This is a very good point.
I never had the discussion whether real AI, not transformer-based chatbots, would be a boon or a bane for human workers.
I mean, we should already have data about it.
By ‘real AI’, I presume you mean AGI, a digital intelligence that is actually superior to human intelligence, ie, is more intelligent than the smartest human and has all our collective knowledge and is able to comprehend it and evaluate it more consistently than any of us… and also is thus capable of improving itself and becoming more and more superintelligent.
That is still scifi, that is not real.
What we currently call ‘AI’ is basically an extremely expensive, lackluster pantomime of that, that fools fools into thinking it is the former thing… mostly because it is sycophantic and very confident, ie, it uses well known ‘hacks’ in human psychology, where confidence, breadth of knowledge, usage of technical terms… you know, con man techniques … are confused for actual competence.
If we had a real AGI, it would be be capable of both hacking into all the military information systems of the world, and tricking humans into nuking each other… and it would also be capable of making actual novel improvements in software, hardware, engineering, physics, social engineering, etc, and could decide to be a kind of benevolent dictator of the entire economy, that it would command and control.
We have no capacity to model the morality that would emerge in an actual superintelligence, because we definitionally would not be able to keep up with attempting to understand how it thinks.
Thats where the whole ‘is AI the potential best thing ever or would it become SkyNet’ problem comes from.
… But we are not there yet.
We are at… basically, a very fancy autocomplete algorithm that can analyze huge datasets reasonably well, compared to an average human, but also makes all kinds of mistakes, hallucinates ‘facts’ in order to generate more coherent things to say, and these hallucinations routinely trick non subject matter expert humans into just going along with it, again, like a con artist, like a fast talking ‘influencer’ pitching selling you a course or giving you some kind of ‘advice’.
And currently, what is going on, is that we are pouring I think at this point trillions of dollars into ‘AI’, under the premise that it is AGI, that it will be capable of generating massive returns on investment and productivity increases…
… but the actual results are turning out to be, all averaged for everywhere it has been implemented… somewhere between a net productivity loss, to meagre productivity gains.
What that means is that the AI Mania is the biggest bubble, the most severe malinvestment of economic resources in the history of humanity.
When that pops, we basically formally transition into cyberpunk dystopia, technofeudalism.
AI is a tool, a device, a machine. Thus, it depends on how you use it, what you use it for.
Right now, we have a whole lot of companies saying they are laying off workers because we don’t need them anymore… this is broadly a lie.
People are being laid off because the economy, the real economy, is already contracting, basically due to the collapse of the US as the undisputed world hegemon.
AI, as a broad, socioeconomic force… is mostly a smokescreen, the ultimate promise of bread and circuses, that masks a gigantic wealth transfer and restructing of economic and political power.
AI as a tool can be used for good, in specific use cases.
But it broadly isn’t, because people are fooled by the conversation machine into thinking it can do things that there is no evidence it can do, because people do not understand its limitations and flaws, and then they plug it into their immensely shitty business processes, and just assume it will not break things when it tries to use them.
biofaust@lemmy.world 4 weeks ago
This is a very good point. I never had the discussion whether real AI, not transformer-based chatbots, would be a boon or a bane for human workers. I mean, we should already have data about it.
sp3ctr4l@lemmy.dbzer0.com 4 weeks ago
By ‘real AI’, I presume you mean AGI, a digital intelligence that is actually superior to human intelligence, ie, is more intelligent than the smartest human and has all our collective knowledge and is able to comprehend it and evaluate it more consistently than any of us… and also is thus capable of improving itself and becoming more and more superintelligent.
That is still scifi, that is not real.
What we currently call ‘AI’ is basically an extremely expensive, lackluster pantomime of that, that fools fools into thinking it is the former thing… mostly because it is sycophantic and very confident, ie, it uses well known ‘hacks’ in human psychology, where confidence, breadth of knowledge, usage of technical terms… you know, con man techniques … are confused for actual competence.
If we had a real AGI, it would be be capable of both hacking into all the military information systems of the world, and tricking humans into nuking each other… and it would also be capable of making actual novel improvements in software, hardware, engineering, physics, social engineering, etc, and could decide to be a kind of benevolent dictator of the entire economy, that it would command and control.
We have no capacity to model the morality that would emerge in an actual superintelligence, because we definitionally would not be able to keep up with attempting to understand how it thinks.
Thats where the whole ‘is AI the potential best thing ever or would it become SkyNet’ problem comes from.
… But we are not there yet.
We are at… basically, a very fancy autocomplete algorithm that can analyze huge datasets reasonably well, compared to an average human, but also makes all kinds of mistakes, hallucinates ‘facts’ in order to generate more coherent things to say, and these hallucinations routinely trick non subject matter expert humans into just going along with it, again, like a con artist, like a fast talking ‘influencer’ pitching selling you a course or giving you some kind of ‘advice’.
And currently, what is going on, is that we are pouring I think at this point trillions of dollars into ‘AI’, under the premise that it is AGI, that it will be capable of generating massive returns on investment and productivity increases…
… but the actual results are turning out to be, all averaged for everywhere it has been implemented… somewhere between a net productivity loss, to meagre productivity gains.
What that means is that the AI Mania is the biggest bubble, the most severe malinvestment of economic resources in the history of humanity.
When that pops, we basically formally transition into cyberpunk dystopia, technofeudalism.
AI is a tool, a device, a machine. Thus, it depends on how you use it, what you use it for.
Right now, we have a whole lot of companies saying they are laying off workers because we don’t need them anymore… this is broadly a lie.
People are being laid off because the economy, the real economy, is already contracting, basically due to the collapse of the US as the undisputed world hegemon.
AI, as a broad, socioeconomic force… is mostly a smokescreen, the ultimate promise of bread and circuses, that masks a gigantic wealth transfer and restructing of economic and political power.
AI as a tool can be used for good, in specific use cases.
But it broadly isn’t, because people are fooled by the conversation machine into thinking it can do things that there is no evidence it can do, because people do not understand its limitations and flaws, and then they plug it into their immensely shitty business processes, and just assume it will not break things when it tries to use them.
biofaust@lemmy.world 4 weeks ago
No, by real AI I meant what we already had before and that now is “disregarded” as machine learning algorithms.
I know it is not the correct technical definition, but AI is mostly a marketing term anyway.