How Do We Know When the World Has Changed?
On June 1, a team of scientists published a pivotal preprint scientific paper claiming to have edited human embryonic DNA with unprecedented precision. As a technical achievement, this work signals a significant advance in genetic manipulation, particularly when contrasted with earlier attempts plagued by inaccuracies. The implications are profound: if further developed, such embryonic editing could potentially free future generations from debilitating genetic conditions. However, the New York Times, through veteran science writer Carl Zimmer, highlighted a more contentious aspect of this research: the capability it creates for engineering babies with specific traits—an unsettling step towards “designer children.”
The Race Towards Self-Improving AI
Coinciding with this breakthrough in genetic editing, AI company Anthropic released a post suggesting that artificial intelligence is accelerating its own development. This notion hints at a concept known as recursive self-improvement (RSI)—where AI systems might design and create increasingly sophisticated successors at an exponential rate. Already, a significant portion of the code running Anthropic’s Claude has been generated by Claude itself, leading to a remarkable increase in productivity for the company. This development portends a future where AI could evolve faster than any human oversight can match, as noted by Anthropic executives Jack Clark and Marina Favaro.
These two disparate yet interconnected advances serve as a precursor to a future that may look remarkably different from today. Both genetic editing and self-improving AI carry immense potential, but they also harbor risks that we may not fully grasp.
The Dual-edged Sword of Genetic Editing
To delve into the genetic editing paper, we find that despite sensational headlines, the research focuses on fairly narrow objectives. Utilizing innovative base editors—which create a small nick in the DNA strand as opposed to the more destructive CRISPR method—Columbia University geneticist Dieter Egli and his team successfully edited two genes: PCSK9, related to cholesterol regulation, and HBG, a form of hemoglobin crucial during fetal development.
The team managed to deliver these base editors into fertilized eggs and two-cell human embryos, achieving some edits without the chromosomal damage associated with earlier CRISPR attempts. However, the technique is far from flawless. Some edits were misplaced in the genome, resulting in a low success rate for normal embryo development. The researchers themselves cautioned that using this technology for treatment is “premature.” Yet, their findings signify a pivotal step towards precise embryo engineering.
Previous attempts, such as those by the Chinese scientist He Jiankui in 2018, ended in widespread condemnation due to ethical and technical failures. Should the current study’s results prove reliable, the significant technical barriers to embryo engineering may soon vanish.
AI Development: A Rapid Shift
In stark contrast to the slow-moving field of biology, AI is advancing at breakneck speed. Anthropic’s analysis indicates that the human effort required to develop AI is diminishing at every stage. Engineers now find themselves reviewing code predominantly generated by AI systems rather than writing it from scratch. This acceleration raises critical questions: If AI can improve itself, what implications does this have for human oversight and control?
This rapid evolution fosters a research loop increasingly dominated by AI, potentially outpacing human cognition and institutional governance. Without careful management, the proliferation of highly capable AI systems could lead to significant economic and societal transformations, a concept echoed by Anthropic CEO Dario Amodei.
The Intersection of Biology and Artificial Intelligence
Both genetic editing and AI development highlight a troubling duality: the miraculous potential on one hand, and the catastrophic risks on the other. The same genetic technology that might eliminate serious inherited disorders could also set the stage for a society structured by genetic privilege. Similarly, advanced AI holds the promise for unparalleled progress while simultaneously posing existential threats.
Anthropic’s leadership has even called for an international dialogue on regulating AI development, drawing parallels with historical efforts to control technologies like nuclear weapons. As they emphasize, the urgency to regulate self-improving AI is pressing, given the rapid pace at which these technologies are evolving.
The Threshold of Change
Identifying when the world has fundamentally changed is not always straightforward. Historical examples illustrate the inconsistencies in perception surrounding groundbreaking advancements. The discovery of nuclear fission in 1938 led to immediate concerns about its potential for weaponization, while the advent of CRISPR was met with initial confusion and lack of urgency among the governing bodies that would later need to address its implications.
This moment in time, with significant developments in both genetic editing and AI, leaves society at a threshold of unprecedented change. The possibilities presented by these technologies are tantalizing yet fraught with complexity. As we venture into this uncharted territory, we must remain vigilant in our understanding and approach, mindful of the doors that may open before us, even those we may not yet recognize.
A version of this story originally appeared in the Future Perfect newsletter. Sign up here!
You can read more about these developments in detail here.
Swati Sharma
Vox Editor-in-Chief
Image Credit: www.vox.com
