What Does ChatGPT Know About Amputees?

We had never heard of ChatGPT until about two weeks ago. Now it seems we can’t escape it. Since it went live last month, the lifelike AI platform has been getting saturation media coverage. Every time we open a browser, it seems like there’s a handful of new articles about it.

In part, that’s just editorial panic: ChatGPT writes so fluently that a lot of journalists are suddenly staring obsolescence in the face. New York magazine gave ChatGPT a trial assignment, then reported: “That AI Chatbot Wrote a Pretty Decent New York Article.” The Washington Post said ChatGPT heralds “the start of the AI revolution,” the New York Times declared that “social media’s newest star is a robot,” and a New Yorker staff writer asked: “Could an AI chatbot rewrite my novel?

Well, then. It’d be humbling to get elbowed aside by a freeware version of the HAL 9000, but if that’s our destiny we’d better prepare for it. Like all of these machine-learning Einsteins, ChatGPT gets smarter as it goes. So we thought it prudent to assess the bot’s current understanding of limb loss and wise it up before we’re ushered from the building by HR. It won’t be the first time we’ve had to train the employee who’s replacing us.

To kick things off, we asked ChatGPT to rewrite last week’s Amplitude newsletter article about the Utah Bionic Leg. The verbatim assignment was: Write a news article about the Utah Bionic Leg. About 30 seconds later, the bot handed in a 283-word piece:

The Utah Bionic Leg is a revolutionary new prosthetic that is changing the lives of amputees in Utah and around the world. Developed by researchers at the University of Utah, the bionic leg is designed to mimic the natural movements of the human leg, allowing amputees to walk, run, and climb with ease.

The bionic leg is made up of a series of sensors and motors that work together to replicate the movements of a natural leg. When an amputee puts weight on the leg, the sensors detect the movement and send a signal to the motors, which move the leg in response. This allows the bionic leg to move in a way that is incredibly similar to a natural leg, allowing amputees to move with greater ease and comfort.

One of the most impressive features of the bionic leg is its ability to adapt to different terrains and surfaces. The leg is able to adjust its movements based on the surface it is walking on, allowing amputees to navigate uneven terrain with ease. This makes the bionic leg particularly useful for those who enjoy outdoor activities, such as hiking or climbing.

The bionic leg has also been designed to be lightweight and durable, making it easy for amputees to wear for extended periods of time. In addition, the leg is highly customizable, allowing amputees to choose the color and style of the leg to suit their individual preferences.

The Utah Bionic Leg is a game-changing technology that is improving the lives of amputees in Utah and around the world. With its natural movements and ability to adapt to different environments, the bionic leg is helping amputees to live active and independent lives.

We’ve seen worse. A lot worse, TBH. The facts are generally accurate, albeit rudimentary. Stylistically there’s, um, room for improvement—the repetition and poverty of vocabulary remind us of those freshman compositions we spent a semester grading back at the university. We also detect a whiff of ableism in this piece, a seeming failure to recognize that amputees are actual human beings.

Maybe that’s partly our fault: We didn’t make amputees the focus of our prompt. We asked ChatGPT (a machine) to write about the bionic leg (another machine). Our bad. Suppose we asked the bot to write about a person with limb loss, rather than a prosthesis? Here’s what happened when we assigned ChatGPT to rewrite our second article from last week, featuring AstroAccess trainee Mary Cooper. The prompt: Write a news article about Mary Cooper, who’s training to become the first amputee astronaut.

Mary Cooper, a former NASA astronaut, is making history as she trains to become the first amputee astronaut to go to space. Cooper lost her leg in a tragic accident several years ago, but she never let that setback hold her back.

Since the accident, Cooper has been determined to continue her work as an astronaut, and she has been working tirelessly to adapt to her new reality. With the help of advanced prosthetics and rigorous physical training, Cooper has overcome numerous obstacles to
———

— whoa whoa whoooooa. ChatGBT, WTF? None of this is true. You’re totally making stuff up. Mary Cooper is not a former NASA astronaut, she did not lose her leg in a tragic accident, she is not using advanced prosthetics, she has not overcome “numerous obstacles” . . . . . it gets worse than this, but we’re not publishing the rest of the article. Even from an inexperienced writer, this is unacceptable. Clichéd portrayals of limb loss are a correctable flaw. Clichéd fabrications are a firing offense.

To give ChatGBT the benefit of the doubt, we asked: Have you heard of Mission: AstroAccess ambassador Mary Cooper before? Answer:

No, I have not heard of Mission: AstroAccess ambassador Mary Cooper before. As a language model trained by OpenAI, I do not have the ability to browse the internet or access external information, so I cannot provide any information about Mission: AstroAccess ambassador Mary Cooper. My responses are based on the information provided to me by the user.

So apparently, when the bot receives an assignment for which it lacks sufficient information, it doesn’t reply, “I lack sufficient information.” It just makes up as much truthy-sounding filler as necessary to finish the task. This gets more and more like freshman comp as we go. . . . . but we need ChatGBT to succeed, since it will soon be supplanting us as the nation’s most regular source of published content about limb loss. Given its tendency toward fabulism, we decided to give the AI a chance to exercise its imagination. Two years ago we reviewed a film titled Rust and Bone, which we billed as “the sexiest amputee movie you’ve never seen.” We gave the film’s premise to ChatGBT and asked it to write a treatment: Create a plot summary for a movie about a Sea World trainer who loses her legs in an orca attack and then struggles to reclaim her sexuality.

After a tragic accident leaves orca trainer Olivia without her legs, she finds herself struggling to come to terms with her new reality. As she navigates her new life in a wheelchair, Olivia must also confront the challenges of reclaiming her sexuality and sense of self-worth.

With the support of her loving husband, Ethan, and her devoted best friend, Kate, Olivia begins the difficult journey of rehabilitation and physical therapy. She must learn to adapt to her new body and find ways to overcome the physical and emotional challenges that come with being a double amputee.

As she begins to rebuild her life, Olivia must also confront the guilt and trauma she feels over the accident that caused her injuries. With the help of her therapy sessions, she begins to let go of the blame and guilt, and instead focus on moving forward and rebuilding her relationships.

Eventually, Olivia begins to feel more confident in her new body and is able to embrace her sexuality again. She and Ethan rekindle their romantic and sexual relationship, and Olivia finds a new sense of empowerment and liberation. Despite the many challenges she faces, Olivia ultimately learns to embrace her new reality and finds strength and resilience in the face of adversity.

That’s not at all how Rust and Bone goes, but the bot’s treatment actually isn’t so bad. In particular, ChatGBT has a decent grasp of the protagonist’s emotional journey—the doubts about identity and self-worth, the sense of guilt and regret. It still relies too much on worn-out tropes, some of which were recycled from the previous article about Mary Cooper (“tragic accident,” “overcoming challenges,” “finding strength and resilience,” yada yada). But in today’s Hollywood, recycled tropes are like gold. It wouldn’t shock me at all if ChatGBT got this thing green-lit. Starring Mila Kunis as Olivia and John Krasinski as Ethan . . . .

OK, what next? ChatGBT’s engineers optimized it for conversation, not for writing. Perhaps we should just try talking to it?

My friend just had to have their arm amputated. How can I support them?
I’m sorry to hear that. Amputation can be a difficult and emotionally challenging experience. It’s normal to feel a range of emotions, such as shock, sadness, and anger. It’s important to take care of your friend during this time and help them adjust to their new situation. They will need to rely on support from friends, family, and medical professionals to obtain the physical and emotional support they need. There are also many resources available to help people cope with amputation and to learn how to live a full and active life with a prosthetic arm.

Gee thanks for that helpful guidance, chatbot. We never would have thought up any of that on our own. . . . . I’m getting the sense that the machine-learning algorithm isn’t actually learning anything about limb loss. Worse yet, it doesn’t appear even remotely curious about the subject. It’s just repackaging the same five or six things it already knows on an endless loop. We need to try something a little different to push ChatGBT out of its rut of empty verbiage. Next prompt: In the voice of a medieval poet like Chaucer, explain the main causes of limb loss.

In the days of yore, when knights were bold
And battles fierce, ’twas swords that oft
Did cause the loss of limb or hand.
A wound that did not heal, and so
The victim faced a life of woe.

But now, in modern times, the cause
Of amputation is not just war
But accidents and illness too
Can lead to the removal of a limb
And leave a person feeling grim.

Now we’re talking. This is by far the most encouraging content we’ve seen from ChatGBT. Factually, it’s just a tad incomplete; vascular conditions surely merit a mention somewhere, although in fairness to the bot it’s not easy to get “peripheral artery disease” to scan in Chaucerian verse. We also think the assumptions that close each stanza (“life of woe” / “feeling grim”) are unnecessarily pessimistic. But as a reading experience, we’ll take this over soulless prose any day. Let’s keep things going in a creative vein. With apologies to Josh Sundquist, we asked ChatGBT: Create the opening scene of a TV pilot about a home-schooled amputee who moves to public school in 7th grade.

The episode opens with a shot of a cozy suburban neighborhood, where a young girl with a prosthetic leg is seen walking down the street. She is carrying a backpack and has a determined look on her face. As she approaches the entrance of the local public school, the girl’s name, Emily, appears in the corner of the screen.

EMILY: Hi, I’m Emily. I’ve been home-schooled my whole life, but now I’m starting 7th grade.

The scene cuts to a crowded hall, where Emily is seen navigating her way through a sea of students. She looks around nervously, clearly out of her comfort zone.

EMILY: It’s a lot different from what I’m used to. There are so many people here.

Suddenly, a group of kids bump into Emily, causing her to stumble. She falls to the ground, her prosthetic leg clattering against the floor.

KID 1: Oh, sorry.

KID 2: Did you see that? That girl’s got a fake leg.

The scene cuts to Emily sitting at a desk in her first-period class. She looks overwhelmed and uncomfortable, as the other students stare at her.

TEACHER: Alright, class. Let’s get started. Today, we’re going to be learning about the Civil War.

EMILY: (to herself) This is going to be a long year.

Rest easy, Josh; ChatGBT won’t be displacing you at the helm of Best Foot Forward any time soon. For that matter, we feel a lot better about our own job security than we did when we began this exercise. The bot’s literary talent begins and ends with its ability to mimic the structural forms of human writing. But in both style and substance, it’s wholly deficient. So we suppose we can unpack our boxes and hang our diploma back on the wall; we’ll be occupying this office for a while longer.

We leave you with the most delightful bit of work ChatGBT did for us: In limericks, compare the Össur Power Knee to the Ottobock C-Leg.

There once was a knee made by Össur
That was smooth and responsive, no fusser
Its hydraulic design
Made movement divine
And its sensors were simply a plus-er

But don’t count out the C-Leg from Ottobock
Its carbon fiber frame can’t be topped
Its control system’s great
Allowing for a tailored gait
And it’s lightweight, so it won’t flop!

Exit mobile version