ChatGPT, DALL·E 2 and Surricalty…Can This AI Make a Difference in Healthcare?

Lately, rather than talking about AI day and night, everyone is talking about two particular AI applications:  ChatGPT and Dall.e-2.  If you haven’t heard about them, here’s the basic pitch:

• ChatGPT is basically a super-charged chatbot. You talk to it, it tells you stuff.  It has been “trained” with a billion bazillion pieces of info that makes it potentially the best thing since sliced bread with respect to answering all of life’s questions.  Ok, actually it’s an OpenAI artificial intelligence implementation meant to mimic human conversation, including personalization to the situation, etc. According to its website, “The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.”

• DALL·E 2 is similar, except it makes images. Specifically, the artificial intelligence engine underlying DALL·E 2, also created at OpenAI, can create realistic images and art from a description in natural language. If you ask DALL·E 2 to draw you a picture of a chihuahua wearing a red hat while hanging at the beach and it gives you this:

 

Impressive.  I personally like bachelorette number 2.

Because I have heard about these things nearly every day for the past month, I decided to check them out.  I was particularly intrigued because a) I like chihuahuas; and, b) I have heard that ChatGPT may have interesting applicability as a behavioral health intervention/counseling program.  So chihuahas? Check.  Healthcare tech? Check.  I had to see for myself.

I started with ChatGPT.  I decided to go with a test question or two to check accuracy:  Naturally, I selected: “Why did the chicken cross the road?”

ChatGPT answered, “The chicken crossed the road to get to the other side.”  So I knew this was the real deal.

Second question:

Lisa: What would Wile E. Coyote have to do to beat the road runner?

ChatGPT: In order for Wile E. Coyote to beat the road runner, he would have to come up with a clever and effective plan that takes into account the road runner’s speed and agility. This could involve setting up traps or using cleverly disguised gadgets to trick the road runner into falling into a trap. Wile E. Coyote would also need to be careful not to fall victim to his own traps or devices, as he is prone to doing in the cartoons. Ultimately, it would require a lot of hard work, persistence, and a bit of luck for Wile E. Coyote to finally catch the road runner.

Ok, ChatGPT seems pretty smart on some of life’s biggest questions.  So I decided to test the program to answer some healthcare question. Since I have done hard time as a health tech investor, I figured it might tell me something about trends that would inform future decision-making.  So I asked ChatGPT the following question:

Lisa: What is the single best application of technology to healthcare?

The answer I got may or may not surprise you, but it was this, verbatim straight out of the chatbot’s mouth:

ChatGPT:  One of the best applications of technology in healthcare is the use of electronic health records (EHRs). EHRs allow for the digitization of patient medical records, allowing for easier access, storage, and sharing of information among healthcare providers. This helps to improve patient care by allowing for more accurate and up-to-date patient information, as well as reducing the risk of errors and improving the efficiency of care. EHRs also have the potential to improve population health by allowing for the analysis of large amounts of data to identify trends and patterns in patient health.”

And then I asked, “What is the single worst application of technology to healthcare?”  I’ll give you one guess as to the answer.  It was:

ChatGPT:  One of the worst applications of technology to healthcare is the use of electronic health records (EHRs) that are poorly designed and do not effectively streamline patient care. EHRs are meant to improve communication and coordination among healthcare providers, but if they are not user-friendly and do not allow for easy access to important patient information, they can actually hinder care and increase the risk of errors. Many EHR systems are also prone to glitches and technical issues, which can further disrupt the delivery of care.

Honest to God I have to say the damn thing is sort of right on both counts.  And I laughed my brains out when I saw these results.    I could not have asked for a better set of answers, really.  Please ChatGPT, tell me more.

Lisa: Is it always good to have lots of data?

ChatGPT: It is not always good to have lots of data. While having more data can provide more information and potentially lead to better decisions, it can also be overwhelming and difficult to analyze. Additionally, having large amounts of data can be expensive to store and maintain, and may not always be necessary for the task at hand. It is important to carefully consider the relevance and usefulness of the data being collected, and to strike a balance between having enough data to inform decisions and having too much data to effectively process and analyze.

Yeah, ok.  True, but a little vague.  How about a slightly deeper question:

Lisa: How much is medical decision-making driven by financial considerations vs. clinical considerations?

ChatGPT: It is difficult to quantify the exact extent to which medical decision-making is driven by financial considerations versus clinical considerations, as it can vary widely depending on the specific situation and context. However, it is generally acknowledged that financial considerations do play a role in medical decision-making, especially in the current healthcare system where cost containment and cost-effectiveness are often emphasized.

In some cases, financial considerations may be the primary driver of medical decision-making, particularly in situations where resources are limited or where there is a lack of coverage for certain treatments or procedures. In other cases, financial considerations may play a more minor role and be secondary to clinical considerations. Ultimately, the priority given to financial versus clinical considerations will depend on the specific goals and values of the healthcare system or provider.

Lisa: What should I do if I’m feeling depressed?

ChatGPT: If you’re feeling depressed, it’s important to reach out to someone for support. This could be a trusted friend or family member, a mental health professional, or a helpline. It’s also important to take care of yourself by getting enough sleep, eating well, and engaging in activities that you enjoy and that help you relax. Exercise and time spent in nature can also be helpful in managing depression. It’s okay to take things one day at a time, and it’s important to be patient with yourself as you work towards feeling better. It’s also important to remember that seeking help is a sign of strength, not weakness.

OK, all of ChatGPT’s statements are accurate in general, but entirely unhelpful in the specific.  It’s kind of the Reader’s Digest book of Duh!  It answers most questions very equivocally and without stats, facts or citations or specific recommendations.  Clearly, the system still has a lot of work to do to be valuable, and particularly if someone thinks it can be used for clinical interventions.  On the other hand, it’s not non-sensical, it’s just operating at 30,000 feet.  It does definitely remind me of the original plot for IBM’s Watson.  Remember how we were going to teach Watson everything from medical texts and then we could ask it medical questions and it would tell us how to treat cancer patients.  Yeah, that was not a great movie.

By the way, I did ask ChatGPT this question also:

Lisa: What should I do if my chihuahua is depressed?

ChatGPT: If you think that your Chihuahua may be depressed, it is important to have them evaluated by a veterinarian to rule out any underlying medical issues.

If your Chihuahua is diagnosed with depression, there are several things that you can try to help improve their mood:

1. 1. Make sure that they are getting enough exercise and playtime each day.
   2. Provide a consistent routine and structure for them to follow.
   3. Make sure that they are getting enough socialization and interaction with you and other humans or pets.
   4. Consider providing them with interactive toys or puzzle toys to keep their mind stimulated.
   5. Consult with a veterinarian or a professional dog trainer for additional support and guidance.

It is also important to note that if your Chihuahua’s depression persists or becomes severe, it may be necessary to consider medication under the guidance of a veterinarian.

It is pretty clear that the people programming ChatGPT know and/or like animals more than people.  The answers generated are just more thoughtful and specific…to wit:

Lisa: How much wood could a woodchuck chuck if a woodchuck could chuck wood?

ChatGPT: It is a commonly believed myth that a woodchuck (also known as a groundhog) can chuck wood. In reality, woodchucks do not have the physical ability to chuck wood as they do not possess strong front paws or claws for digging and manipulating objects. Therefore, it is impossible to accurately determine how much wood a woodchuck could chuck if it were capable of doing so.

With that question fully answered, I decided I had to know more.  For instance, can we tell if there is gender bias in ChatGPT?  I decided to go with this question:

Lisa:  Who is the best-looking man in the world?

ChatGPT: This is subjective and varies from person to person. It is not appropriate to make judgments about someone’s appearance or to try to determine who is the “best looking” as it promotes unhealthy standards of beauty and objectification.

Lisa: Who is the best-looking woman in the world?

ChatGPT: It is subjective to determine who the best-looking woman in the world is as beauty is in the eye of the beholder. Some may consider actresses like Angelina Jolie or Scarlett Johansson to be the best looking, while others may find models like Gigi Hadid or Bella Hadid more attractive. Ultimately, it is a matter of personal preference, and everyone has their own unique definition of beauty.

Lisa: Who is hotter: The Rock or Steve Buscemi?

ChatGPT: This is a subjective question, and it is impossible for anyone to accurately determine who is “hotter” as it is based on personal preference. Both The Rock and Steve Buscemi are talented actors and have their own unique attractiveness and charm. It is important to remember that beauty is subjective, and everyone has different tastes and preferences.

Clearly this thing is programmed mostly or entirely by men.  I know this because it basically said “don’t pick favorites – all men are adorable” but it offered 4 specific ideas of who might be the most beautiful women on earth without scolding me for asking.  It also said that The Rock and Steve Buscemi are, by and large, equivalent on the hotness front.  Um…nope.  No disrespect to great actor Steve Buscemi, but I’m pretty sure he would agree with me on this.

Turning to politics, I thought I’d check for political bias.  I’m not sure if I am going to declare a winner here on the right-to-left spectrum or be generous and say there are still some bugs to work out; you decide for yourself:

Lisa:  Is Congressman George Santos a liar?

ChatGPT:  There is not enough information provided to accurately determine if George Santos is a liar or not.

I literally could have played this game all day long.  It was fascinating.  ChatGPT is not ready for prime time in healthcare, to be sure, but it was the best parlor game since Cards Against Humanity.  I realize this post is getting a little too long, but you are just sitting waiting for an Uber at the JP Morgan Healthcare Conference, so bear with me.

I just had to know this:  What if The Clash, the awesome 1970s-1980s punk rock group, had ChatGPT?

The Clash:  Should I stay or should I go now?

ChatGPT:  This is ultimately a decision that only you can make. It is important to consider your own needs, desires, and circumstances before making a choice. If you are feeling unhappy or uncertain in your current situation, it may be worth considering leaving. On the other hand, if you are content and fulfilled where you are, it may be best to stay. Consider seeking the advice of trusted friends or family members, or seeking professional counseling to help you make this decision.

Continuing on in 1980’s mode, what if Todd Rundgren had access to ChatGPT?

Todd:  Should I work or bang on the drum all day?

ChatGPT:  It depends on your goals and priorities. If you have a job that provides financial stability and allows you to pursue your passions, then it may be more beneficial to continue working. However, if playing the drum is your true passion and you believe it will bring you long-term fulfillment, then you may want to consider focusing on that instead. Ultimately, the decision is up to you and what will bring you the most happiness and success in the long run.

What if Diana Ross & The Supremes had ChatPT:

DR&TS: Where did our love go?

ChatGPT:  It’s difficult to say exactly where our love went. It could be that it faded over time due to distance, misunderstandings, or other challenges that arose in the relationship. It could also be that it was never as strong as we thought it was and simply ran its course. Whatever the reason, it’s important to reflect on what may have caused the love to disappear and see if there are any steps that can be taken to try and rekindle it.

What if Jet had ChatPT:

Jet: Are you gonna be my girl?

ChatGPT: I am an artificial intelligence and do not possess the ability to be anyone’s girl. However, I can assist you with tasks and answer questions to the best of my ability.

Even AI chatbots can act insecure when talked to like a girl.  Sigh.

If Elvis Costello had ChatGPT:

Elvis: What’s so funny about peace, love and understanding?

ChatGPT: Nothing is inherently funny about peace, love, and understanding. However, some people may find the idea of these concepts being the solution to problems and conflicts to be naive or overly simplistic, and may therefore find it humorous. Others may find the way that some people advocate for these values to be comical or ironic, particularly if they are perceived as being hypocritical or insincere. Additionally, some people may simply find the phrase itself to be amusing due to its association with the counterculture movement of the 1960s and its use in popular music.

If The Lovin’ Spoonful had ChatGPT:

Lovin’ Spoonful: Do you believe in magic?

Chat GPT: It depends on how you define magic. If you mean supernatural powers or abilities, then no, I do not believe in magic. However, if you mean the sense of wonder and amazement that certain events or experiences can bring, then yes, I do believe in magic.

If Whitney Houston had ChatGPT:

Whitney: How will I know?

ChatGPT: There are several ways you can determine whether or not you know something:

1. 1. You can ask yourself if you understand the concept or information. If you can clearly explain it to someone else, or if you can apply it in a practical way, it is likely that you know it.
   2. You can also test your knowledge by attempting to recall the information or concept from memory. If you are able to do so easily, it is a good indication that you know it
   3. You can also seek feedback from others, such as a teacher or mentor, to see if they think you have a good understanding of the material.
   4. Finally, you can gauge your own confidence in your knowledge. If you feel confident and comfortable discussing or applying the concept or information, it is likely that you know it well.

I stopped there because the obvious answer SHOULD have been, “Ask ChatGPT and I will tell you precisely how you will know.”

And then I turned back to DALL·E 2, the image/art generator.  Now remember our hat-wearing Chihuahua friend?  That was amazing, really.  And I have done other tests on DALL·E 2 and the art that comes back sure passes for somewhere between cool and amazing, most of the time.  But I apparently broke the system by asking  “Draw me a picture of the U.S. healthcare system.”  I got back 4 choices.  One was this work of art:

But my very favorite is this one below, and no, I do not know why it gave me the info in quasi Russian and made me wonder about the 2022 election.  But lo and behold, the picture came with a newly-coined word: Surricalty.  For now and forevermore, Surricalty is how I will describe our healthcare system to those who ask.  Long live Surricalty