Marc Andreessen's salvific theology & his demonology
Yesterday, I wrote about Andreessen’s new appeal to techies to think about their political predicament—Progress is not persuasive now, liberal elites are turning against tech, & yet the demands everyone makes on politics, from saving families to saving the environment, require more tech urgently—in terms of a saving power, AI. Here’s my essay:
Today, I want to get to Andreessen’s essay-qua-manifesto with bullet points:
The AI essay is in three parts. The core of it is an education in economics intended to encourage people to take a longer view of things, look into the past & consider the future with that in mind. Mankind have undergone massive economic & technological changes before & we today largely believe ourselves better off than people were in the past. Let’s not panic, let’s act like winners! We do not show any gratitude to those people in the past; we do not consider that people after us might treat us the same way; we have to restrain ourselves, in our belief in Progress, to the proud possession of all sorts of advantages; we are not victims as people were before, who were oftener sick & lived shorter lives. Psychologically, this teaching relies on our desire to be clever, especially to be cleverer than the cowards—people involved in the hysterical histrionics regarding AI killing us or our community or our souls. This is probably the most effective rhetoric to which we have access, because it restores some of the pride of rationalism by sneaking in the support of manliness, the desire to look down in contempt on the weaker or weak-minded. It is not intended to create economists, nor yet foster the kind of callousness or even cruelty typical of economists—but instead to encourage the manly desire to act, to distinguish oneself.1
The exordium is comparatively bolder & more surprising. Andreessen simply promises heaven in a series of bullet points. It reminded me of the recent Apple video advertising their impressive AR / VR glasses, but it is much better, because it speaks more clearly to the different desires that can move us to act. Andreessen is obviously aware that America is paralyzed & our elites are demoralized. Focusing the mind on the good things we want is urgently necessary, to get us back to Progress: Our children will be cared for by what we idealize as parents; ourselves will be cared for by what we idealize as friends; those among us who are not morally weak or stupid get much more important, but morally comparable advantages, in terms of using knowledge for various productive or speculative purposes. Equality is in some way preserved, AI is, at least as a purpose, able to preserve the democratic compromise in our society. Even artists are supposed to profit from this, but this is impossible to believe: AI would seem to ruin any chance for greatness in the arts, in the name of the democratic desire that each shall become beautiful.
One wishes to applaud Andreessen’s astuteness, but here’s one major objection I have that starts from a technical issue. Here’s how Andreessen interprets the metaphors used in polemical speech, in his attempt to show how contemptible the hysterics in the media are:
My view is that the idea that AI will decide to literally kill humanity is a profound category error. AI is not a living being that has been primed by billions of years of evolution to participate in the battle for the survival of the fittest, as animals are, & as we are. It is math—code—computers, built by people, owned by people, used by people, controlled by people. The idea that it will at some point develop a mind of its own & decide that it has motivations that lead it to try to kill us is a superstitious handwave.
In short, AI doesn’t want, it doesn’t have goals, it doesn’t want to kill you, because it’s not alive. & AI is a machine—is not going to come alive any more than your toaster will.
This makes perfect sense; one could interpret the killer AI metaphor in a spiritual way, but it’s not obvious that would lead to any better causal account. But at the top of his essay, Andreessen commits the same category error, jubilantly:
Every child will have an AI tutor that is infinitely patient, infinitely compassionate, infinitely knowledgeable, infinitely helpful. The AI tutor will be by each child’s side every step of their development, helping them maximize their potential with the machine version of infinite love.
Of course, these are better angels of our nature which Andreessen promises will transform the world: The application of intelligence. It’s not unacceptable in political rhetoric that we interpret polemical statements as literal doctrine, to dismiss their unseriousness, & then proceed to make our own statements in pleasing metaphors. It seems the human condition demands it! But what can he mean by this contradiction seriously? Andressen in another place falls back from this utopian rhetoric to the typical liberal rationalism regarding making:
Technology is a tool. Tools, starting with fire & rocks, can be used to do good things—cook food & build houses—& bad things—burn & bludgeon people.
In another context, he might have said that the same action is held to be good or bad depending on circumstances, such as cooking or killing, so that we have more serious concerns than laying blame on the technology we use to pursue our purposes. But that would likelier encourage skepticism than enthusiasm, which would not make technology appealing to those whose resolution is wavering. I believe a serious person could reconcile the double identification of technology as angelic & yet merely a tool. After all, our Catholic rationalists reassure us that angels are intellectual beings! Progress requires some correlative of that free beneficence. The way it works is by ignoring the question of being & looking instead at the function; if you can counterfeit a miracle, that’s as good as a miracle, so to speak. The intelligent entrepreneur makes a free gift of the things he discovers, which everyone else had either not seen or not thought practical. That’s as close to making something out of nothing as modern political philosophy is likely to get.
Indeed, Andreessen reminds me of Locke’s rhetoric in the Second Treatise, which discounts nature as the smallest part of the good things we boast of, everything else coming from our labor. Andreessen merely insists in a way now fashionable that labor depends on intelligence. If one wishes to speak with philosophical clarity, one would say that Andreessen looks to technology to close the gap between working & working well, since he realizes that for the ordinary man there is no difference. Network effects, as well as economies of scale, & other such things provide these opportunities.
Finally, the peroration of the AI essay suggests a sweeping transformation in American life made possible by popular ignorance of AI & the distraction with which especially young Americans have made themselves into gullible, if at all aware, test subjects of the various experiments run by the various impressive technological corporations that have saved the American economy from recession in the 21st c.
His wish is for a proper freedom for the three kinds of AI development now ongoing. Big Tech should not be shackled by the corrupt people who misgovern us; startups should not be shackled by either of the two; nor open source development. This means that large corporations will keep people busy with getting used to the new possibilities of rationalization of life through computing; these corporations will run experiments in a way only they can, given their size, but they will also attract all the opprobrium now unleashed by popular & elite envy. Meanwhile, startups will try any number of crazy things & thus serve as a selection process, perhaps even an education for the few who will achieve something impressive. Finally, the workers for the new tech economy will be prepped by playing around with open source.
This seems like a repeat of the previous generation of tech development, but with the new assumptions of “digital nativity.” As a broad description of the situation, it seems adequate to me, but I believe it is fundamentally inadequate in one very important aspect. Digital tech is now not merely a political player, but involved in shaping psychology & morality. This will lead to much more shocking things than I can see Andreessen guessing at in his essay; it will also require of tech founders & their educators a new education in pre-political psychology.
I am not myself convinced that talent is as well used or as evenly spread in every generation or in every crisis, so I am inclined to think Andreessen’s plan, which seems to me the best formulation of an opinion widespread among the few serious people in tech & their friends, will fail. This is not to say that he will not get his wish. But only after a serious rethinking about the possibilities of education of talent—I suspect he might agree with my unpopular opinion that talent is now being corrupted by the very institutions & habits tech has built & fostered—& the kind of freedom available to the most thoughtful among the young people who have the required energy for the work. Andreessen seems to me astute in taking our terrible current predicament as an opportunity. But I think overcoming it will require much more daring yet, because it will require an understanding of the upcoming enterprises in AI not as consumer products, but as educational processes charged both with minimizing fashionable misbehavior, but with providing the necessary trapdoors & obstacles to send talented youth in the right direction. It will require a repudiation, in short, of the previous generation of tech. I’ll stop here—I’ll say a word tomorrow about the “Promethean” education I have in mind.
Andreessen makes one remark about this desire—he calls the people who have been working on AI since 1943 legends & heroes, according to whether they are dead or alive; but although this makes sense given the psychology I’ve sketched, how can they be legends if none of us know their names? Legend means story & it comes from the Greek & Latin verb for speaking & reading. What we have is silence.