/biz/ - Business & Finance

>geeky lanky dork gets the spotlight for 5 minutes over stupid trend

Gee i wonder whats next

major dev conference at end of the year. nothing matters much up until that point, that's where they'll announce it

>>58115791
guy doesn't do shit, his researchers do. And they'll not crack AGI using retarded LLMs so good luck with that

>>58115791
to continue to trick redditors and normies who think computers are magic and think that LLMs will demonstrate human-level intelligence any time soon

>>58115828
>>58115851
Super weird samefag.
>they just won't do it... OK!

He'll hand over the keys to OpenAI to the Elders of Zion for their final trick

>>58115810
>Literal fucking technological singularity
>stupid trend
ok boomer

>>58115895
I work in the valley and studied ai/ml for 6 years so I know what I'm talking about. these are all probalistic models..but I'm still making money off it so who gives a fuck. It IS hype though and a scam but only like the top 5% of programmers/researchers can determine this

>>58115970
GPT 4 has been pretty fucking useful for my startup, literally bootstrapped months of research and programming and then some into a few weeks with a handful of people. We went from some meme lean startup to a meme zero to one, we're just doing the finishing touches now. Their API is a fucking godsend.

Should I get myself a bag of Worldcoin? I saw one of his orbs scanning people the other day. This dude is gonna rule the world

>>58115970
Once again...
>it just won't work because..... its just "probalistic", OK! Trust me, I've been a programmer for 6 whole years!

>>58116014
yeah that sounds about right. i mean i think it's cool and useful (hence the hype) but ive found it really hard to build shit with outside of scaffolding a project

github copilot is even worse with its crappy random type fix suggestions that simply don't work because the feedback loop for typescript actually includes what package version is available/if a bunch of packages are out of date the typing for a project will be broken

there's just still a ton of manual human intervention you have to do that you can either hire a jeet to do or do yourself

>>58116057
more like 15 years of actual engineering experience and 6 years of research simultaneously, then realizing ai was gay late 2018 because its so compute constrained and you just become an adhoc data scientist in any real job so i moved on to building businesses/actually trying to make money

>>58116165
Didn't ask. You're like a chemical engineer trying to give input on if nuclear fusion will be acheived soon.

>>58116165
you fucked up the timing and now are resentful toward LLG breakthrough huh

>>58115791
Where did he even come from? Before openai he was basically known for running the scam that got reddit out of its deal with those tarnnies from that shit web publishing place.

>>58115791
I tried watching the lex interview, this guy has vocal fry and such an affect that it makes me hate his guts.

>>58116271
He is the stereotypical homosexual with an overprotected mother.

>>58116271
Sorry anon, but "beta nerds" have created the technological society we live in now

>>58116242
>>58116224
watch the yann lecun interview w lex fridman

>>58115791
AGI waifus pls SAMA

>>58115791
release whisper 4 with even more hallucinations than whisper 3, then change his name to josh, pierce his nipples and build the world's greatest waterslide park

>>58115791
I asked chatgpt for a list of words that end in “eth” for the ethereum name service and it gave me meth, beth, and a whole list of words like death, breath etc. gpt is getting stupider not smarter. I was gonna cancel my sub.

>>58116165
>2018
That’s the equivalent of 100 years ago in terms of AI advancements retard

>agi by 2030
>asi by 2040
>longevity escape velocity by 2045
all conservative estimates.

>>58116714
sorry, all AGIs are going to be put on biotech research plantations so that we can make Sam immortal. you get an AI waifu after that, maybe.

>>58118114
Nope, you are fucking 100% wrong. we STILL don't have any actual image classification algorithms (a simple task like image classification) that are beating humans and we've had 10 years of advancements. Something as simple as "is this image similar to this one?" or "what is in this image?" that beats a human. Most of this is shit that existed 10 years ago and the improvements have not been exponential. We went from 0-70% in 2010-2014 for image recog, then 70-75% 4 years later, then 75-80% 2 years later but nothing beating humans (this is NOT exponential growth)


but redditors like you with no idea who don't even even know the basics will say agi will be here in the next decade and that LLMs which are based around supervised learning, gradient descent and attentional mechanisms (much like how image recognition is based around convolutions + attention) will say we are close to agi

K

>>58115946
>dude spellcheck is going to change the universe
I'm going to assume you are actually retarded and need a computer to make bad decisions for you

dunno who tf this guy is but i'll assume he's an important figure in the crypto space or a content creator that is slightly influential in some way
if he has a lot of money and influence, i hope he puts it on loaf cat and promotes it so i can make my x5 into a x10 soon enough and take my deserved vacations in europe