The first week of June saw the publication of two significant scientific advancements, one in human embryonic gene editing and another in artificial intelligence, that could fundamentally reshape the future of humanity. The findings, published in disparate fields, raise complex questions about control, ethics, and the pace of societal adaptation to transformative technologies.
Precision in Embryonic Gene Editing
A preprint scientific paper released June 1 detailed a new method for editing human embryonic DNA with unprecedented precision. A team led by Columbia University geneticist Dieter Egli employed base editors, which make targeted alterations to DNA strands, to edit genes associated with cholesterol levels (PCSK9) and fetal hemoglobin (HBG). These edits, delivered into fertilized eggs and early-stage human embryos, largely avoided the chromosomal damage seen in previous CRISPR-based attempts. The research demonstrated the potential to correct genetic mutations linked to conditions such as high cholesterol, sickle cell disease, and beta-thalassemia. While the researchers stressed that the technique is "premature" for clinical use, the paper marks a notable step towards minimizing technical obstacles in embryonic gene editing. This advance contrasts with the widely criticized 2018 work of Chinese scientist He Jiankui, whose use of CRISPR on human embryos resulted in technical flaws and genetic damage.
AI's Accelerating Development
Concurrently, the AI company Anthropic published findings asserting that artificial intelligence is already accelerating its own development. Their research suggests that AI systems, like Anthropic's Claude, are increasingly writing the code that runs them, leading to faster product development cycles. While acknowledging that Claude is not yet capable of full self-direction, Anthropic posits that this trend may signal an early stage of recursive self-improvement (RSI), where AI systems can design and build successors at an ever-increasing rate. The company's engineers have reportedly shipped eight times more code in the last two years, partly due to AI assistance. Anthropic has even proposed that AI companies consider pausing frontier AI development to allow societal structures and alignment research to catch up, citing the decades it took to establish international regimes for technologies like nuclear weapons.
Dual-Edged Potential
Both advances possess a characteristic of "bivalence"—offering both immense benefit and profound risk. The precision in gene editing that could eradicate inherited diseases might also pave the way for a genetic caste system. Similarly, AI's capacity to accelerate scientific discovery could also lead to human disempowerment or existential risks if its development outpaces our ability to govern it. The speed of AI development, in particular, raises concerns that it could soon exceed the capacity of human oversight.
The implications of these developments echo historical moments of scientific revelation. The discovery of nuclear fission in 1938 was immediately understood for its potential to create a bomb, while the initial publication of CRISPR technology in 2012 garnered little attention before its profound impact became clear. The current advancements, however, present a more complex dual potential, leaving experts and the public at a threshold where the ultimate consequences remain uncertain. The danger lies not only in potentially choosing the wrong path but in proceeding without fully recognizing the significance of the doorway.