Let’s talk about breaking tech.
On April 10 at 5:07 PST, four crew members of NASA’s Artemis II moon mission returned to Earth. The Artemis II mission was the first to travel toward the moon in more than fifty years. This crew of four set a new record for the farthest distance ever traveled from Earth. NASA recovered the astronauts from the Orion spacecraft in the Pacific Ocean. Reentry, or any entry onto planetary surfaces, is one of the riskiest parts of spaceflight. Most of my time at NASA’s Jet Propulsion Laboratory (JPL) was spent testing for the entry, descent, and landing (EDL) teams on various spaceflight hardware. The process of stress-testing software and hardware for planetary reentry takes years of human spacecraft interaction to find the limitations of their designs. Identifying technological limitations is the first step toward human control. This is the lesson I learned on my assignment with the Orion spacecraft.
Fifteen years ago, I was assigned to shoot footage of Orion’s waterdrop tests. It was a frigid February week at the Aberdeen Proving Grounds, north of Baltimore, Maryland. Our role was to set up high-speed, stereoscopic 3D cameras around Orion to record drop tests into an almost frozen, manufactured lake. At the time, I weighed about half the payload of the camera rig, which consisted of a Ronford Baker tripod with two fully accessorized Red One cameras, each with a long zoom lens, mounted side-by-side. I rigged the payload to the small three-person barge, using every ratchet and bungee in our arsenal. The tests involved hoisting Orion into the air by crane and releasing it into the water about twenty-five feet from the barge
I heard the countdown from the crane operator over the walkie-talkie. “Five-Four-Three…” The barge captain yelled at me, “Hold on, girlie!”
“Two-One-Drop Craft.”
Orion dropped from the sky into the lake, creating a wave so big that I went flying off the back of the barge, barely managing to grab an anchored oar attached to the boat. As I pulled myself up, I looked at the cameras and panicked. Did I strap down NASA’s only two Red Ones well enough? Thankfully, I did. I got hit with another wave and flew through the air again. In hindsight, what I did not account for was my own well-being.
The waves continued to crash against the barge from the direction of the spacecraft until they hit the retaining wall behind us and returned. The barge captain and I now had waves hitting us from both sides. He laughed hysterically as he watched me being tossed around. I was a little banged up, but I did not get seasick.
We did this test ten times.
That day, I learned a few lessons. Hold onto the barge for dear life, do not trust that the testing pool’s water is safe enough to touch skin (It was not; it burned), and keep testing until you find the limitation.
One of the best lessons I learned during my time at JPL was the purpose and process of testing new hardware. With each drop test and every parachute test, engineers and scientists are looking for the breaking point of their designs. If we don’t know the limitations and, with that, the scope of the hardware or software’s potential, we can’t fully control operations in our day-to-day activities. Finding the breaking point of a design is never a failure for these engineers; it is an opportunity to reimagine design and workflow, ultimately strengthening the project. This process for Orion’s team took more than fifteen years. Like our own industry, it is not usual for the payoff of our labor to be celebrated years later.
Fifteen years later, I am back in Baltimore, Maryland. It is the week of the Artemis II reentry. This time, I am working with a group that is trying to integrate AI (artificial intelligence) into their workflows safely and ethically. This hands-on, practical workshop, in partnership with the American Federation of Teachers (AFT), gave this group an entry-level taste of how AI can optimize their time-consuming workflows without displacing their jobs.
One member of the class admitted that his top takeaway from the workshop was the encouragement to experiment and to find the model’s breaking point. He also said that, so often, when we learn new technology, the process feels fragile. We don’t want to break it, so learning new products can be intimidating or off-putting. When he found an endpoint to the process he was trying to prompt, he felt reassured. Being encouraged to find a model’s frustrations or faults made him feel more comfortable that artificial Intelligence wasn’t going to take over his life. At this point, AI still needs human intervention to fully carry out tasks that feel authentic to humans.
I was flying back from Baltimore when Orion made its splash in the Pacific. There was something poetic to me that fifteen years later, I was not in the water next to the spacecraft, but in the sky it came from. There was a huge sense of relief when I watched the broadcast and the four crew members returned to Earth safely. That assurance and that safety came from years of human testing and evaluation of technology’s limitations, and from the creative redesigning and reimagining to course-correct and offset the spacecraft’s limitations—all done by humans.
In Solidarity,
Jillian Arnold
PS October 31, 2000, was the last time all of humanity was on Earth. Since then, there has always been at least one person in space.