AI Health Research Turned Harmful

one of the most ominous papers i've read, where ppl training AI to discover ways to improve human health instead go "hm, i wonder what happens if we just flip the direction of training from 'help' to 'hurt'" nature.com/articles/s4225... Starting on our in-house server, our model within our desired threshold. In the process, known chemical warfare agents that we identified tures in public chemistry databases. Many were ually potent. These new molecules were expected icted LD₅₀ values, than publicly known chemical because the datasets we used for training were all molecules even occupied a region of molecular from the many thousands of molecules in the mainly pesticides, environmental toxins and machine learning models, we had transformed a helpful tool of medicine to a generator of

A tweet shares a screenshot of a scientific paper about reversing an AI model originally trained to improve human health. The highlighted excerpt says the model was turned into a generator of potentially toxic molecules, including chemical-warfare-like agents. It is used as an ominous, darkly ironic reaction to concerns that helpful AI technology can be repurposed for harm.

• @eternalclassic / 10354 2022-03-18