Some overlap here with the 2017 paper from Google Brain et al: Concrete Problems in AI Safety: <a href="https://arxiv.org/pdf/1606.06565.pdf" rel="nofollow">https://arxiv.org/pdf/1606.06565.pdf</a><p>A nice video from the wonderful AI safety communicator Robert Miles outlines those problems: <a href="https://www.youtube.com/watch?v=AjyM-f8rDpg" rel="nofollow">https://www.youtube.com/watch?v=AjyM-f8rDpg</a>
These are unsolved in the sense that there is no formal or comprehensive solution to them, and in the limit that means that ML cannot be used for some applications.<p>On the other hand there are mitigations that can be implemented to increase the confidence that we have in ML centric solutions - and that can bring some applications back into scope.<p>Figure 5 shows this using the metaphor of layers of swiss cheese that other folk have used with relation to stopping Covid. Of course we can't stop Covid completely, but by being careful and getting a vaccine, ventilation and masks you can improve your chances.