Archive note: This essay was originally published in 2015. It is preserved here as a snapshot of how I was thinking about artificial intelligence and cloud robotics at the time, not as a current technical forecast.

Back in December, I was on a phone interview for an article with The Daily Dot called Stephen Hawking thinks artificial intelligence will destroy us, where they asked my opinion on whether artificial intelligence would destroy humanity if left unchecked. The interview came about because of my involvement with co-founding MegaBots, Inc., a robotics startup focused on building a new sports league of giant fighting robots.

In the article, I sided with Hawking’s opinion—who am I to argue with the smartest person on the planet?—but caveated that “it really comes down to how companies and people utilize the technology that will dictate its future.” The article covers various roboticists’ opinions on the topic, spanning academia and the private sector.

Lately I’ve spent some time thinking about this topic again, sparked by a conversation with Matt Carney, a friend at the MIT Media Lab Center for Bits and Atoms, who worked at places like IDEO and Meka Robotics before starting his PhD program. While throwing back a couple of tasty beverages at the Muddy Charles Pub, we were chatting about robotics—as all good nerds should—and ended up discussing the next technology leaps in robotics. We referenced a previous conversation Matt had with Gill Pratt, the DARPA program manager who managed the DARPA Robotics Challenge, about leveraging the cloud for a robot’s operating system.

The concept is robots sharing common computing power required to perform higher-level functions. If those functions, which often require significant computational resources, could be shared through the cloud, the speed of robotic advances could increase drastically. The idea is also supported by open-source systems such as ROS—the Robot Operating System.

I know, I know: when is this guy going to get to the good stuff—where we all die by the robots we create?

That’s just it. This is where it could happen. As a unified approach to robotic development through open-source collaboration converges with the cloud as an operating system, an AI singularity could emerge. At that point, we’re talking about robots across the globe sharing a singular “awareness” that communicates and acts as one system.

Consider a scenario. You are engaged with Robot A in the United States, which shares its cloud intelligence with thousands of robots across the globe. Robot B in China experiences something different that forces the shared algorithms to re-process the engagement. Robot A instantaneously changes its behavior with you based on zero influence from your interaction, updating instead because of its connection to the cloud intelligence.

Scary, right?

It’s like having infinite access to data instantaneously and constantly improving on that information regardless of your interaction with the physical world—something we’re not accustomed to and that feels eerie.

This is where Skynet comes in. It might not take much to create the tipping point where machine-learning algorithms are presented with a scenario in which humans are identified as an unnecessary element in tasks, engagement, or utility. That could lead to the scary post-apocalyptic universe we’ve been treated to on the silver screen.

Should you be scared? Probably. But like any global social issue, it takes education, awareness, and action to draw clear lines around what is acceptable. In this case, that includes the added complexity of balancing or focusing technological advancement. At least be comforted that smart minds like Elon Musk and Stephen Hawking have spoken against the use of artificial-intelligence weapons, hopefully a step in the right direction.