The Download: threats from space mirrors and credit for AI drugs

Insilico Medicine

The case of Insilico Medicine's AI-designed drug for pulmonary fibrosis brings a fascinating legal dilemma to the forefront of intellectual property. While the company proudly announced the molecule was "discovered by" generative AI, its patent application listed only human inventors. This discrepancy underscores a fundamental challenge in current patent law, which is designed to recognize human inventorship exclusively.

As AI systems become increasingly sophisticated in generating novel designs and solutions, the traditional framework struggles to accommodate their role, potentially disincentivizing transparency about AI's contribution or creating a legal void for AI-driven innovations. This situation forces a re-evaluation of what constitutes "invention" and who can claim credit in an era of advanced AI.

If AI can autonomously propose complex drug candidates, the line between human guidance and AI's independent creative output blurs significantly. The current legal stance could lead to a system where human names are merely placeholders for AI's work, or it might push for new legal interpretations that acknowledge AI's role, perhaps through new forms of intellectual property rights or revised definitions of inventorship that consider AI as a tool rather than a mere assistant.

Data Center Backlash

The growing political and public opposition to data centers, highlighted as a "must-read" story, represents a significant challenge to the bipartisan consensus that once supported America's AI buildout. Once championed for economic development, data centers are now facing scrutiny across the political spectrum due to concerns over their environmental impact, particularly energy consumption and water usage, as well as their physical footprint and noise.

This backlash is scrambling midterm elections and forcing politicians to reconsider their stance on the infrastructure essential for AI's expansion. This shift in public sentiment could have profound implications for the future of AI development. If communities resist the construction of new data centers, the physical capacity for training and deploying large AI models could become constrained, potentially slowing innovation or driving development to regions with less stringent environmental or community opposition.

The debate signals a broader societal reckoning with the hidden costs of AI, moving beyond abstract ethical concerns to tangible local impacts that affect real people and their environments. Addressing these concerns will be crucial for sustainable AI growth and public acceptance.

LLMs

A novel approach to understanding large language models (LLMs) is emerging, where researchers are treating these complex systems "like aliens" or "city-size xenomorphs." This perspective acknowledges that even their creators don't fully grasp how LLMs work, their capabilities, or the mechanisms behind phenomena like hallucinations. By adopting a biological or neurological research paradigm, scientists are attempting to dissect and analyze LLMs as if they were studying vast, unknown living organisms.

This "strange new science" aims to move beyond simply observing LLM outputs to understanding their internal processes and underlying logic. Such research is crucial for developing more reliable AI, establishing effective guardrails, and building trust in these powerful models. By gaining a clearer sense of what LLMs can and cannot do, and the "under the hood" mechanisms of their unexpected behaviors, this biological analogy could unlock new methods for controlling, predicting, and ultimately improving AI systems, making them safer and more aligned with human intentions.