The Case For AI Optimism
Marc Andreessen says AI will Save the World. He's not wrong.
Fortunately, I am here to bring the good news: AI will not destroy the world, and in fact may save it. - Marc Andreessen
This Marc Andreessen declaration, that AI will save the world, is an essay well worth reading. I will summarize it, break it down and comment on it, but it is important and well-written and worth a full read on its own as well.
This essay is a helpful sequel to Andreessen’s 2011 essay that “Software is Eating the World,” where he explained the pervasive power of software to change our world. Now comes the next iteration of technology acceleration with AI. As the next major step in technology progress, he sees AI as powerful, important and hugely and mostly beneficial.
AI is quite possibly the most important – and best – thing our civilization has ever created, certainly on par with electricity and microchips, and probably beyond those. - Marc Andreessen
It’s also an important rejoinder to the AI Doomers and AI worry-warts, who in Andreesen’s words have instigated a ‘moral panic’ that has over-emphasized the risks of AI, as well as the AI Skeptics who discount the value of AI.
AI Augments Human Intelligence
He is firmly in the AI Optimist camp when he describes AI as possibly “A way to make everything we care about better,” and expresses many benefits and opportunities in the era of AI.
He places AI in the arc of human achievement by noting that our human intelligence has been the technology level that raised our standard of living this far, and AI is now an augmentation of that:
What AI offers us is the opportunity to profoundly augment human intelligence to make all of these outcomes of intelligence – and many others, from the creation of new medicines to ways to solve climate change to technologies to reach the stars – much, much better from here.
By explaining AI as a technology that augments human intelligence, instead of merely displacing it or competing with it, he sees it a lever and magnifier. There will be “magnification effects of better decisions by leaders,” more science progress by scientists who “greatly expand their scope of scientific research,” as the productivity of all intellectual work will go up.
“Productivity growth throughout the economy will accelerate dramatically, driving economic growth, creation of new industries, creation of new jobs, and wage growth, and resulting in a new era of heightened material prosperity across the planet.”
In short, augmenting human intelligence with AI will magnify intellectual productivity, accelerating all intellectual endeavors and speeding technology progress. As we have written about earlier about “The Singularity is Inevitable,” the outcome of this will be an acceleration in technology progress that feeds on itself, as AI itself rapidly evolves and improves.
Scientific breakthroughs and new technologies and medicines will dramatically expand, as AI helps us further decode the laws of nature and harvest them for our benefit. - Marc Andreessen
This heightened productivity will mean not just technology and science accelerating, but an even more humanizing effect, as there will be an expansion of creative arts and human creativity.
The creative arts will enter a golden age, as AI-augmented artists, musicians, writers, and filmmakers gain the ability to realize their visions far faster and at greater scale than ever before. - Marc Andreessen
Rebutting the Risks of AI
He directly rebuts these AI risks, specifically these 5 categories of concern:
“AI will kill us all” is a myth and a 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, and 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 and decide that it has motivations that lead it to try to kill us is a superstitious handwave.
The AI rising to kill us makes for great Science Fiction and is long-standing in culture, but as I’ve noted before, AI will only have the agency that we will give it. It will be optimized to be useful to us, not optimized for its own power.
‘hate speech and misinformation’ and AI Safety and Alignment
The real issues regarding AI Safety and AI Alignment are whether the outputs of AI can be helpful or harmful. Beyond obvious limits against AI promoting illegal actions or inflicting actual harm, there are grey areas of what harm means. Who is to judge?
He draws lessons from the experiences with social media platform content censorship to conclude that any system to regulate AI output will evolve to “demand ever greater levels of censorship and suppression of whatever speech they view as threatening to society.”
AI Alignment and Safety ‘experts’ are typically elites who presume to know better than others what “good for society” means. He describes this progressive world-view as a niche morality, and any attempt to impose it faces resistance in the form of a healthy desire for free speech and a diverse world that does not all align on those concerns and cultural biases.
If you don’t agree with the prevailing niche morality that is being imposed on both social media and AI via ever-intensifying speech codes, you should also realize that the fight over what AI is allowed to say/generate will be even more important – by a lot – than the fight over social media censorship. AI is highly likely to be the control layer for everything in the world.
In this last sentence, he is letting slip why AI is deemed as such a risk. AI’s power to help us, including teaching and informing us, is AI’s power to influence all we believe. To control AI is to control what people believe.
We should default towards free speech and openness on AI models and platforms, not because there won’t be harms, but because that’s a far lesser harm than a world of central controlled and censored AI. He puts it bluntly: “In short, don’t let the thought police suppress AI.”
Will AI Take All Our Jobs? No!
He argues that, contrary to the “Lump Of Labor Fallacy,” we have unlimited demand for labor, and that productivity gains will create ever more demand for new and different economic goods, and hence labor. As we have seen with every prior technology wave, automation lowers costs, which increases demand and creates more jobs than it eliminates. This can continue in a virtuous cycle.
AI … may cause the most dramatic and sustained economic boom of all time, with correspondingly record job and wage growth. - Marc Andreessen
Economic history and theory both support his optimistic view over the long-term; jobs will arise to replace the jobs lost. However, AI’s dizzying evolution may accelerate technology change to a point where some people’s skills are mis-matched to the jobs needed. What jobs will former truck drivers and Uber drivers take up? Will it benefit some but leave others behind? Hence the next concern - inequality.
Will AI Lead To Crippling Inequality?
He points out that most of the added value of AI will not go to the creators of AI, but to the users of it. With several billion people on smart phones, and the motivation to sell AI to the largest market possible, AI will be cheap and widely available, making things more equal, not less.
everyone gets the thing - as we saw in the past with not just cars but also electricity, radio, computers, the Internet, mobile phones, and search engines. The makers of such technologies are highly motivated to drive down their prices until everyone on the planet can afford them.
Because of this wide-spread adoption, AI will be an economic equalizer, not a source of inequality.
Will Bad People Do Bad Things with AI?
Yes, but bad people have always used technology for bad ends, AI as a tool and technology is not unique.
Since “the AI cat is obviously already out of the bag,” limiting AI is futile, so he suggests using AI as a defensive tool. That means using AI to build systems to identify fake or AI content and address challenges in cyber-defense and other areas.
Baptists and Bootleggers
He used the experience of Prohibition and used the analogy of Baptists and Bootleggers for what can go wrong in regulating an industry. Well-meaning “Baptists” try to impose a moral code through regulations, but self-interested opportunist “Bootleggers” always manage to profit from the laws and regulations and undermine the goal.
Some in the AI business use the mantle of expertise to make calls for AI regulation in the name of AI Safety, but they are in fact self-interested efforts to restrict competition. Yet even if the calls for AI Safety are well-meaning, his point is that the chance is high that AI regulation will end up leading to “regulatory capture, insulation from competition, the formation of a cartel.”
His warning is to be skeptical of these efforts.
His proposal is to lean in the other direction, with free market AI development by large AI companies and AI startups. Both should “be allowed to build AI as fast and aggressively as they can.” As for open-source development: “Open source AI should be allowed to freely proliferate and compete with both big AI companies and startups. There should be no regulatory barriers to open source whatsoever.”
The China Risk
China is the United States’ most powerful rival and biggest challenge in the world right now, and AI is the most important technology advance of our time. Andreessen highlights that this leads to a logical consequence: “The single greatest risk of AI is that China wins global AI dominance and we – the United States and the West – do not.”
As the world’s second largest economy, China is also the second most prolific publisher of papers on AI. China has vast capabilities in this technology and leading firms that are using AI in ways not adopted in the US. China also has self-imposed impediments because their authoritarian Government imposes censorship on AI.
We are in a potentially dangerous place if the way China adopts and applies AI helps their form of Government and their economy in ways we don’t. We face a challenge from China that is as much economic as military, and the technology race is woven into both challenges. We can’t afford to let China lap us on AI.
We should seek to win the race to global AI technological superiority and ensure that China does not. - Marc Andreessen
Summary
Andreessen’s essay is a great contribution to the AI discussion.
First, he is correct on most important points. Both in describing the benefits of AI and debunking unwarranted fears, he puts AI in the proper context of history, economics and technology, in place of hand-waving and hyperbole.
Second, his AI optimism is a useful antidote to over-concern and AI dooming that has pervaded discussion of AI’s impact on society recently. Dwelling on AI risks is useful in that we need to put in the guardrails to avoid serious problems, but the focus may cause many to mistakenly conclude AI might have more risks than benefits. His description of the tremendous value of AI is a good corrective on that point.
Third, because there is great uncertainty to what is to come with rapidly evolving AI, we need a vibrant discussion with differing perspectives about what AI will do for us and how to manage powerful AI. While there is risk of over-hyping the “new big thing” from Silicon Valley VCs, if anything, people are under-stating the profound long-term impacts that AI will have.
His final thoughts were to commend the “legends and heroes” building our AI future and to call upon everyone to go ahead and build it.
I believe we will do just that. The AI accomplishments still to come in this decade will blow your socks off. We have just seen the warm-up act, and the best is yet to come.
Postscript
I described the four main attitudes towards AI briefly in my article on the Open letter to pause AI, including the AI Enthusiast / AI Optimist, who believe that AI’s impact on humanity will be significant and largely beneficial. Andreessen clearly fits squarely in the AI Optimist camp. I’ll say more about these four categories in a followup.
