Artificial intelligence won’t end civilization, argues Marc Andreessen. Just the opposite. It is quite possibly the best thing human beings have ever created.
Like all technology and scientific discoveries, AI is morally and ethically neutral. Its application is not, and humans have a habit of converting instruments of progress into antisocial or criminal devices or weapons of war. We regularly turn plowshares into swords. AI may be neutral, but we're not. I fear the uses that AI will be put to. To quote Shakespeare's Cassius, "The fault, dear Brutus, lies not in our stars but in ourselves.
“Every child / person will have an AI tutor / assistant / therapist ” and so forth -- this has to be satire. No one can be that tone-deaf or that ignorant of the need for human connection and relationships.
I’ve worked as a engineer and researcher in machine learning (now we’re calling it AI) for the past decade and I want to fact check this.
> AI is a computer program like any other—it runs, takes input, processes, and generates output.
True. But he omits that the rules are encoded in 1 billion to 1 trillion parameters and that no human truly understands *how* it processes that input to create an output.
> This moral panic is already being used as a motivating force by a variety of actors to demand policy action
True. Sam Altman is definitely a bootlegger. Most of the policy proposals from Sam Altman are designed to build himself a moat.
> My view is that the idea that AI will decide to literally kill humanity is a profound category error.
I find it unlikely that ape-derived brains will still dominate the earth in ten thousand years. We’re safe from existential risk for now though.
> My response is that their position is nonscientific. What is the testable hypothesis? What would falsify the hypothesis?
Red Teaming Deep Neural Networks for Trojan Re-discovery is a promising direction. I think it can be extended to reinforcement learning.
When did it become a conspiracy theory? This is a straw man argument to make anyone who disagrees with him look dumb.
> AI is going to improve warfare, when it has to happen, by reducing wartime death rates dramatically
Now he’s just messing with us.
> There is a whole profession of “AI safety expert,” “AI ethicist,” “AI risk researcher.”
I write under a pen name, but my day job is not about safety. It’s about profit. There are definitely risks.
> They are paid to be doomers, and their statements should be processed appropriately.
Andreesen as an investor is paid to be an optimist and his statements should be processed appropriately.
I wrote up a piece on Forecasting Authoritarian Uses of Large Language Models if anyone is interested [https://taboo.substack.com/p/authoritarian-large-language-models]. I think the immediate risk in the next generation or two is just the same as always. Power hungry tyrants using new technologies to control us. And investors making misleading statements minimizing the risks of these technologies.
Given the amount of information and intelligence available in todays world and the level of understanding that too many of the younger generation have or don't have it is doubtful A.I. will penetrate the obtuse minds of these "educated" kids. Many of them don't even know the timeline of recent history or any of the basic building blocks of an intelligent life. They seem immune to acquiring knowledge or information. Just interview the next graduating class at a major University and try getting an answer to basic questions. Good luck. A.I. will become a tool of the elites and government to consolidate their power and control. It will affect the general population but without their consent or knowledge. And so it goes.
"It is quite possibly the the best thing human beings have ever created." - Please tell me that AI did not write the title of this article with a typo. :)
Do you not think something is already lost in our humanity as people wander around with eyes glued to their smart phones never bothering to look at the world around them. The internet is so addictive that some people never interact with others-much less nature. People are brainwashed. My feeling is that we are already less as human beings and this directly relates to technology. As for AI? I see it as another nail in the coffin of humanity. I hope I’m wrong.
This fool actually typed that with a straight face.
He seriously considers that long list of bullet points "the positive argument" FOR this nonsense! Every thing he envisions and looks forward to, I believe any sane human should regard as an absolute nightmare.
A. If it was created by man, then mankind can pervert it (Biden and COVID anyone?) and use it against the average normal human being.
B. The ability to use for spying and selecting those you disagree with for punishment, reeducation, or even elimination is too great to be trusted. Maybe even aborting those babies that the superior elite determine should not be allowed to live. Or monitoring your consumption, usage of goods, or spoken resistance, and determine you are a problem. Think not, well can you say China!
C. Facebook and Ticktock would never collect information to use against you right? Suckenberg can be trusted, just look at that face and tell me you don't trust him.
How does AI help the average person get a decent job and be able to feed their family and just live a free life? How would it help end the endless wars in Africa and the hunger and famine in the world? I distrust anyone saying they know its the best thing ever for mankind. You don't know shit until it all plays out!
When I read about the attorneys who relied on AI to write a legal brief, and AI made up case law to argue a point, and the attorneys submitted it to the court and opposing side without checking it, and the judge and opposing attorneys asked to see the full citations for the case law, weelllllllll now....the attorneys wiped egg off their faces and are having to pay a fine. A few more stories like this, and AI’s value will be seriously damaged.
Won't it matter who does the programming? If you ask AI to sort out issues in the gender wars, won't its answers reflect the ideological input of the programmers? The same thing for having it tutor in history for example. And even in supposedly scientific fields like medicine where it will push certain kinds of therapies determined by the scientific orthodoxy. That is already happening with gender affirming care for dysphoria. It is also happening in the realm of so-called scientific journals.
So from where I sit, your view of AI would create a frightening and far-reaching totalitarianism. I hardly consider that making the world better.
Responding to this article - do I believe him, or my lying’ eyes? Humans everywhere, medicated with prescribed or illegal drugs, criminals given free range to operate at will, homeless and mentally ill choking once beautiful urban cities, taxes rising on those working to achieve the waning American Dream while masses of less fortunate or less industrious live on subsidized incomes that will never lift them out of poverty. AI will only increase the fortunes of the Uber-rich and powerful, while subjugating , intoxicating and suppressing compliant underlings. Would gladly eat my words, but like my eyes, throat not cooperating either.
Garbage in garbage out. I’m a natural optimist but this was over the top. Unfortunately many of the “bad guys” are our leaders and elites. We have to pursue AI responsibly but there will be serious downsides.
Like all technology and scientific discoveries, AI is morally and ethically neutral. Its application is not, and humans have a habit of converting instruments of progress into antisocial or criminal devices or weapons of war. We regularly turn plowshares into swords. AI may be neutral, but we're not. I fear the uses that AI will be put to. To quote Shakespeare's Cassius, "The fault, dear Brutus, lies not in our stars but in ourselves.
"Create a place for people to live like human beings, instead of slaves to some bullshit concept of Progress that is driving us all mad."
— Hunter S. Thompson
“Every child / person will have an AI tutor / assistant / therapist ” and so forth -- this has to be satire. No one can be that tone-deaf or that ignorant of the need for human connection and relationships.
I’ve worked as a engineer and researcher in machine learning (now we’re calling it AI) for the past decade and I want to fact check this.
> AI is a computer program like any other—it runs, takes input, processes, and generates output.
True. But he omits that the rules are encoded in 1 billion to 1 trillion parameters and that no human truly understands *how* it processes that input to create an output.
> This moral panic is already being used as a motivating force by a variety of actors to demand policy action
True. Sam Altman is definitely a bootlegger. Most of the policy proposals from Sam Altman are designed to build himself a moat.
> My view is that the idea that AI will decide to literally kill humanity is a profound category error.
I find it unlikely that ape-derived brains will still dominate the earth in ten thousand years. We’re safe from existential risk for now though.
> My response is that their position is nonscientific. What is the testable hypothesis? What would falsify the hypothesis?
Red Teaming Deep Neural Networks for Trojan Re-discovery is a promising direction. I think it can be extended to reinforcement learning.
https://arxiv.org/pdf/2302.10894v2.pdf
> a conspiracy theory about math and code
When did it become a conspiracy theory? This is a straw man argument to make anyone who disagrees with him look dumb.
> AI is going to improve warfare, when it has to happen, by reducing wartime death rates dramatically
Now he’s just messing with us.
> There is a whole profession of “AI safety expert,” “AI ethicist,” “AI risk researcher.”
I write under a pen name, but my day job is not about safety. It’s about profit. There are definitely risks.
> They are paid to be doomers, and their statements should be processed appropriately.
Andreesen as an investor is paid to be an optimist and his statements should be processed appropriately.
I wrote up a piece on Forecasting Authoritarian Uses of Large Language Models if anyone is interested [https://taboo.substack.com/p/authoritarian-large-language-models]. I think the immediate risk in the next generation or two is just the same as always. Power hungry tyrants using new technologies to control us. And investors making misleading statements minimizing the risks of these technologies.
This piece needs to be followed by Mr Andreesen’s statement on conflicts of interest. Those who control and profit from AI are going to control us.
Given the amount of information and intelligence available in todays world and the level of understanding that too many of the younger generation have or don't have it is doubtful A.I. will penetrate the obtuse minds of these "educated" kids. Many of them don't even know the timeline of recent history or any of the basic building blocks of an intelligent life. They seem immune to acquiring knowledge or information. Just interview the next graduating class at a major University and try getting an answer to basic questions. Good luck. A.I. will become a tool of the elites and government to consolidate their power and control. It will affect the general population but without their consent or knowledge. And so it goes.
"It is quite possibly the the best thing human beings have ever created." - Please tell me that AI did not write the title of this article with a typo. :)
Do you not think something is already lost in our humanity as people wander around with eyes glued to their smart phones never bothering to look at the world around them. The internet is so addictive that some people never interact with others-much less nature. People are brainwashed. My feeling is that we are already less as human beings and this directly relates to technology. As for AI? I see it as another nail in the coffin of humanity. I hope I’m wrong.
"the machine version of infinite love..."
This fool actually typed that with a straight face.
He seriously considers that long list of bullet points "the positive argument" FOR this nonsense! Every thing he envisions and looks forward to, I believe any sane human should regard as an absolute nightmare.
A. If it was created by man, then mankind can pervert it (Biden and COVID anyone?) and use it against the average normal human being.
B. The ability to use for spying and selecting those you disagree with for punishment, reeducation, or even elimination is too great to be trusted. Maybe even aborting those babies that the superior elite determine should not be allowed to live. Or monitoring your consumption, usage of goods, or spoken resistance, and determine you are a problem. Think not, well can you say China!
C. Facebook and Ticktock would never collect information to use against you right? Suckenberg can be trusted, just look at that face and tell me you don't trust him.
How does AI help the average person get a decent job and be able to feed their family and just live a free life? How would it help end the endless wars in Africa and the hunger and famine in the world? I distrust anyone saying they know its the best thing ever for mankind. You don't know shit until it all plays out!
That’s a lot of kool-aid. With Kamala Harris as the AI czar what could possibly go wrong?
AI in the hands of a big and growingly intrusive government.
What could possibly go wrong?
When I read about the attorneys who relied on AI to write a legal brief, and AI made up case law to argue a point, and the attorneys submitted it to the court and opposing side without checking it, and the judge and opposing attorneys asked to see the full citations for the case law, weelllllllll now....the attorneys wiped egg off their faces and are having to pay a fine. A few more stories like this, and AI’s value will be seriously damaged.
Won't it matter who does the programming? If you ask AI to sort out issues in the gender wars, won't its answers reflect the ideological input of the programmers? The same thing for having it tutor in history for example. And even in supposedly scientific fields like medicine where it will push certain kinds of therapies determined by the scientific orthodoxy. That is already happening with gender affirming care for dysphoria. It is also happening in the realm of so-called scientific journals.
So from where I sit, your view of AI would create a frightening and far-reaching totalitarianism. I hardly consider that making the world better.
Responding to this article - do I believe him, or my lying’ eyes? Humans everywhere, medicated with prescribed or illegal drugs, criminals given free range to operate at will, homeless and mentally ill choking once beautiful urban cities, taxes rising on those working to achieve the waning American Dream while masses of less fortunate or less industrious live on subsidized incomes that will never lift them out of poverty. AI will only increase the fortunes of the Uber-rich and powerful, while subjugating , intoxicating and suppressing compliant underlings. Would gladly eat my words, but like my eyes, throat not cooperating either.
Garbage in garbage out. I’m a natural optimist but this was over the top. Unfortunately many of the “bad guys” are our leaders and elites. We have to pursue AI responsibly but there will be serious downsides.