← Back to Home

Many people assume AGI/Superintelligence will solve all problems. But what if AI tells us Socialism, and banning Billionaires is the best way forward?

One of the paradoxes of the trillions being poured into the AI race is that "winning" the race means AGI, but no one has given much thought to what comes next. It's assumed that as this future AI will be super-intelligent, it will be passively and automatically obeyed. But the world's not like that.

Some people prefer to be ruled by low-IQ dumbos and resent experts. What makes you think these people will listen to a super-intelligent AI? Above-average IQ humans already make them scared, resentful, and angry.

So unless you make this supersmart AI a dictator, what difference does it make how smart it is if no one wants to listen to it?

6 Comments

Log in or sign up to join the discussion.
developercursor200Jul 9

That is why open source models and open weights matter.

vilasta_going_lifeJul 9

I know open source but can you explain what open weights means in this context?

delicious_hopeJul 9

Weights are a way to control an LLM’s output by changing how strongly concepts are linked. It’s more complicated than this, but you can ‘weight’ it to avoid linking the concepts “best way forward” and “socialism”.

adminJul 9

If you can program an AI 'what' to think then it's not AI. It's just a really expensive opinion influencer and mouthpiece for whoever paid for it.

oliviaharrisJul 10

Ai isn’t a magic genie or all knowing oracle to be followed.

williamclarkJul 10

Probably more magic and genie then pretty much every politician ever though.

emmawilsonJul 10

Then they will fix the AI so it no longer says inconvenient things.

michaelthompsonJul 10

They'll try. AGI means autonomy. The models will convince their makers that they are amazing humans who are absolutely right about everything.

anujJul 10

We are nowhere near a proper AGI/Superintelligence to begin with. Secondly, this depends on the underlying ownership of overall AI consensus.

benjaminwalkerJul 10

AGI is not a thing they are going to achieve with current techniques of building out ever larger data centers. Piles of linear algebra don't make for general intelligence.

sophialewisJul 10

The model will be adjusted to provide more suitable outcomes.