As long as humans continue to discover new natural phemomena and invent new technologies based on using those phenomena, we will have to continue worrying about what our creations might do to us, taking our jobs and making us slaves to machines. People are both right and wrong to worry about artificial intelligence (AI). On the one hand, AI potentially offers major positive changes in to our lifestyles. On the other, in the hands of greedy corporate bosses and power addicted elitists, it could be a significant weapon in the battle for power and control.
On the other hand, complex intellectual machines working without human interaction could have very dangerous consequences, Mark Bishop, a University of London expert in cognitive computing, said in a radio interview.
“A group of semi — autistic nerds experts in America predict that within two decades 43 percent of jobs in engineering and factory work and even in the realm of cognitive activity could be automated. They also believe that certain jobs would be hard to automate,” professor Bishop noted.
He added that he had experience of intelligent machines engaged in procurement and in calculating the efficiency of certain jobs, the need to replace some personnel and things like that. The Daily Stirrer’s information technology expert compared this to his own experiments with A.I. and the illogical results algorithms often deliver. He also asks why, if Amazon’s A.I. is as smart as they say, does it frequently suggests as his next purchase in the Kindle store, the book he has just finished reading, that was bought from the Kindle store.
Mentioning Elon Musk and his letter of warning about the dangers inherent in the use of AI, Mark Bishop said that things like human creativity and mathematical insight are fundamentally not computable. Same goes for Stephen Hawking’s warnings about Artificial Intelligence leading to machines that surpass our intelligence and eventually destroy us. Some people point to the existence of machines which can be Grand Masters at Chess or leading players at the fiendishly complex Chinese game Go. Such talk is irrelevant, the best Chess or Go players are potentially capable of writing poetry that evokes an emotion, playing a silly game with their children, or laughing at a joke but the computers that beat them at their own game are only programmed to do one thing.
“Another theory is that computers don’t really understand the meaning of symbols, that they don’t really understand anything. The third argument in this critique is that computers cannot digitise and make use of the sensory inputs that are so important to living things, including humans. Sight, sounds, scent, taste and touch, sensations which have a profounf effect on our perception of situations, cannot be programmed into software, an expression in your face, in the tapping of raindrops on the window, or the fragrance of a flower, these have an effect on our mood and thous on our responses to stimuli. This is quite true, but such non binary information is not available to Artificial Intelligence. Bishop says computers work by parsing vast amounts of information incredibly quickly and filtering out certain strings. If anyone believes this is how humans thing, those people are obviously lacking sufficient intelligence to qualify as human.
Computers can never give rise to sensations, they work solely on digital input. Considering all these three things, one could say that a computer could really behave like a terminator because it has no sensations. Many preditions have been made about how the future will be ruled by algorithms but algorithms have to be programmed by a human. One recent scare story (sensation sells, remember,) warned that algorithms are already influencing the way we think. This may be partially true, if people look only at the information that appears in their newsfeed on websites like Facebook, they will be seeing a highly controlled version of current events reporting. All we have to do to defeat this is continue being human, being random and unpredictable, and being as sceptical about what we see on the web as we are about what we read in the tabloids.
When asked about the development of human -level artificial intelligence, professor Bishop said that a big question is whether people can engineer artificial intelligence capable of seamlessly deploying knowledge from one domain across to another. Hmmm? After a long career in Information Technology, during which academics have constantly predicted ‘True Artificial Intelligence is just a few years away,’ we are no closer now to true artfical intelligence in spite of all the advances in speed, miiturisation and functionality. The machines may do a lot more that the old mainfraim on which I loaded my first COBOL program from punched paper tape, but they still only do what they are programed to do. Computers have no autonomy, which is probably a good thing. We have seen too often the consequences of autonomous beings that fair to respond to the constraints on behaviour our senses suggest to us.
I have often said and written, “The only way machines can ever be truly intelligent is if we radically redefine what we mean by intelligence.