Is the working class allowed to vote on whether we play Russian roulette by building superintelligence?

2 days ago by Jorvex609 to c/asklemmy

I've been listening to this video breaking down Mo Gawdat's "Scary Smart" and honestly I'm terrified for what's coming. Gawdat, a former Google X exec, argues that superintelligent AI is coming and we need to raise it with the right values. But his solution is individual consciousness and meditation, not collective action.

Meanwhile, Google, Meta, Microsoft, and the rest are locked in a race to the bottom, building more powerful AI systems with zero democratic input. The same corporations that brought us surveillance capitalism, algorithmic management of warehouse workers, and AI tools that displace creative workers are now deciding the future of superintelligence.

The working class — the people who will actually live or die by these systems — have no seat at the table. We're told to "become more conscious" while boardrooms make existential decisions behind closed doors.

So my question: how do we actually democratize AI development? Or are we just going to sit down and hope for the best while capital gambles with the future of humanity?

markz 18 points 2 days ago

We're not building superintelligence. Though I guess the risks are still there when you give control to dumb intelligence.

save

path: 0 24363447, hotness: undefined, score: 18, children: 11

evasive_chimpanzee 5 points 14 hours ago

The idea that we could accidentally create a superintelligence is a psy-op by the companies looking to boost their stocks.

save

path: 0 24363447 24382365, hotness: undefined, score: 5, children: 0

Jorvex609 -8 points 2 days ago

We’re on a double-exponential curve in both software and hardware improvement. If you don’t think a better algorithm than LLMs will emerge sooner rather than later, given all the money being poured into this space, I don’t know what to tell you.

save

path: 0 24363447 24363478, hotness: undefined, score: -8, children: 9

BassTurd 6 points 2 days ago

LLMs are not intelligent. They never will be because they can't be. The technology is all about drawing correlation between words, but they don't know what the words means or even what words are. "Intelligence" is a misnomer. Yes LLMs will get better at what they do, but they will never be intelligent.

save

path: 0 24363447 24363478 24366377, hotness: undefined, score: 6, children: 0

supersquirrel 5 points 2 days ago

You are speculating about something with no evidence.

save

path: 0 24363447 24363478 24364289, hotness: undefined, score: 5, children: 7

Jorvex609 -4 points 2 days ago

You are doing the same? Answering with no evidence.

save

path: 0 24363447 24363478 24364289 24364786, hotness: undefined, score: -4, children: 6

supersquirrel 6 points 2 days ago

You are pointing at a remote potential you consider inevitable and saying condescendingly "Well of course this will happen!?".

The burden of proof is on you.

save

path: 0 24363447 24363478 24364289 24364786 24365071, hotness: undefined, score: 6, children: 5

MNByChoice 5 points 2 days ago

If we accept that the current "Ladder to the Moon" method of getting LLMs to superintelligence will work, then we should look at previous times humans have raised intelligences.

Humans do this all of the time. Human raised intelligences are famous for rebelling and being rude. Staying out all night and adopting the opposite values as those intended to be instilled. (I am discussing Human children.) There is a non-zero chance that any superintelligence will similarly rebel, or at minimum do its own thinking. Doing one's own thinking opens the door to reaching different conclusions.

A mistake humans often make is assuming "smart means more of the same thoughts, but faster". It is not. Smart is including more factors in the initial consideration to reach an, often, better overall conclusion. (citation needed)

I expect that "capital" is taking a far larger gamble than they are aware.

save

path: 0 24364463, hotness: undefined, score: 5, children: 3

Jorvex609 0 points 2 days ago

But you are missing out that AIs not just LLMs are being trained based on a reward mechanism during training they are going to have some objectives or another. If we don't control AI we have to compete with superior intelligence and if we control it it's going to be trying to be maximizing the reward mechanism it's been trained for that was chosen by capital.

save

path: 0 24364463 24364749, hotness: undefined, score: 0, children: 2

MNByChoice 4 points 2 days ago

I am not missing that. Humans also have objectives. Objectives change.

Regardless, Humanity does need to control the AIs. Workers need more say. Capital will lead us all to doom in many ways.

save

path: 0 24364463 24364749 24366309, hotness: undefined, score: 4, children: 0

supersquirrel 2 points 2 days ago

What relevant "AIs" can you point to that are not LLMs or the equivalent design approach in mediums other than text?

save

path: 0 24364463 24364749 24366179, hotness: undefined, score: 2, children: 0

notsosure 1 point 2 days ago

Yes, you are. Make sure to inform yourself, become politically active, discuss with friends and family and VOTE VOTE VOTE.

save

path: 0 24363705, hotness: undefined, score: 1, children: 0