3 ms·

Marc Andreessen: “AI Will Save the World”

2 points by fortran77 3y ago

rossdavidh 3y ago
As I recall, Marc Andreessen has been quite positive on blockchain, as well. This doesn't mean the current crop of "AI" will be more like blockchain than it is like the internet, but it does mean that Marc Andreessen being enthusiastic about something, mostly just means he is a technophile and this is the new tech buzzword.
- jjtheblunt 3y ago
  That's definitely polite. In the general case: Might $FamousTechPersonality also be described as publicly enthusiastic about things that he has invested in already, so he feigns certainty as a mind game to get others to invest themselves, to his earlier investment's benefit?
noduerme 3y ago
This sets up a really weak straw man argument on the other side. There are many valid concerns about this technology well short of AI robots killing people. >> What AI offers us is the opportunity to profoundly augment human intelligence He then lists a bunch of tutoring and assistant applications, with the explicit assurance that they will be infinitely patient and intelligent. Nit: So far, patience for questioning does not seem to be a hallmark of LLMs. But really? What machine aid or technological assistant - or even human teacher or assistant - has ever augmented intelligence? Knowledge, maybe. Efficiency, for sure. But human intelligence? Reading something in a book doesn't increase intelligence unless you understand it. Learning how to do it from basic principles is more likely to increase overall intelligence than reading it. Having a teacher guide you through reading it or an assistant help you to make it does nothing whatsoever to increase your intelligence, if intelligence is measured by your capacity to think your way out of an unknown box. In fact, everything that exists to help you solve puzzles without having to think for yourself of the best way to tackle a problem is a lost opportunity for exercising your intelligence. But we're not talking about life hacks or game cheats here, where the stakes are low. We're talking about basic reasoning. Andreessen is touting a technology that would (if it were patient, and didn't hallucinate or become psychotic when pressed for proof of its assertions), intercept children and "augment" their lives by giving them answers without the crucial experience of discovering any answers for themselves. He mistakes knowledge, or access to knowledge, for intelligence. Intelligence is the native or learned skill of being able to seek knowledge, to reframe problems, to discover for oneself. If invention is proof of intelligence, and necessity is the mother of invention, what intelligence can there be where necessity is eradicated? Nowhere in the history of our species has existing technology done anything other than facilitate incremental, iterative improvement. The actual exercise of true intelligence requires the absence of an existing "test harness". It's the development of the harness for testing an hypothesis that requires exercising intelligence.

2 points by fortran77 3y ago

As I recall, Marc Andreessen has been quite positive on blockchain, as well. This doesn't mean the current crop of "AI" will be more like blockchain than it is like the internet, but it does mean that Marc Andreessen being enthusiastic about something, mostly just means he is a technophile and this is the new tech buzzword.

jjtheblunt 3y ago
That's definitely polite. In the general case: Might $FamousTechPersonality also be described as publicly enthusiastic about things that he has invested in already, so he feigns certainty as a mind game to get others to invest themselves, to his earlier investment's benefit?

noduerme 3y ago

This sets up a really weak straw man argument on the other side. There are many valid concerns about this technology well short of AI robots killing people. >> What AI offers us is the opportunity to profoundly augment human intelligence He then lists a bunch of tutoring and assistant applications, with the explicit assurance that they will be infinitely patient and intelligent. Nit: So far, patience for questioning does not seem to be a hallmark of LLMs. But really? What machine aid or technological assistant - or even human teacher or assistant - has ever augmented intelligence? Knowledge, maybe. Efficiency, for sure. But human intelligence? Reading something in a book doesn't increase intelligence unless you understand it. Learning how to do it from basic principles is more likely to increase overall intelligence than reading it. Having a teacher guide you through reading it or an assistant help you to make it does nothing whatsoever to increase your intelligence, if intelligence is measured by your capacity to think your way out of an unknown box. In fact, everything that exists to help you solve puzzles without having to think for yourself of the best way to tackle a problem is a lost opportunity for exercising your intelligence. But we're not talking about life hacks or game cheats here, where the stakes are low. We're talking about basic reasoning. Andreessen is touting a technology that would (if it were patient, and didn't hallucinate or become psychotic when pressed for proof of its assertions), intercept children and "augment" their lives by giving them answers without the crucial experience of discovering any answers for themselves. He mistakes knowledge, or access to knowledge, for intelligence. Intelligence is the native or learned skill of being able to seek knowledge, to reframe problems, to discover for oneself. If invention is proof of intelligence, and necessity is the mother of invention, what intelligence can there be where necessity is eradicated? Nowhere in the history of our species has existing technology done anything other than facilitate incremental, iterative improvement. The actual exercise of true intelligence requires the absence of an existing "test harness". It's the development of the harness for testing an hypothesis that requires exercising intelligence.