Robotic limitations

Issue section: 
Issue: 
(423)

Joseph Choonara’s article “If robots took our jobs, could they do them?” (March SR) made for refreshing reading. I work as a software developer and there is much talk of the impact AI will have on employment levels in our line of work. Much of it seems to follow the bleak “robot overlords” narrative that has dominated the debate so far.

I was pleased that Joseph raised the profitability problems inherent in automation under capitalism. Crucial arguments such as this one have been swept under the carpet by AI proponents.

Another issue that seems to go unmentioned is that of NCPs or “Non Computable Problems”. There are many problems that cannot be solved by any algorithm. A good example is the “halting problem” discovered by the late, great Alan Turing. He proved mathematically that it was impossible to write an algorithm that could analyse other algorithms and predict if they would end up stuck in an infinite loop, repeating the same section of code forever. Humans are able to complete this task very easily.

I am therefore a little more sceptical about AI and think that we are quite some time away from replicating the complexity of human intelligence. Equating the human brain to an algorithm belittles the limitless potential of human beings and their ability to change the world, which I suppose is what our rulers want.