HUMAN ROBOTS

The problem is not that robots will become more intelligent than humans but that humans will become as stupid as robots.

Current debates on robotics and associated Artificial Intelligence (AI) focus on either the low level issue on how robots may replace human factory labour, or the high level issue on how AI endowed robots may become even smarter than (extra-)ordinary humans (witness the latest GO-game program that beat a champion player but note that simple rule-based games like GO, while able to generate billions of possible moves, can nevertheless be analysed and subjected to predictive algorithms).

On the low level issue we have known ever since the industrial revolution, that machines are supposed to make our lives easier in that they perform tedious and/or dangerous tasks. With regards to the latter, the military applications excite the fancy of many a militarist even if it takes away the number one reason for staging a war, i.e. to die for a cause. The robot that cleans the house, cooks and serves the meals, changes nappies and generally replaces the domestic slave is of course a nice idea that excites the middle classes who lack the means to employ their own service workers – in the knowledge that they ought to be able to mimic the upper classes in this respect. Of course they are respectful of the machinations employed by the oligarchs, monarchs and other super-stars of the cleptocracy, admiring their vast celebrity resources from a virtual distance.

Ever since Descartes quite sensibly declared that humans are nothing but biological machines, however sophisticated, scientists have sought to design and build machines that edge ever closer to the human model – but as Feynstein says ‘if I cannot make it I cannot understand it’. The main stumbling blocks are, one, quantum-biology and two, language. Of course the two may be intimately interlinked as for example studied in the field of bio-linguistics, and whilst there is almost no empirical knowledge about how language can arise from the brain, there are other far less discerning attempts to model language by machine.

The most advanced of these attempts is so-called machine translation which is based on statistical models of language use. Simply put, the program will search for previous translations of the word(s), phrases, sentences that serve as input, and voilà, you have the translation (and every time somebody uses the translation machine, more data is added). This works fine for metropolitan languages that have accumulated very large data bases for such purposes. It comes down to the truism that many if not most ordinary speech events (especially in written form) have been uttered again and again. As such it’s no problem to convert the English sentence ‘what’s for dinner?’ into German within a fraction of a second – faster than a human translator could do. Even so the permutations of even simple and often repeated utterances can confound the translation machine. Cultural relativism is not easy to program – just think about the cultural baggage the term ‘dinner’ has. Equally the rules of complex syntax confound all computational algorithms, hence practically all reasonably sophisticated sentences will be translated into syntactical gibberish (just try to input this very sentence into a translation machine, say for German, and then input the German to arrive back at English, and see what the results are!)

There is however a time honoured way around all this cultural and linguistic diversity, namely to apply a heavy dose of cultural and linguistic imperialism. If we all think, speak and act the same, the statistical probabilities will be very high that translation and knowledge transmission can be programmed successfully – i.e., we all will behave like a robot. The German language has a nice word for this process: gleichschalten (which Google Translate renders as ‘to force in line …’). Nazi-Germany, by applying brutal force, tried but failed. Now it’s the English world’s turn, using more subtle devices. No wonder ‘dinner for one’ is one of the most watched English-language comedy clips in Germany. Many a German speaker will simply adopt ‘dinner’ as a German word just like the French accommodate ‘le weekend’.

English as a modern lingua franca is the medium of globalization, comparable perhaps to how Latin was the scholastic lingua franca in pre-industrial Europe. The consequences are clear: if all the world’s scientists program AI in the English medium, progress will be rapid. As with many science applications, the outcomes can be either beneficial or more often than not turn out to be weapons of mass destruction. At the lower level we can look forward to applications as described below (Johnson et al. 2016):

A researcher from Ryerson University in Toronto studied the use of playful robots in language practice for children with autism spectrum disorder (ASD). After recording interactions between the learners and robots, the researcher concluded that robots enable children with ASD to communicate because of their low stimulus levels and predictable behaviors.

Note the advantage of ‘predictable behaviours’. Applied to higher levels this should have alarm bells ringing, à la Chomsky who debunked ‘language as behaviour’ comparing it to fascist practices we just described in German as gleichschalten as (to force in line …). As such the danger of AI is not that robots installed with AI will outsmart human intelligence but that the human intelligence of the masses will be reduced to AI, as predictable human machines that operates on the level of Pavlovian stimulus and response – the dream come true for the advertising industry and associated demagogues. Wilhelm Reich explained all this very well in his seminal Mass Psychology of Fascism. The AI programmers will become science nerds that will create an Orwellian nightmare in which the billions of human robots perform tasks at the pleasure of a few human masters – who themselves remain aloof of AI. All the same there will also be quite a few genuine robots that perform all the tasks even human robots are unable to do, like assembling precision machinery that can land on comets and do genome testing in three seconds flat.

The only hope is that the human masters themselves will engender a group of revolutionaries who will liberate the masses from oppression and restore their human intelligence to the level of shared social justice and common wealth. That the human robots will rise up, Spartacus-like, is an ever diminishing scenario exactly because AI is such an effective weapon to control and subjugate vast masses of people. Just ask yourself why the proverbial 99% remain utterly despondent in the face of an ever widening gap between the very few super-rich oligarchs and the billions of the ‘prekariat’ (a term invented by a German sociologist in order to up-date the ‘proletariat’). Only human intelligence can defeat the pseudo-intelligence that makes up the 1% who use AI as well as age-old primitive violence in equal measure to maintain the status-quo. Of course, in the meantime, there will be many a success story à la Orwell’s 1984 whereby the lone revolutionary reads The Theory and Practice of Oligarchical Collectivism only to find that such works are written by agent-provocateurs in order to catch people who might harbour criminal thoughts against Big Brother. Note that in 2016 ‘Oligarchical Collectivism’ is a very apt term to describe the global madhouse, or is it not?

Johnson, L., Adams Becker, S., Cummins, M., Estrada, V., Freeman, A., and Hall, C. (2016). NMC Horizon Report: 2016 Higher Education Edition. Austin, Texas: The New Media Consortium.