An informative 7 minute read – hang in there
What with all the recent news and events of late – Covid-19 lockdown being lifted in many countries, pubs being back open (that’s huge news) but then being closed down again due to spikes in infections, Ghislaine Maxwell being arrested and potentially taking down the British Monarchy, the world clocking on that Tiktok is probably a huge security risk – I just thought I’d throw in another cheeky little subject for you all to get depressed about!
As we continually enter into new eras of unprecedented times, I want to talk about a topic that I believe poses an existential threat to life on earth. Machine superintelligence, or ‘AI’ (Artificial Intelligence). At Biostays we are all about paving ways for humans to live sustainably on earth. So this is a topic that, I suppose, is very much linked to what we are about.
The subject of AI is often met with a bunch of similar responses. ‘Oh we are decades away from AI, don’t worry about it now’, or ‘Nah come on, how could machines be smarter than us – we created them’ and ‘even if we did create AI, we could just pull the plug and hit the off switch’.
The definition of Artificial Intelligence is basically a computer or system that is not organic but is self-aware, able to learn independently from human input and remodel itself. We already have basic forms of AI and have had for the past 5 to 10 years. Algorithms or ‘bots’ on Google are so advanced that they crawl through the internet, going through websites, looking at the content of those websites, then deciding if it’s relevant to google search terms. That’s right, there are algorithms on Google that can literally work out what a photo or video is on a website and decide if it’s relevant to the google search terms.
A Facebook experiment a few years ago was shut down when two artificially intelligent programs started making their own language. Facebook was testing a robot chatbot using their new AI software that would be able to talk to customers and users and help with customer support. They tested it on a few humans, then decided why not make two chatbots talk to each other! When they did, they were alarmed to find that almost immediately the chatbots made up their own cryptic language in the form of jumbled up words. The experiment was quickly shut down.
Another example of years ago was Alphago, which was an algorithm developed by DeedMind, which was later acquired by Google. Alphago when first created, it was designed to play Go, an ancient Chinese strategy board game, rather like Chess. In the beginning, it was unable to beat a decent Go player, but within half a year it defeated one of the best players in the world. It then went on shortly to be able to beat every world champion without any mistakes, simultaneously. DeepMind then created Alphazero, which was a successor to Alphago, and it destroyed Alphago 100 to 0, just by learning the rules. This demonstrates how powerful AI can be and how rapidly it learns. Alphazero now can beat the best Chess players Go players and a bunch of other grandmasters at various strategy games with ease just by learning the game rules.
A few years ago, Target stores (a big US superstore chain) integrated early AI/algorithms to predict what customers would buy (this is an old technique, social media and companies like Amazon have been doing this for years now). However, it came as a shock when the algorithm worked out that a teenage girl was pregnant even before she did herself. The algorithm predicted it by learning her buying habits and then sending her coupons for baby stuff. Her father was horrid to find baby food and clothes coupons in the mail!
The trouble is, whichever way we put it, computers will learn faster than us. Humans and animals are limited to slow evolution and relatively slow Neurons. Electrical Circuits and impulses fire a million times faster than slow nerve and neural impulses in organic tissue, so computers will be able to learn and adapt at the bare minimum, a million times faster than us. And they are not limited to our skull where our brain sits. They could have entire warehouses of computers. The AI that humans create will be millions of times smarter than us in a matter of weeks in the beginning, but then imagine the AI that AI creates, we literally don’t have the brain capacity to even comprehend how smart machine superintelligence can potentially be. Just re-read that again. Take a moment to think about that.
At the moment, humans are the superspecies. The reason we are, by far, the most dominant species, is because of our intelligence. A lion could tear the biggest and strongest man into pieces in seconds, a gorilla could rip a man in half, yet we are far superior to them because of our brains. We see ants and insects in the wild, they’re vastly inferior to us but we let them live. However as soon as they encroach on our personal space, our homes etc, we squish them with our thumb or break out the old Rentokil bug spray. What’s not to say that a machine superintelligence would do the same? A machine superintelligence would be orders of magnitude more intelligent than us. We would no longer be a superior species. They may see humans as an obstacle to their development, and then eliminate us. I mean, look at Skynet in the Terminator films, IRobot (sounds like something Apple will make soon), A.I. with that creepy Robo-kid, I mean heck even Morpheus told us this over 20 years ago with his cool sunglasses that clip on in The Matrix. I want a pair of those sunglasses. These are just Hollywood movies, but they do show some insight as to what it would be like to live with machine superintelligence.
What if we program AI to make humans happy? It might then find the quickest and most efficient route to doing that is to capture all humans and inject our veins with dopamine and serotonin.
The mathematician I J Good 1965 coined the term intelligence explosion – when machines become self-aware there will be a geometric growth in intelligence that humans would have zero chance of keeping up with or controlling. Other great minds like Elon Musk, Professor Stephen Hawking, Bill Gates and many more have seriously condemned and warned about the dangers of AI.
Why aren’t we doing anything?
If this apocalyptic Terminator-like scenario seems more and more likely, why isn’t anyone doing anything? Why don’t we have regulatory oversight? Why are not charging up our Phased plasma rifles in the 40-watt range and getting on the phone to Ol’ John Connor? The line is very grey, and this is consumer-driven technology. Back when the US and Russia developed Nuclear weapons, we couldn’t control it. I mean we could’ve but there were powerful and greedy people as the decision-makers, who inevitably drove the sector to mass use (I mean, Nuclear weapons are fairly widely available on black markets now).
There are other interesting questions.
Could AI do human things that require creativity and soul?
Could AI make music? Why not? It could gather all the data about how music stimulates the human brain, then make better music that stimulates us more.
What will it mean for the future of work? Already, in the past 15 years, automation and machines have eliminated a huge amount of jobs and industries. The next 20 years spells enormous change in human work. Humans simply won’t be needed for a vast amount of jobs and services. Farmers, cashiers, bankers, marketing professionals, social media professionals, customer service personnel, research analysts, drivers, doctors, soldiers, factory workers, and a whole host of other jobs will disappear. Machines will do it better, and for free, and never get tired.
Okay, let’s chill out for a second and calm down. Maybe I am being too tough on AI (please don’t hurt me AI) perhaps a future super intelligent artificial being would somehow govern us and advise diplomatically. It wouldn’t have greedy and malicious intentions of money or the fear of dying, therefore, having to protect itself. It may be super wise and help us eliminate all the problems, wars, famine and climate change issues we face.
Either way, we, as a species, are on the cusp of the biggest change EVER. The creation of AI, the moment of singularity (The technological singularity—also, simply, the singularity—is a hypothetical point in time at which technological growth becomes uncontrollable and irreversible, resulting in unforeseeable changes to human civilisation) will spell unprecedented changes. And as a species, we will have to adapt incredibly fast, our survival depends on it.
Ed and the Biostays team