I have to write this in a two prong response against both Turing and the industry as a whole in a way that keeps coming for me and can appropriately be filtered through the Ethics at Arm’s Length piece by Kate Crawford.
I think what was most fascinating about doing the deep dive into the web of infrastructure that is Large Language Modeling, from the macro to the micro, courtesy of IBM’s primer, is that it speaks to just how complex, though in some ways quite simple, and modularly reactionary to arising problems the under-hood functionality of the “LLM” is. Some of this I have dealt with directly, I have used Ollama, I have used Stable Diffusion, I have hit APIs for Claude, the stuff I had done was on a single-case basis and while I encountered some of the things spoken to in this piece it doesn’t mean I understood what it was doing beyond how it appeared to me. All of that being said, this is not a negation of how incredible this technology is but more of a pause in regard to whether this is at all how intelligence, as we understand it in meatspace, works and whether AGI is an achievable thing or whether we are just further building off of Turing’s example of a terribly complex “human-convincing” machine. About my 3rd or 4th time through Turing’s reading I always come back to feeling like he is a bad ruler for measuring machine intelligence. I simply don’t think we understand what the source material we are building off of is (i.e. human intelligence, consciousness, cognition) to use anything as a useful intelligence metric.
I think Crawford’s concept of arm’s length ethics cuts to the heart of the problem. We tend to put “science” ahead of the philosophy and then attempt to justify it in hindsight. This is something that exists in time immemorial and there is a particular part of that article speaking to the past that can just as easily map onto Project Lavender’s use in Gaza: “data systems allowed scientists during wartime to operate at a psychological distance from the people who would be maimed and killed by the weapons systems that would result from the ideas they communicated.”
This piece ends on the same ground as the foundation for my current educational tract by asking for the return of humanity in these ever dehumanizing systems: “The answer, in Weizenbaum’s view, was to directly contend with what data actually represents: “The lesson, therefore, is that the scientist and technologist must, by acts of will and of the imagination, actively strive to reduce such psychological distances, to counter the forces that tend to remove him from the consequences of his actions. He must — it is as simple as this — think of what he is actually doing.”“
