Future of artificial intelligence
How will future Artificial Intelligences (AIs) reshape our economy and our society? Will we live in a future where we coexist with AI-robot beings (ala Star Wars) or will we instead persecute and enslave AI beings (Bladerunner)? Learn insider secrets about the future of artificial intelligence.
Future of the economy
Will humans live peacefully in a future dominated by artificial intelligences? - Future of artificial intelligence P6
When it comes to humanity, let's just say we don't have the greatest track record when it comes to cohabitating with ‘the other.' Be it the genocide of the Jews in Germany or the Tutsis' in Rwanda, the enslavement of Africans by Western nations or the Southeast Asian indentured slaves now working in Middle East Gulf nations, or even the current persecution experienced by Mexicans in the US or Syrian refugees in select EU countries. In all, our instinctual fear of those we perceive as different than us can lead us to take actions that either control or (in extreme cases) destroy those we fear.
Can we expect anything different when artificial intelligence becomes truly human-like?
Will we live in a future where we coexist with independent AI-robot beings, as seen in the Star Wars saga, or will we instead persecute and enslave AI beings as depicted in the Bladerunner franchise? (If you haven’t seen either of these pop culture staples, what are you waiting for?)
These are the questions this closing chapter of the Future of Artificial Intelligence series hopes to answer. It matters because if the forecasts made by leading AI researchers are correct, then by mid-century, we humans will be sharing our world with an abundance of diverse AI beings—so we better figure out a way to live alongside them peacefully.
Can humans ever compete with artificial intelligences?
Believe it or not, we can.
The average human (in 2018) is already superior to even the most advanced AI. As outlined in our opening chapter, today’s artificial narrow intelligences (ANIs) are immensely better than humans at the specific tasks they were designed for, but hopeless when asked to take on a task outside of that design. Humans, on the other hand, along with most other animals on the planet, excel in our adaptability to pursue goals across a wide range of environments—a definition of intelligence advocated by computer scientists Marcus Hutter and Shane Legg.
This trait of universal adaptability doesn’t seem like a big deal, but it demands the ability to assess an obstacle to a goal, plan an experiment to overcome that obstacle, take an action to execute the experiment, learn from the results, then continue to pursue the goal. All life on the planet instinctively executes this adaptability loop thousands to millions of times each day, and until AI can learn to do the same, they will remain lifeless work tools.
But I know what you're thinking: This whole series on the future of artificial intelligence forecasts that given enough time, AI entities will eventually become just as smart as humans, and shortly after that, way smarter than humans.
This chapter won't dispute that possibility.
But the trap a lot of commentators fall into is thinking that because evolution took millions of years to produce biological brains, it will be hopelessly outmatched once AIs reach a point where they can improve their own hardware and software in cycles as short as years, months, maybe even days.
Thankfully, evolution has some fight left in it, in part thanks to recent advances in genetic engineering.
First covered in our series on the future of human evolution, geneticists have identified 69 separate genes that impact intelligence, but together they only affect IQ by less than eight percent. This means there could be hundreds, or thousands, of genes that impact intelligence, and we'll have to not only discover all of them, but also learn how to predictably manipulate all of them together before we can even consider tampering with a fetus' DNA.
But by the mid-2040s, the field of genomics will mature to a point where a fetus' genome can be thoroughly mapped, and edits to its DNA can be computer simulated to accurately predict how changes to its genome will impact its future physical, emotional, and most important to this discussion, its intelligence attributes.
In other words, by the mid-century, around when most AI researchers believe AI will reach and possibly surpass human-level intelligence, we'll gain the ability to genetically modify entire generations of human infants to be significantly smarter than the generations that preceded them.
We’re heading toward a future where super intelligent humans will live alongside superintelligent AI.
Impact of a world filled with super intelligent humans
So, how smart are we talking about here? For context, the IQs of Albert Einstein and Stephen Hawking scored at around 160. Once we unlock the secrets behind the genomic markers that control intelligence, we could potentially see humans born with IQs surpassing 1,000.
This matters because minds like Einstein and Hawking helped spark the scientific breakthroughs that now line the bedrock of our modern world. For example, only a tiny fraction of the world's population understand anything about physics, but a significant percent of the world's GDP is dependant on its findings—technologies like the smartphone, modern telecommunication system (Internet), and the GPS cannot exist without quantum mechanics.
Given this impact, what kind of advances could humanity experience if we gave birth to an entire generation of geniuses? Hundreds of millions of Einstein's?
The answer is impossible to guess since the world has never seen such a concentration of super geniuses.
What will these people even be like?
For a taste, just consider the case of the smartest recorded human, William James Sidis (1898-1944), who had an IQ of about 250. He could read by age two. He spoke eight languages by age six. He was admitted into Harvard University by 11. And Sidis is only a quarter as smart as what biologist theorize humans can one day become with genetic editing.
(Side note: we’re only talking about intelligence here, we’re not even touching on genetic editing that can make us physically superhuman. Read more here.)
In fact, it's very possible humans and AI can co-evolve by creating a kind of positive feedback loop, where advanced AI helps geneticists master the human genome to create increasingly smarter humans, humans who will then work to create increasingly smarter AI, and so on. So, yes, just as AI researchers predict, the Earth could very well experience an intelligence explosion the mid-century, but based on our discussion so far, humans (not just AI) will benefit from that revolution.
Cyborgs among us
A fair criticism to this argument about super intelligent humans is that even if we master genetic editing by the mid-century, it would take another 20 to 30 years before this new generation of humans mature to a point where they can contribute significant advances to our society and even out the intellectual playing field alongside AI. Wouldn't this lag give AIs a significant head start against humanity if they decided to turn ‘evil'?
This is why, as a bridge between today's humans and tomorrow's superhumans, starting in the 2030s, we'll see the beginnings of a new class of human: the cyborg, a hybrid of human and machine.
(To be fair, depending on how you define cyborgs, they technically already exist—specifically, people with prosthetic limbs as a result of war wounds, accidents, or genetic defects at birth. But to remain focused on the context of this chapter, we’ll focus on prosthetics meant to augment our minds and intelligence.)
First discussed in our Future of Computers series, researchers are currently developing a bioelectronics field called Brain-Computer Interface (BCI). It involves using a brain-scanning device or an implant to monitor your brainwaves, convert them into code, and then associate them with commands to control anything that’s run by a computer.
We’re still in the early days, but through using BCI, amputees are now testing robotic limbs controlled directly by their minds, instead of through sensors attached to their stump. Likewise, people with severe disabilities (such as people with quadriplegia) are now using BCI to steer their motorized wheelchairs and manipulate robotic arms. But helping amputees and persons with disabilities lead more independent lives isn’t the extent of what BCI will be capable of.
What in the 2030s will look like a helmet or hairband will eventually give way to brain implants (late-2040s) that will connect our minds to the digital cloud (Internet). Eventually, this brain prosthesis will act as a third hemisphere for our minds—so while our left and right hemispheres manage our creativity and logic faculties, this new, cloud-fed, digital hemisphere will facilitate near-instant access to information and enhance the cognitive attributes where humans often fall short of their AI counterparts, namely speed, repetition, and accuracy.
And while these brain implants won’t necessarily boost our intelligence, they will make us far more capable and independent, just as our smartphones do today.
A future filled with diverse intelligences
All this talk of AIs, cyborgs and super intelligent humans opens up another point to consider: The future will see a much richer diversity of intelligences than we have ever seen in human or even Earth's history.
Think about it, before the end of this century, we’re talking about a future world filled with:
- Insect intelligences
- Animal intelligences
- Human intelligences
- Cybernetically enhanced human intelligences
- Artificial general intelligences (AGIs)
- Artificial superintelligences (ASIs)
- Human super intelligences
- Cybernetically enhanced human super intelligences
- Virtual human-AI hybrid minds
- A few more in-between categories that we encourage readers to brainstorm and share in the comment section.
In other words, our world is already home to a diverse range of species, each with their own unique types of intelligences, but the future will see an even greater diversity of intelligences, this time expanding the higher end of the cognitive ladder. So just as today’s generation is learning to share our world with the insects and animals that contribute to our ecosystem, future generations will have to learn how to communicate and collaborate with a wide diversity of intelligences that we can barely imagine today.
Of course, history tells us that ‘sharing’ has never been a strong suit for humans. Hundreds to thousands of species have gone extinct due to human expansion, just has hundreds of less advanced civilizations have disappeared under the conquest of expanding empires.
These tragedies are due to the human need for resources (food, water, raw materials, etc.) and in part, to the fear and distrust held between foreign civilizations or peoples. In other words, the tragedies of the past and of the present are due to reasons as old as civilization itself, and they will only worsen with the introduction of all these new classes of intelligences.
Cultural impact of a world filled with diverse intelligences
Wonder and dread are the two emotions that would best summarize the conflicting emotions people will experience once all of these new types of intelligences enter the world.
'Wonder' at the human ingenuity used to create all these new human and AI intelligences, and the possibilities they might create. And then 'dread' from the lack of understanding and familiarity current generations of humans will have with future generations of these ‘enhanced' beings.
So just as the world of animals is entirely beyond the comprehension of the average insect, and the world of humans is entirely beyond the comprehension of the average animal, the world of AIs and even super intelligent humans will be way beyond the scope of what today's average human will be able to understand.
And even though future generations will be able to communicate with these new higher intelligences, it’s not like we will have a whole lot in common. In the chapters introducing AGIs and ASIs, we explained why trying to think of AI intelligences like human intelligences would be a mistake.
Briefly, the instinctual emotions that drive human thought are the evolutionary biological legacy from several millennia worth of human generations who actively sought out resources, mating partners, social bonds, survival, etc. Future AI won’t have any of that evolutionary baggage. Instead, these digital intelligences will have goals, modes of thinking, value systems entirely unique to themselves.
Likewise, just as modern humans have learned to suppress aspects of their natural human desires thanks to our intellects (e.g. we limit our sexual partners when in committed relationships; we risk our lives for strangers due to imaginary concepts of honour and virtue, etc.), future superhumans may overcome these primal instincts entirely. If this is possible, then we really are dealing with aliens, not just a new class of humans.
Will there be peace between the future super races and the rest of us?
Peace comes from trust and trust comes from familiarity and shared goals. We can take familiarity off the table since we already discussed how non-enhanced humans with have little in common, cognitively, with these super intellects.
In one scenario, this intelligence explosion will represent the rise of an entirely new form of inequality, one that creates intelligence-based social classes that will be near impossible for those from the lower classes to rise up from. And just as the widening economic gap between the rich and poor is causing unrest today, the gulf between the different classes/populations of intelligences could generate enough fear and resentment that could then boil into various forms of persecution or all-out war. For the fellow comic book readers out there, this might remind you of the classic persecution backstory from Marvel's X-men franchise.
The alternate scenario is that these future super intellects will just figure out ways to emotionally manipulate the simpler masses into accepting them into their society—or at least to a point that avoids all violence.
So, which scenario will win out?
In all likelihood, we’ll see something play out in the middle. At the onset of this intelligence revolution, we’ll see the usual 'technopanic,' that technology law and policy specialist, Adam Thierer, describes as following the usual social pattern:
- Generational differences that lead to fear of the new, especially those that disrupt social mores or eliminate jobs (read about AIs impact in our Future of Work series);
- "Hypernostalgia" for the good old days that, in reality, were never all that good;
- The incentive for reporters and pundits to fear-monger about new tech and trends in exchange for clicks, views, and ad sales;
- Special interests elbowing each other for government money or action depending on how their group is impacted by this new tech;
- Elitist attitudes from academic and cultural critics fearful of new technologies the mass public adopts;
- People projecting the moral and cultural debates of yesterday and today onto the new technologies of tomorrow.
But like any new advance, people will get used to it. More important, while two species may not think alike, peace can be found through mutually shared interests or goals.
For example, these new AI can create new technologies and systems to improve our lives. And in return, funding and government support will continue to advance the interests of AI overall, especially thanks to the active competition between Chinese and US AI programs.
Likewise, when it comes to creating superhumans, religious factions in many countries will resist the trend to genetically tamper with their infants. However, practicality and national interest will gradually break down this barrier. For the former, parents will be tempted to use genetic editing tech to ensure their children are born disease and defect-free, but that initial goal is a slippery slope towards more invasive genetic enhancement. Likewise, if China begins to genetically enhance entire generations of their population, the US will have a strategic imperative to follow suit or risk falling behind permanently two decades later—and so will the rest of the world.
As intense as this entire chapter reads, we need to remember that this will all be a gradual process. It will make our world very different and very weird. But we will get used to it, and it will become our future.
Future of Artificial Intelligence series