Zack Kass, The new RenAIssance: AI and the expansion of human potential

Zack Kass, one of the first 100 employees at Open AI and head of the Go-To-Market division, is a widely recognized speaker, consultant and, now, author.

Kass is an unabashed advocate of AI and believes it will be the key to sparking a “new renaissance” by creating an “unlimited supply of intelligence.” He offers a parallel to his argument by citing John Maynard Keynes’ article, “Economic Possibilities for Our Grandchildren,” published just months after the stock market crash of 1930.

Keynes argued that innovation and technology would soon solve the economic problem of survival by creating an age of material abundance, a time when people could focus on “living wisely and well.”

Kass believes that “unlimited intelligence” will spark something similar.

“The first renaissance was not merely a burst of art or science; it was a cultural rebirth. A fusion of breakthroughs across disciplines. Gutenberg’s printing press catalyzed literacy. Da Vinci’s anatomical studies advanced medicine. Copernicus redefined our place in the cosmos. Each discovery reinforced the others, building momentum that lifted humanity into an era of awakening.

Today it feels like we’re standing at the edge of a similar inflection point. AI is an intelligence multiplier. Properly harnessed, it could democratize education, revolutionize health care, and accelerate innovation. It could help us tear down barriers that have long constrained human potential and reshape the fabric of society itself.”

Kass is not a Pollyanna seeing the future through rose colored glasses.

“But progress is earned, not granted. Every great transformation brings both opportunity and risk. For AI to truly serve humanity, we will be forced to solve radical new ethical dilemmas, unprecedented economic disruptions, daunting technical challenges, environmental collapse, dehumanization, the loss of identity, and above all, terrifying uncertainty.”

The central thesis of the book, in two paragraphs:

“AI is going to revolutionize industries, sciences, and daily life. It will solve the existential problems that have confounded us for millennia, and those we’ve created for ourselves in just a few generations. It will rewrite social constructs and redefine what it means to be human.

I’m also obligated to tell you the ways in which AI could (and likely will) compromise our lives before I tell you how it will reduce suffering and increase joy.”

The book does a credible, but in my opinion incomplete, presentation of the benefits and problems that will arise from widespread availability and adoption of AI. Along the way he does introduce some interesting insights. For example, when discussing how AI is going to displace to many workers (some estimate a 10-20% unemployment rate in the near future) he eschews the economic arguments and notes that, in the U.S. and perhaps most of the industrialized world, your job is a major factor in determining your personal identity. The problem is not just how to buy groceries, but how to understand who you are.

There are four areas where I have strong disagreements.

First, a fundamental misunderstanding of “intelligence.” This takes multiple forms:

1. Equating the operation of a neural net with the human brain that is at best a caricature; focusing exclusively on electrical activity, while ignoring chemistry (hormones) and the “embodied mind.”

2. De facto equating human and machine intelligence. He does use the modifier ‘machine’ once in the book and frequently seems to make a distinction between what the computer brings to the table and what a human does. Overall, the distinction is glossed over; as when he speaks of AGI (artificial general intelligence) and “superintelligence.” Machine intelligence is, at most, a kind of idiot savant—extremely capable in a narrow aspect of human intelligence; one that make it a very effective tool; when employed by a capable human being.

Second, misidentifying the kinds of problems that AI can solve. Kass asserts that unmetered “intelligence” will lead to solutions to definite problems: those with known parameters and describable constraints.” Examples he gives are curing cancer, building a nuclear reactor, and removing carbon from the atmosphere. Except for, perhaps, building a reactor I fail to see how these problems have known parameters and constraints. But the general point is valid; many problems are in principle solvable via formal methods and algorithmic approaches.

Indefinite problems are those with no clear parameters at all. Examples he gives include the emergence of consciousness, what lies beyond the observable universe, meaning of life, and why we cannot just ‘get along’. Kass believes these are intractable for humans because, “we can’t imagine what we don’t know how to imagine.” Somehow, AI will be able to do this kind of imagination for us. I doubt it.

Kass cites examples of solved problems that have eluded human-driven solution, e.g., Alpha Fold solving protein folding. This kind of problem may be incredibly difficult; but because it exists in a context of a deterministic system it is solvable by formal methods utilizing massive computational resources. Such problems are not indefinite, merely as yet undefined.

Problems posed in the context of complex systems, e.g., defining optimal socio-political or economic organization, establishing a sustainable environment, or establishing a health care system grounded in prevention rather than curing cannot be addressed with formal methods. The Software Engineering Institute at Carnegie-Mellon University published a study of Ultra-Large-Scale Systems that exposed how traditional engineering approaches are incapable of developing. These kinds of problems are characterized by unknown and unknowable, variables and relations, conflicting requirements, emergent properties, and unpredictability. They are “wicked” problems that computers cannot solve, but humans with 50,000 years of adaptation and experience can, at least “good enough.”

Third, underestimating resistance. Early in the book, Kass takes account of the human tendency to assume the worst, to see and fear danger, and prioritize caution over curiosity. All the result of simple biological evolution in a dangerous world: negativity bias. He also notes that as much as humans dislike danger, they really do not like change.

He sees this bias for the status quo as the primary factor of resistance to AI and its “obvious” benefits.

Overlooked are, in my opinion, the real forces arrayed against his vision of a renaissance: the ultra-wealthy, essentially every politician in every government, bureaucrats, organized labor, and organized religion.

Fourth, and most importantly, failing to recognize the current inadequacy of human intelligence.

Kass speaks of a grand collaboration between AI’s unlimited machine intelligence, as a resource and tool, and human intelligence characterized by imagination, curiosity, and critical thinking. An abundance of AI’s seems assured, but where is an equivalent abundance of human collaborators?

Kass does spend time talking about “idiocracy;” humanity’s intellectual decline—a decline spurred by AI as a tool. He notes historical parallels: the pocket calculator would cause students to “forget math” and the printing press would eliminate human ability to memorize texts. (The latter echoes Socrates’ bemoaning written language eliminating human ability to memorize.) He cites numerous statistics documenting declines in reading, comprehension, and math skills. Decline in physical abilities are also documented.

What he fails to note: how the education system, in the U.S. specifically, has for several decades engaged in a race to the lowest common denominator, reducing standards, focusing on test scores only, and exploding bureaucracy prioritized over teaching. All of these evident from kindergarten to university. Not mentioned, and critically important, the 30-year emphasis on STEM and computational thinking—the very areas that machines can, and arguably should, equal or exceed human abilities.

To be fair, Kass does attempt to address the need for an education system that will nurture human intelligence. He cites examples like Waldorf Schools and Alpha Schools, the latter offering 2-hours a day of AI-driven individualized instruction in “subject matter” with most of the focusing on “social learning.” The problem: such schools are few in number, very expensive, and focus on pre-school through 6^th grade only.

Ignored are problems that include: how to scale the model; how to extend the model up to and including university education; and, significantly, how might the hundreds of thousands of students already in, and graduated from, the existing system to be “retrofitted?”

Perhaps the most poignant discussion the book concerns time. Kass believes, correctly, that AI and computing technology in general, will offer humans a lot of time and, at least potentially, the ability to utilize that time in ways that increase individual and collective happiness. He also asserts that happiness is dependent on inter-personal relationships: real-lie friends instead of screen “likes” and “followers;” engagement with nature over “binge watching;” and cultivation of lives of “virtue, meaning, and joy.” Interestingly enough, I constantly see the last phrase repeated in communities and publications associated with non-Trump conservativism.