the future for running applications sometime in the mid 90's and I dismissed the guy because he was a BS-talking marketdroid - and also because the idea was completely idiotic on its face. Yet here we are...
We tend to remember the hits and forget the misses. They said we'd all be plugged into VR and riding around on Segways, too, etc. Those things settled into mature, but minor, technologies, and no doubt genAI will too--but that's not going to generate enough revenue to justify the out of this world valuations of OpenAI, et. al.
Please no…
I'm not saying it's desirable, I'm saying don't dismiss it because stupid shit happens when enough stupid people with money want to make it happen. And Sam Altman is loaded.
I know, and it’s fucking scary…
Just today I saw this post on Mastodon, and cannot fathom how they’ll want to run entire apps flawlessly inside LLMs, and at what ecological cost!?
LOL, I give up, I can't even figure out how it got 15.
Easy. You got 1 and 2, which is obviously 12. Then you add 3, because it is a sum, so 15 comes out. Don't forget to like and subscribe!
Almost works in JS. +("1" + 2) + 3 is 15, and "1" + 2 - -3 is 15, but "1" + 2 + 3 is "123". 🙄️
I'm still not convinced Javascript is a serious language rather than a prank.
At a guess, when people ask it to "sum the numbers above", they usually test it on the sequence 1,2,3,4,5. It's an LLM, it's doesn't process its input, it returns one of the most probable tokens based on what it's seen before. If it actually becomes a "thing", crashing the global economy is the least of our worries.
If it actually becomes a "thing"
It is actually already a thing: https://support.microsoft.com/en-us/office/copilot-function-5849821b-755d-4030-a38b-9e20be0cbf62
Also, see this article from last week: https://www.theverge.com/news/787076/microsoft-office-agent-mode-office-agent-anthropic-models
Microsoft says its Agent Mode in Excel has an accuracy rate of 57.2 percent in SpreadsheetBench, a benchmark for evaluating an AI model’s ability to edit real world spreadsheets.
Hmm... -.-
There's a "it's in the product" thing, and a "people actually, seriously, use it fir actual work" thing. We've got the first, I'm hoping that enough people get burnt by it being wrong, in non-serious ways, that noone tries to use it seriously. Hope and expectation are different things though. sigh
Its added 4 and 5 also. I think its solving a trianguar number pattern: n(n+1)/2. These things are in maths tests all the time where you need to find the next two in the sequence
Im curious if a =COPILOT formula gives the same results on that sheet today as it would next year after the LLM has changed with the extra input. Will it reassess that every time the sheet is opened and there is an internet connection?
Nondeterministic spread sheets
Excel never automatically recalculates. Even if you use RANDOM it will still be the same. But touch the cell and it will be different
Oh no
We've been trying to abstract hardware since... C. We've had much better virtual machines, but they never catch on.
Adoption is a feature you can't design.
But for LLMs digging any deeper than they already have, lol no. Microsoft bet the farm and demanded a whole new keyboard key. People see it as an unreliable convenience at best. It's not getting any better until after the bubble pops.
We’ve been abstracting away hardware details since the invention of punchcards. “Assembly code” is a remarkably high level abstraction above microcode, which is a remarkably high abstraction above logic gate arrays.
He’s still kinda wrong though. With the exception of corporate desk jobs, the vast majority of computing is done on phones/tablets these days and on those platforms apps are still king.
Well, it's how the personal terminals in star trek are used most of the time. They don't even have keyboards, not only in the cabins but also the one in Picard's ready room.
On TNG only the nerdiest nerd of all, so much of a nerd to be an android, had a computer with a physical input interface in his cabin, Data.
ChatGPT... the everything app? I've heard that one before.
The thing companies don't seem to understand is that the reason why "everything" apps are so popular in the Global South is because people in the Global South dont have the space on their phones nor the resources to have 10 different apps for ten different services so having one service do it all is more convenient for them.
They keep trying to replicate it in the Global North is because having one app to do it all would make whatever company that manages it very very rich, but they fail to consider the other factors, as always.
Thats makes absolutely zero sense. At this point it is only marketing nonsense.
The danger to OpenAI here is people will see that and then start to question whether some of their precious insane claims are also just marketing nonsense.
The man behind the curtain should stop humming and tapping his foot if he doesn’t want people looking.
Hey ChatGPT, where are the project files for the client I’ve been working on?
Now you have to remember to save and backup your homework.
With AIOS™©® you have to make sure your homework is kept in context, look at a hallucinated Facebook page for too long and all your work is gone.
And on this server farm we used all the water, AI AI OS
With a drought drought here and a drought drought there...
Please process my payments for me, too.
And my online shopping in general.
No problems here, nuh huh.
Perfect Simpsons reference. Sick.
Tell him to go discover the Black Marker and promptly sink into the ocean already.
I won't be around for the Unitologist takeover anyway.
I just think they're neat!
