Hello, World!
Here’s a joke generated by this AI: “According to a recent poll, newsletter writers are more likely to not glow.”
I mean…I guess…?
Whether you glow or not, welcome back to Hold the Code! To kick off our new season, we discuss AI’s capabilities in comedy and healthcare. Our weekly feature dives into some of the legitimate dangers that are posed by current AI technologies. Happy reading!
Computer, Tell Me a Joke!
Humor is often considered a fundamentally human trait, but could a machine be funny as well? Today, robot comedians are already a reality. For example, Jon the Robot, an android created by engineering professor Naomi Fitter, can perform a pre-written stand-up routine and even vary its comedic timing in response to its audience. But while Jon’s jokes were written by people, many AI researchers hope to develop machines that can generate their own material.
The Challenge of Funny
Of course, incorporating humor into AI comes with many challenges. Machines learn by searching for patterns in data, but the things we find funny are often unpredictable. Humans also rely on vast catalogs of shared experiences and cultural references when hearing or telling jokes. Machines are limited to the information that humans give them, and thus have a hard time replicating the nuances of our humor.
On the other hand, comedy may be more predictable than we think.
“Rodney Dangerfield: ‘I get no respect.’ That’s a formula,” notes comedy writer Joe Toplyn. “There are plenty of formulas in comedy, and some of them are right on the surface.”
From this perspective, it seems conceivable that machines, if given enough data, could learn these formulas and use them to produce their own jokes.
How else can AI impact comedy?
Regardless of the feasibility of a funny AI, many researchers argue that there is intrinsic value in pursuing the quest for one. For example, the process might lead us to machines that enhance comedy in other ways, such as by generating prompts for human comedians. Additionally, an understanding of humor is really an understanding of how people perceive the world. An AI that can crack jokes could almost certainly be useful in many other applications. At the end of the day, the power of technology comes from how we choose to interact with it.
Is AI really transforming healthcare?
AI is a buzzword in various industries, including in healthcare. But is the buzz around AI transforming healthcare premature?
“There are some incredible uses right now, but I think the whole concept of it transforming health care in its present state—we probably ought to pump the brakes just a little bit on that,” said Brett A. Oliver, MD.
Quality and data soundness are among the top concerns that medical professionals have about the development and adoption of AI technologies in healthcare. Before these products are deployed, users must be sure that they are not harmful and are robust to failure, especially considering they deal with sensitive data and assist in making potentially life-altering medical decisions.
The issue of the AI Black Box
Additionally, the AI black box (the concept that algorithms take in data and, somewhat magically, output a solution) poses a risk to the healthcare industry. Oliver states that physicians want to know why an algorithm makes a certain decision, and without this knowledge getting them to adopt this technology could be a struggle.
So what can AI do right now?
There have been a number of successful AI-powered systems introduced to healthcare recently, but some have yet to be thoroughly tested. One app helps to identify which COVID patients can be discharged from the hospital, and another works to automatically turn doctor-patient conversation into physician notes.
“We’re really excited about [automatic physician notes],” Oliver says. “...[I]t’s getting clinicians away from the computer and engaged again with the patient.”
Weekly Feature: Haywire! AI’s Worst Case Scenarios
Okay, we’ve all seen a sci-fi movie — or heard from crazy Uncle Joe at the Thanksgiving table — about sentient computers and robots with human-like intelligence taking over the world one day. Luckily, some experts don’t think that day will come. Others are not as optimistic:
As machine learning engineer Malcolm Murdock writes in his novel The Quantum Price, “AI doesn’t have to be sentient to kill us all. There are plenty of other scenarios that will wipe us out before sentient AI becomes a problem.”
Ah, lovely, isn’t it!
Real talk though
Although we don’t have to be quite pessimistic about our future in an AI age, this article by IEEE Spectrum talks about 6 worst-case scenarios the growing prominence of AI technology may lead us to:
Deepfakes, disinformation, and crumbing trust: AI are getting terribly good at generating fake and false information that is getting increasingly difficult to discern from real data. Deepfakes, false data, and fake claims make it continuously harder for us to discern fiction from reality.
Speed vs. safety: Yes, AI is powerful. This is pretty common knowledge. In the rush to create and next most powerful AI technology before competitors, enemies, rival countries, would there be sufficient testing and safety protocols in place? Would there be sufficient ethical deliberations done to prevent serious unintended consequences?
End of privacy and free will: As stated in the article, “Today, AI-enabled systems predict what products we’ll purchase, what entertainment we’ll watch, and what links we’ll click. When these platforms know us far better than we know ourselves, we may not notice the slow creep that robs us of our free will and subjects us to the control of external forces.”
A bad outlook for your screen time: Social media platforms are already using AI to maximize time spent on their apps and products. Would there be a day where AI is so good at getting our attention, that we forget our lives outside of the screen?
The tyranny of AI design: “When individuals who fall outside of the biased norm are neglected, marginalized, and excluded, AI turns into a Kafkaesque gatekeeper, denying access to customer service, jobs, health care, and much more,” the article notes. “AI design decisions can restrain people rather than liberate them from day-to-day concerns. And these choices can also transform some of the worst human prejudices into racist and sexist hiring and mortgage practices, as well as deeply flawed and biased sentencing outcomes.”
Oops! Did we scare you too much?
Fear and ignorance of AI only hinders us from reaping its potential benefits. This is also why we are so passionate about this newsletter: if we all can learn about and stay conscious of AI’s benefits and risks, we will be on track to build a better AI-enabled society and a more equitable future.
Written by Michelle Zhang, Molly Pribble, and Larina Chen
Edited by Molly Pribble
I appreciated the notes about Deepfake, blackboxes and the overall application to healthcare. Keep up the great work. I'm trying to keep up with what all this means for all of us.