It is electrifying. My hair stands on end. My mind races.

I strive to comprehend what no human can truly grasp. I attempt it nonetheless. I try to visualize it.

A chemical weapon, so toxic that a mere droplet on the skin suffices. Like a virus, where even the finest aerosol infects. Instantly lethal.

Madness. And was it really that simple? Just two years ago, I graduated from university and joined this pharmaceutical conglomerate. Using cutting-edge AI, we decipher protein folding patterns that conventional methods would have kept secret for decades—perhaps forever. These protein folds hold the promise of new cancer medications, so desperately needed as more and more people succumb to the disease. I take pride in my work.

Yet there it is. Merely adjusting the target parameters, and the proteins no longer heal.

They kill.

Of course, it’s not quite that straightforward. Just as in drug research, having a blueprint isn’t the same as having an instruction manual. The journey to producing a novel protein structure is long and complex. Without the necessary expertise, suppliers, and facilities, progress stalls. The machine doesn’t know any of this. It doesn’t ponder healing, killing, production lines, or geopolitics. It simply works. It does its job, much like all of us. It lacks a conscience. And me? I think of my family. I do. I do!

I reset the target parameters. I recalibrate them for cancer medications and remain silent. And I hope. Because there are countless others like this AI. And countless more like me.

Background

The main character in this short story is confronted with a conflict as old as humankind. Using tools is a characteristical trait of humankind, and the way the way the tools are used depends on the user soley. A simple stone can be used to grind grain – and it is also suscpected as corpus delicti for the first murder on earth [1]. As humankind evolved, our tools evolved as well. But did our collective conscienece keep up? What are the ever more powerful tools are going to be used for? That is the simple question the short story vividly compresses into few lines, based on the real example of protein folding research with AI.

Google’s AlphaFold revolutionized structural biology

Google’s AI subsidiary DeepMind has developed a system called AlphaFold that can accurately predict the 3D structure of proteins from their amino acid sequences. This is a major breakthrough for biology, as protein structure determines protein function and interactions. AlphaFold uses deep learning and attention mechanisms to model the complex folding process and the physical and chemical constraints of proteins. AlphaFold has achieved remarkable results in the biennial CASP (Critical Assessment of protein Structure Prediction) challenge, where it was recognized as a solution to the protein-folding problem by the scientific community. AlphaFold’s predictions have been made freely available to researchers via the AlphaFold Protein Structure Database, in partnership with EMBL’s European Bioinformatics Institute.

Protein Q8I3H7: May protect the malaria parasite against attack by the immune system. [3]

AlphaFold has been used to advance research on various topics, such as malaria vaccines, cancer drug discovery, and plastic-eating enzymes. DeepMind and Isomorphic Labs are working on the next generation of AlphaFold, which will expand its coverage and accuracy to other biological molecules, such as ligands, nucleic acids, and post-translational modifications. These molecules are essential for understanding the cellular mechanisms and pathways, and have potential applications for drug design, protein engineering, and synthetic biology. The next generation of AlphaFold will enable a new era of ‘digital biology’, where AI can help unlock the mysteries of life. Already now AlphaFold has speeded up the research and discovery of new stable protein structures many times compared to pre-AI methods.[4]

„True crime“

„AlphaFold found thousands of possible psychedelics“ [2]

–> WIP

Artikel von P.M.

References

[1] How (and Why) Did Cain Kill Abel? The First Murder – Understanding the Bible
[2] AlphaFold found thousands of possible psychedelics. Will its predictions help drug discovery? (nature.com)
[3] AplhaFold public database.
Based on research published in:
Jumper, J et al. Highly accurate protein structure prediction with AlphaFold. Nature (2021).
Varadi, M et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Research (2021).
[4] A glimpse of the next generation of AlphaFold – Google DeepMind