> Also, I doubt that a rogue strong AI would struggle with opening a regular online bank or brokerage account. I very much doubt disembodied strong AI is possi

> Also, I doubt that a rogue strong AI would struggle with opening a regular online bank or brokerage account. I very much doubt disembodied strong AI is possible in the first place. If nature and physical reality in general is anything to go by, intelligence and agency correlate strongly with physical interaction with the environment (e.g. plants are less "intelligent" than most animals). Even though there's an argument to be made that such interactions can be simulated, research has shown that optimisation algorithms tend to exploit the imperfections of the simulation instead of developing generalised solutions or just become instances of Clever Hans [1]. I'm puzzled how some ML/AI researchers come to the conclusion that an algorithm without access to external information and with no means of manipulating its environment can do _anything_ harmful at all (aside from wasting processing cycles or bandwidth). Malware has explicitly specified functionality (e.g. delete or encrypt files, etc.) that didn't just "evolve" out of nowhere and requires very specific target environments (OS, APIs, access rights, etc.). [1] https://techcrunch.com/2018/12/31/this-clever-ai-hid-data-from-its-creators-to-cheat-at-its-appointed-task/ https://techcrunch.com/2018/12/31/this-clever-ai-hid-data-fr...