5 reasons why robots won’t take over the world, yet
This is probably a cliche topic at this point. With big and influential names like Elon Musk, Stephen Hawking, Mark Zuckerberg and a few others chiming in their opinions on a speculative topic such as this, it becomes more than necessary for the world to wonder about the same. Generally, the progression of science and technology refrains from banking on “opinions” and leans on facts, data and inference but when we enter uncharted territories like Artificial Intelligence and Machine Learning, well not everything can be predicted correctly.
If you had asked me “Will the robots take over the world?” about 7 or 8 years ago… I would have probably laughed at your face and made some relatable joke from a generic sci-fi movie like any mediocre meme page out there. Certainly things have changed. Considerable number of research agencies and groups are conducting extensive studies on the impact of the growing robots being employed by companies instead of humans. According to Oxford Economics, around 20 million jobs will be replaced by robots by 2030. That’s only 9 years from now. We have more and more AI assistants like IBM Watson involved in data analytics and user interaction, taking care of the heavy weight tasks. You’d be surprised to know the extent to which we are relying on automation in all sorts of fields. If you think Self-Driving cars are scary then wait for Automated diagnostics where AI will be able diagnose your symptoms leveraging Machine Learning.
Well of course, nobody wants to “accidentally” be diagnosed with Cancer. And the doctors working on the research are also quite aware of the limitations of an AI assisted diagnosis. To be very pragmatic, in reality there are very few cases where AI or Robotics has been able to completely automate a process without any sort of human intervention. If we start considering these scenarios we find that there are certain reasons why robots will not take over the world any time soon:
As of now, robots can only replace non value added jobs
We sometimes talk about activities and refer to them as, “Mindless actions”. Updating inventory, logistics, packaging, manufacturing, setting up equipment, inspecting a part, and billing a customer are some of the industrially prevalent activities inside small and big companies. These would definitely come under the category I described. Needless to say that they are easy to automate and robots will definitely replace humans there. Even soft jobs like content skimming, SEO, data analytics etc are being automated quite rapidly. But that’s really about it.
All that is great but is it possible for a robot to replace Elon Musk as the CEO of Tesla? Executive decision making, delegation, intuitive risk taking, creative problem solving are a few of the things humans will still need to do while they are going to be surrounded by an army of robots, which ultimately will answer to them. It is great to think about hypothetical and what-if scenarios where we do actually create robots capable of doing these tasks, and even then: humans in power won’t let someone else take the wheel. Call it greed, lust for power, arrogance, or just common sense. It is highly unlikely to happen in the coming years.
We achieve AI Singularity, then what?
A hypothetical incident in time that is predicted to happen with a certainty of “not if but when”: AI singularity. Considering the recent development of AGI and self improving intelligent systems, we can quite positively say that eventually we will have artificial or human augmented systems that surpass our abilities. There are obviously plausible arguments against this prediction like the complexity brake suggested by Paul Allen. Even then reaching singularity seems very much possible. However, while I was reading “The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future” by Martin Ford, I realized that technology is not independent of human agency.
The problem is that we don’t have enough incentive right now (we may in the future) to develop AI capable of self improvement to such an extent that it will surpass reasonable limits. Even if we do, what would be the purpose of such a technology? I can tell you from experience, technology doesn’t exist of it’s own accord. We innovate for the sake of aiding ourselves. With the exception of terrorists who may innovate to destroy others. But I highly doubt terrorists would develop Skynet, anytime soon. That is largely owing to the fact that developing it would require a capital investment that won’t go unnoticed by security councils all across the globe.
Considering all this, it is hard to imagine an AI assistant that would learn to hate humans and starts improving itself so much that it kills me just because I asked her to call me Dr. Sexy. Not that I asked Cortana to actually call me that…
Economics and trade 101 for an all powerful AI
Regardless of human or machine, if you want an upgrade you need resources. Considering the fact that we live in a economically vibrant society it is an obvious yet rarely discussed aspect of self improving AI. It is quite evident that if a machine wants to improve, at each milestone of the learning curve it will need additional resources to expand. As it starts reaching milestones where it has to overcome more complicated hurdles, it will require more expensive resources. Such expenses are usually procured by funding or philanthropic investors who give a chunk of their wealth to the good cause of “Scientific Advancements”. But if we are to talk about an all powerful AI, we know that even those investors don’t have pockets deep enough. So will the self sufficient AI procure resources on it’s own? How? It is not an economically capable entity which is allowed to trade or hold assets. So will it procure resources by unfair means? Depends on the whether it’s source code allows it to commit crime. Maybe of it were a seed AI that could alter its source code to override its prerogative of not committing a crime while prioritizing self improvement? Again, I don’t think human greed and common sense would let it do that.
Creative or repetitive?
Talking about trade, let’s consider an AI that could take on the world with it’s magnificent, artistic abilities. We have seen machines capable of producing music, making self portraits and writing very funny haiku. Their work of art carries this innate beauty that attracts millionaires and it is sold for a hefty price, but ultimately it is more or less a copy of what humans can do regardless of what machine learning approach the system takes.
No no, don’t get excited about that one robot who made that one painting. It would be naive to think that people would consider buying paintings made by AI when it has lost it’s novelty. They are attractive now, because we have not seen anything like it before, but it won’t replace a human artist. That is simply because of the nature of human consumers. Once they realize that that particular piece of art is a product of predictable and transparent learning process, people will lose interest. Art is attractive because of it’s illusive origin hidden deep in the mind of the author or artist which has been left open to interpretation for the people. Whereas anything a machine builds has a defined meaning and purpose that can be traced back to it’s origin and thus it is boring.
You can always switch it off.
Just remember to put a switch!
I know it seems like an oversimplification of a larger aspect related to human agency and involvement in the development of AI but it lays it down simply. We have been very active in discussion regarding the dangers of technology and thanks to that we have developed a sense of responsibility that allows us to consider counter measures. The purpose of such conversations and debates is to come up with solutions and contingencies to this inevitable threat. We learned from our mistakes of the past, and we are definitely going to build a better future.
