Why creating an AI to defeat us might be mankind’s solely hope

One look at the state of affairs and it is obvious that humanity has turned to a corner. On the one hand, we are smart enough to create machines that can learn. On the other hand, people in Texas are dying because elected officials want to keep the government out of Texas. Chew on it for a second.

What we need is a superhero better villain.

independence Day

People fight. Whether you believe it is an inalienable part of our mammalian psyche or that we are capable of restraint but unwilling, the fact that we are a violent species is inevitable.

And it doesn’t seem like we’re getting any better as we develop. University of Iowa researchers conducted a study of existing material dealing with “human aggression” in 2002, and their results, as expected, painted a rather unpleasant picture of our species:

Aggression, in its most extreme form, is an unmatched human tragedy. Hopes that the horrors of World War II and the Holocaust would create a global aversion to killing were dashed. Since World War II, homicide rates have increased rather than decreased in a number of developed countries, particularly the United States.

The rational endgame for humanity is complacent extinction. Whether through climate change or through mutually assured destruction by military means, we are stuck against progress.

Fortunately, humans are very adaptable beings. There is always hope that we will find a way to live together in peace and harmony. Typically, these hopes are abstract – if only we can solve world hunger with a food replication machine like Star Trek, maybe, just maybe, we can achieve peace.

But all of human history is evidence that this ever happened. We are violent and competitive. After all, we now have the resources to feed all of the people on the planet. We just decide against it.

So we need a better enemy. Choosing yourself as our greatest enemy is self-destructive and stupid, but no one else came forward. At this point we even start kicking the coronavirus in the ass.

Put simply, we need the aliens from Independence Day to come down and just attack the crap out of us.

Or … killer robots

To be very clear, we are not advocating that aliens come and exterminate us. We just need to focus all of our adaptive intelligence on an enemy other than ourselves.

With regard to artificial intelligence, we need a real generative adversarial network in which humans are the learners and aliens are the discriminators. That’s pretty much the storyline of Independence Day, the 1996 film with Will Smith (spoiler):

  • People are so bellicose that two adorable men, Will Smith and Harry Connick Jr., are forced to become war fighters in the military
  • Aliens who have likely seen The Fresh Prince of Bell Air and Little Man Tate decide this is a travesty and come to destroy us
  • Humans unite in a united front and defeat the aliens

The aliens created a problem and asked us to solve it. The only solution was optimization. We tweaked and resolved the problem and got an acceptable output. Anything but total cooperation and our species would have failed the discriminator test, and the aliens would have beaten our attempt like a cosmic Dikembe Mutombo.

The actual problem

People are often evil, bigoted, and full of malice. But we’re still human. The actual problem is that aliens don’t just cooperate and attack us. It would be difficult to focus on something like Brexit, a US election, or whether Google’s recent plan to look at ethics in AI is good with aliens currently firing lasers at cities around the world.

We can’t control aliens. In fact, it is possible that they don’t even exist. Aliens are not reliable enemies.

However, we have complete control over our artificial intelligence computers and systems. And we should definitely teach them to constantly challenge us.

The simple solution

With AI we can determine how powerful an opponent becomes with an intelligent, rapid development. We could avoid all the laser shooting at cities and slowly work our way towards the rally part where we all work together to win.

The current paradigm for AI development is to create things that help us. And maybe that hurts us in a void. Our soldiers use AI to fight other people. Our teachers and managers use AI to organize classrooms and workplaces. Some of it is good on the surface, some is supposed to be for the common good.

But how much easier can we make it to be human before we get a full case of Wall-E syndrome? We should develop artificial intelligence that challenges each of us along with life-saving and life-affirming AI.

Take the domain of chess, for example. There is no question whether humans or AI dominate the chessboard. The greatest human chess players can be beaten by robots running on smartphone processors. Ladies and gentlemen, it’s a wrap. And that’s good.

We have used chess as an analogy for war strategy for centuries. Now, if we get into a battle with a future evil sentient AI, we know that it can likely tactically defeat us.

We need to focus on training our troops to fight an enemy who is stronger and more capable than humans, and most importantly, on developing methods of defense that rely not on killing people, but on protecting everyone.

People obviously crave challenges. We compete for sports and fun. The only reason there are billionaires and trillionaires is because the idea of ​​being “the best” seems, through mere human hubris, a better idea than “the best for all of us”.

Perhaps a technological shift towards developing ways to challenge people in ways that we cannot challenge each other could turn the tide in our favor.

Imagine a video game that constantly challenges you in a very personal way, or a job evaluation system that adapts to your unique personality and experience and always challenges you to do your best job.

The overall goal would be to create a system that would replace tribalism with cooperation. Our DNA itself appears to be laden with human birth trauma, and we have spent the past 5,000 years (at least per recorded story) figuring out ways to get around it. But with a focused diversion, our passion for adversity could potentially become a strength for our species.

Maybe we need an AI opponent who is our “huckleberry” when it comes to the urge to compete. If we cannot make most people nonviolent, perhaps we could direct that violence at a tangible, nonhuman adversary whom we can all well defeat.

We don’t need killer robots or aliens for that. All we need is for the AI ​​community and all of humanity to stop caring about making it even easier to do all the violent things we’ve always done to ourselves and give each other something else that has to do with all of these harmful intentions.

Maybe it’s time we stopped fighting the idea of ​​robot overlords and came up with some robot overlords to fight.

Will Smith could not be reached for comment.

Published on February 18, 2021 – 20:11 UTC

Leave your vote

0 points
Upvote Downvote

Related Articles

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.