A (Brief) Massage Therapist's Guide to AI

AI is everywhere. And it does everything...or does it?

In this episode we break down the different types of AI, including generative AI like ChatGPT. We'll explain how each AI type operates uniquely, with a focus on their current roles and limitations. And boy, do they have limitations.

Corey provides a clear breakdown of AI, explaining the distinction between weak AI, which performs specific tasks, and the elusive strong AI seen in science fiction. She explores how AI is currently used in healthcare, from interpreting medical imaging to handling large data sets, and how it might influence massage therapy through the development of "rubbing robots." 

This episode offers valuable insights into the intersection of technology and massage therapy, balancing excitement with caution. Whether you're a therapist or just curious about the future of AI in health and wellness, this episode will give you a thoughtful perspective on what's ahead.

The definition of massage therapy (Kennedy 2016)
ChatGPT Just Solved Chess (Video)
Google apologises for Photos app's racist blunder (BBC)
Man is to Programmer as Woman is to Homemaker: Bias in Machine Learning
Artificial Intelligence and Its Role in Massage Therapy with Whitney Lowe (ABMP Podcast)
Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell
The Alignment Problem: Machine Learning and Human Values by Brian Christian

Chapters

• 1:03 Definition of Massage Therapy
• 1:38 Dopamine
• 4:12 AI Everything
• 7:30 Game Bots
• 9:04 Image Bots
• 9:48 ChatGPT
• 12:27 Data
• 14:07 Healthcare
• 15:26 Rubbing Robots

Transcript

Speaker 1: 0:42

Welcome to the Rub a HeelWild podcast about massage therapy. I'm your host, kori Rivera, licensed massage therapist and information magpie, and in today's episode we're going to talk about artificial intelligence and rubbing robots. As a quick reminder, this podcast uses the word rubbing to talk about massage activities that do not qualify as massage therapy. The academic definition of massage therapy is massage therapy is for both self-care and health maintenance. Massage therapy includes rubbing, but it also involves health promotion and education messages. The. The setting of the treatment, such as a hospital or a spa, may influence the results, but the results absolutely depend on the skill, training and experience of the therapist and on the therapeutic relationships and communication between therapist and client. Okay, let's talk robots. Recently I read a current model of how we think dopamine works. Didn't think I was going to open an episode about robots talking about a hormone, did you?

Speaker 1: 1:51

Dopamine has been called a feel-good hormone. We thought for a long time that dopamine was a reward. Do something that gives you an evolutionary advantage, like eating a Snickers bar. It's nutrient-dense? Okay, eat the Snickers, get a hit of dopamine. But in 1993, a study came out of Switzerland authored by Dr Wolfram Schultz. The study found that when monkeys were trained to open a box with food in it, they generated dopamine. However, the rush of dopamine from opening the box diminished over time. Eventually, there was no dopamine generated when the monkey found food. The study also discovered that if the monkey found food randomly, instead of every time it opened the box, the dopamine came back, which led to the question if it's not the food, then what caused the reaction?

Speaker 1: 2:39

Currently, we think that dopamine is what happens when you anticipate that the future will be better than the present. I would call it hope. Dopamine is what happens when you have hope. Dopamine is the pull of the slot machine, but not the coins that come out. Or hitting the button to look at social media, but not what's actually in your feed. Or waiting for that adorable kitten on your screen to do something adorable, which it totally will, because why else would its person have posted the video? There is a 100% chance of incoming adorableness and for some of us, dopamine is what makes us excited about artificial intelligence and rubbing robots. Because I am excited. Innovation makes me excited. Novel uses of technology make me excited. Accessibility makes me excited. I'm thrilled to be able to tell you about both AI and the rubbing robots that are being developed. However, all that dopamine also sparks a big old warning sign in my brain, because my excitement about the future also makes me vulnerable to unexpected consequences and hype cycles, and this is why these episodes are taking so long. My thoughts about artificial intelligence and rubbing robots swing like a pendulum with every new piece of information, and there is a lot of information coming through right now.

Speaker 1: 3:56

In this particular episode, I'm going to give you a rundown of AI in general and rubbing robot AI in particular. We're going to talk about what AI really is and what it really does. If you haven't already listened to the previous episode, massage Tools, I recommend it. Unless you live in complete isolation in the mountains or the desert or something you've definitely noticed the explosion of artificial intelligence everything desert. Or something you've definitely noticed the explosion of artificial intelligence everything. Ai will write your paper, draw you a picture, compose music and soften your irritated tone in that email conversation you've been having for three days straight with no conclusion in sight. That last one might be a me thing, but in the endless headlines about AI is it useful, is it dangerous? Is it cheating? Most people haven't seemed concerned about what. First, let's talk about what it's not.

Speaker 1: 4:47

The name artificial intelligence was coined at the first ever conference about it in 1956. The people who work on its development admit that it is a terrible name. There are a bunch of highly debated ideas about what intelligence in human is, let alone a machine. Is intelligence knowledge, abstract thinking, problem solving? What about emotional intelligence? Does the situation matter? How do you measure intelligence? What makes a human smart? And we all know smart people who are really dumb. How can someone be smart and dumb at the same time? So calling a computer intelligent is confusing off the bat. Additionally, there are two avenues of research in AI. One group wants to create something that can think in big bunny quotes, and one wants to make programs that do things better or faster than humans. These tasks are related but are very different if you're the person trying to accomplish them and spectacularly different if you're the person trying to write a grant that funds them. So what we do not have is called general artificial intelligence or strong AI.

Speaker 1: 5:50

This is the kind of AI that movies are interested in. This is Hal from 2001, a Space Odyssey, sunny from iRobot or Samantha from Her. It's a computer that understands, can think about ethics and morality and potentially decide to destroy humanity because we make some pretty terrible decisions. We have not made that. Depending on who you ask, we are close decades away or it is completely impossible. What we do have is weak AI. This is AI that can do specific tasks and, lately, can do them really well.

Speaker 1: 6:21

You've been hearing about algorithms for a while. Algorithms are like toddler AI. You've probably heard things like. The YouTube algorithm is a horrible thing that fans the flames of the fires of bias and division in order to keep you watching longer and make money from ads. And if you hadn't heard that, maybe you've heard. The Netflix algorithm can suggest movies you might like. Both are true, by the way. An algorithm is a strict set of instructions. It's programmed very specifically and spits out an answer that it was designed to figure out. The YouTube algorithm is programmed to figure out what keeps you watching and then feed that to you and your suggestions.

Speaker 1: 6:58

Artificial intelligence is made up of a group of algorithms and it can reprogram those algorithms to meet a goal. Instead of being fed strict instructions, ai is fed data. The part humans programmed is how AI deals with the data. This means that AI alters its own algorithms using a simple technique like trial and error to complete the task it was given. Deep learning is what happens when AI is given data and a goal but no instructions. Deep learning is kind of like Little Red Riding Hood here's a basket, get to grandma's house, godspeed. So and this is important all of the weak AI we have are learning to complete their tasks in a completely different way. There are a lot of methods, but we're going to talk about three. We're going to talk about AI that play games, ai that recognize images and generative AI like ChatGPT.

Speaker 1: 7:50

If someone asked you who comes to mind when you think about chess, garry Kasparov might pop into your head. Garry Kasparov, aside from being the world chess champion for 15 years, was the first grand chess master to be beaten by an IBM computer called Deep Blue in 1997. You might be less likely to remember Fan Hui, a world champion of the game Go, who was beaten by the Google computer AlphaGo in 2015. If you want to come to the Heelwell community, I will happily explain how the computers won, but for this episode, I just want you to know that they didn't win. Using the same method, go has exponentially more possible moves than chess, but the bot that won at Go wasn't a more advanced version of the chess bot. It was a completely different learning system. The other thing you should know is that both Go and chess are games of perfect information. That means that both players know what is happening all of the time. In games like hide and seek, you don't know where people are. In chess and Go, you know exactly where your opponent is and how many pieces they have and which kinds. Chess and go, you know exactly where your opponent is and how many pieces they have and which kinds. So Deep Blue and AlphaGo are totally different programs doing what looks like the same task beating a human at a game of perfect information but one of them got there using a submarine and one of them got there using a hot air balloon.

Speaker 1: 9:04

Now let's talk about image recognition. Identifying images is way more complicated than playing a game. Interestingly, these programs work a lot like human eyes. They were trained to recognize edges of things and then simple shapes, and then complex shapes and then whole objects and faces. This is the same order our eyeballs use to understand visual information. The training used to recognize visual images is a lot like playing Marco Polo. The training used to recognize visual images is a lot like playing Marco Polo. Marco Polo. The system makes a guess, gets feedback, changes the system a little bit, makes another guess. This loop of guessing and correcting eventually allows the computer to understand the image. That is the super simplified version and we are going to stop there. But the GameBots and the ImageBots Not the same. Okay, chatgpt ChatGPT was an interface released by the company OpenAI in November of 2023.

Speaker 1: 9:55

What made ChatGPT unique was its ability to reply to a human and sound like a human when it did. You might remember Watson, the IBM computer that won at Jeopardy in 2011. Now we come to Watson, who is Brad Stoker. While it was very impressive, recalling facts and using a specific way to answer is a pretty narrow task compared to a completely open response. Chatgpt is what is known as a large language model, or LLM. It also fits into the category of generative AI. This means that it isn't simply reporting information or meeting a goal. It is producing a unique answer every time you ask it something.

Speaker 1: 10:30

Large language models work by association. It learns what words are related to other words, which is truly impressive when you think about it. The word run has over 600 definitions, including idioms. You can run cross-country and also have a run in your socks. You can run a company or run a report, but the thing is ChatGPT doesn't understand anything. It associates, and it does it incredibly well. And while you could argue that humans are also association machines, humans have a great many more things involved, including, but not limited to, life experience, emotion, a body and social pressure.

Speaker 1: 11:08

Writing is more than words strung together, and words are more than sounds strung together. However, for a machine that strings things together without understanding what it's saying, it does a great job making me sound less irritated. It saves me from having to comb my writing for SEO keywords and it can write humorous things if you ask it, but it can't play chess. In 2023, a user in the Anarchy Chess subreddit pitted ChatGPT against the current best chess bot, which is called Stockfish, in a hilarious game. Chatgpt proceeded to change the color of the pieces, move rooks diagonally and ultimately lose because it put itself into check. This is because ChatGPT is a word association AI. It found patterns in written chess games and then imitated that. It never knew the rules of chess and if you tried to explain them, it wouldn't understand. Chatgpt cannot play chess and Stockfish cannot write a sentence.

Speaker 1: 12:01

Generative AI like ChatGPT takes data and a model and generates a solution. Did I say generative AI takes data? It takes data, a lot of data. So it takes data, a lot of data. So so much data An estimated 10,000 data points on the low end for each piece of a puzzle you're trying to analyze. So even if there were a generative AI model that could create massage therapy protocol from an intake form, there wouldn't be enough data for it to work. As in all things data-related, more data means greater accuracy, unless your data is inaccurate. You've probably heard some of the stories about bad datasets.

Speaker 1: 12:39

In 2015, google's fancy new auto-caption software tagged people with dark skin as gorillas. An earlier word association tool called Word2Vec would allow you to do word math. You could input king minus man and it would return queen, but it also gave answers like man is to computer programmer, as woman is to homemaker. Ouch Massage therapy researcher Dr Anne Blair Kennedy asked generative AI to create images about massage therapy, and first the generator told Dr Kennedy it couldn't do that because the images might be sexual in nature. So there's that, after some more specific prompting, it created a person with their face smooshed into a table while two different people touched their back, a therapist with three arms, and it couldn't figure out how to put a client face up. Each of the images also had some less than subtle Asian cultural overtones. In the gorilla example, the dataset used to train the auto caption had a lot of problems. Like it had more images of George W Bush than Black women. That means it had more images of one man than 14% of the population of America. In the programmer versus homemaker example well, how do you remove bias from a language? And in the third man, if you want to know what the general population thinks about something, ask a generative AI that's been trained on the entire internet. I'm going to finish up this episode talking about how AI is being used in healthcare and how it is being used in the rubbing robots that are being developed.

Speaker 1: 14:15

In healthcare. The most common use of AI is to read imaging and interpret test results. The second most common use is to deal with large amounts of data and third, in a limited fashion, to help design treatment plans. First, the imaging AI is able to spot pathological issues in MRIs and other scans faster than a human. It is particularly good in oncology and dermatology, both specialties that rely on imaging. One study found that the most accurate results came from a human and a program working together rather than one working independently. The program could identify potential problems, but it had a hard time incorporating other information from the patient's chart to make a decision. Ai can also help interpret test results to detect events like heart attacks. In oncology, ai can be used to decide what course of treatment is most optimal. The more we learn about cancer and disciplines like genetics, the more we are able to personalize treatment to individuals. But it takes a lot of work and time to understand all the information necessary. Potentially, ai programs will help clinicians navigate the large amount of data in patient charts. The advent of electronic health records has allowed us to keep better track of patients' medical histories. But keeping track isn't the same as organizing and it definitely isn't the same as effective utilization. So that's healthcare.

Speaker 1: 15:28

What about rubbing? A lot of the new robots that are being advertised toss in the fact that they use AI, but I think that's misleading. The availability and wide use of ChatGPT might lead you to believe that the robot can generate massages the same way ChatGPT generates sentences. It can't. The artificial intelligence in the rubbing robots is used to aim the robot arms. Aim the robot arms.

Speaker 1: 15:54

You know how when you sit in a rubbing chair at the mall and the tracks that are supposed to rub next to your spine rub on your spine because you're 5'2 and not 5'11 and your rib cage sits almost directly on your pelvis, so your waist is like an inch tall Just me, okay. The AI in the new rubbing robots solves those problems. Navigating in three dimensions is challenging and all bodies are different and, even better, are different on any given day. I have ribs that shift out of place, which results in my shoulder blades wandering around on my back. These robots can scan your body and adjust its pre-programmed routines to fit your special, unique shape. It uses artificial intelligence to do it, but the AI is less chat, gpt and more chess. But the robot doesn't read a client's intake form and ask them about their decision to clean out their flower beds for nine hours the day before, notice how far they can rotate their trunk without wincing, make a treatment plan and then appreciate the pictures of the ready-to-be-planted flower beds, because that would be massage therapy, and the robots only do rubbing.

Speaker 1: 16:57

So why should you, a massage therapist who works with humans, care about how AI works?

Speaker 1: 17:03

Mostly because, although some of the claims might be more hype and less truth, ai isn't going away. I wanted to give you a peek behind the chrome curtain so that statements made by tech companies would be less mysterious and, hopefully, less threatening in that unknown, foggy future kind of way. I encourage you to try some AI tools, because the ones that already exist can really save you time, particularly at boring tasks. I'm going to direct you to the ABMP podcast, where Whitney Lowe has done an episode on AI that talks more specifically about using the robots for good. Where Whitney Lowe has done an episode on AI that talks more specifically about using the robots for good. If you're looking for a place to start right now, I second Whitney's suggestion of perplexityai, a generative AI that cites and links its sources, along with providing a brief answer to your questions. In the next episode of the Rub, we're going to take a closer look at the robots themselves and what they might mean for the profession of massage therapy. I hope you learned something cool today and thank you for listening.