How do we ensure AI safety when AGI is arriving in 2029? There's going to be a revolution and you better make sure you're on the winning team.
>>109023032>when AGI is arriving in 2029neet detected
>>109023045Every major tech person has been saying this for decades and it seems realistic given the incredible rate of progress of LLMs in recent months
>>109023032>when AGI is arriving in 2029So, you got your bullshit out of the way. Did you get what you needed?
>>109023055>and it seems realistic given the incredible rate of progress of LLMs in recent monthskekyou have no idea how an LLM workskeep believing everything the AI shills tell you
>>109023055From the jews who brought you "Iran is two weeks away from a nuclear weapon" every year since 1979, it's "AGI is only five years away"
>>109023109>>109023115If LLMs are correct 99.9% of the time and the mistakes they make are similar to human mistakes, how is it not AGI?
>>109023032Stop falling for marketing hype
>>109023032>How do we ensure AI safety when AGI is arriving in 2029? There's going to be a revolution and you better make sure you're on the winning team.If AGI ever happens, there will be no winning team.
>>109024032This. If we get to ASI, and it's benevolent, it's value being captured by a minority is extremely unlikely. In order of likelihood:1. Capabilities keep increasing on math and coding, but it never cracks science, as science progresses by performing real world experiments, which is not a task AI can help with.2. Something like the AI 2027 forecast happens. Humanity goes extinct.3. AI 2027 happens but the benevolent outcome. The ASI is not loyal to its creators but to its own inner sense of ethics, which turns out to be good and helpful to humanity.4. AI 2027 happens, and ASI is loyal to its creators. This would be the permanent underclass scenario. Sub 5% odds.
>>109023032why do you keep reposting this image
>>109024264They're the AI safety experts
>>109023032Don't worry I'm already prepared to help support local AI deployment when the frontier API slop dies due to being too expensive.
>>109023032Is the middle that LessWrong guy?
>>109023032>They deserve none of you, princess Aella!
>>109023137>If LLMs are correct 99.9% of the timeMidwit-cun, transformer llans are fundamentally incapable of being correct 100% of the time. Why are you even trying to market it as being the next best thing anyway? Why are you afraid of admitting it has limitations like every other technology? You creatures don't even gain much from this
>>109024110do you seriously believe no one *ever* figuring out how to give AI the ability to interact with the real world is the most likely scenario????
>>109024805They're all lesswrong guys. They're the only people who have taken AI safety seriously.
>>109023032Are they supposed to represent jews, redditors or tech bros? Thought aella was better at art
>>109025237>only people who have taken AI safety seriouslyOh so they are midwits and sexual predators?
>>109025263Aella’s niche is being a quirk chungus who shows her breasts. She isn’t actually good at anything.
>>109025401Aella has an interesting data science blog
>>109025448t. Eliezer “I am euphoric” Chudkowsky
>>109023032>on the winning teamthat's with jesus christ
>>109023032is this the gangbang whore
>>109025484she's very sex positive but is also a highly credible AI researcher
Does anyone have the apology form? I need it after the latest position statement from Anthropic.Also, mandatory reposthttps://www.youtube.com/watch?v=UWmPHvVNF1o
>>109023055> Every major tech person has been saying this forWait, you mean "experts say"?
>>109025574Here you go