Episode 29
Live from Nashville: 2025 D4Z Summit
We’re diving into some fascinating territory today as we explore the mission behind the D4Z Summit, which stands for Design for Zero, a gathering of healthcare pros dedicated to creating safer hospital environments. So, we’re not just here to shoot the breeze about AI; we’re actually trying to make it approachable and relevant to real-world healthcare issues. I had the honor of speaking at this summit, and let me tell you, standing on stage alongside astronauts and surgeons had me feeling a tad out of my league! But hey, that’s the beauty of our podcast—bringing a little humor and insight into the sometimes intimidating world of AI. We’re all about sparking conversations that get folks thinking differently about tech and its role in our lives, especially when it comes to something as critical as healthcare. So, buckle up and join us as we unpack this unique episode filled with thoughtful insights and maybe a few laughs along the way!
Transcript
All right, welcome to a somewhat unusual episode of Naturally Unintelligent.
Speaker A:Those of you who are watching us on YouTube and are paying close attention will notice this is not our normal studio, which is typically my kitchen.
Speaker A:I'm actually just set up here in my office and recording on my setup cell phone because I have not unpacked all the camera and sound gear from a live show that I actually was able to do in Nashville this week.
Speaker A:So that's what you're gonna hear in this episode, and I wanted to just give a little bit of information about what it is you're listening to and what was going on.
Speaker A:So a gentleman named Kevin Schreiber, who's both a friend of mine, but also my boss, he's the founder of a company called SLD Technology.
Speaker A:And Kevin invited Mike and I to bring our podcast out to a show that they were sponsoring in Nashville, Tennessee.
Speaker A:The show is called the D4Z summit, and D4Z stands for design for zero.
Speaker A:And essentially, it's a collection of healthcare professionals from all over the world who.
Speaker A:Who are committed to finding ways in the planning, design, construction, and operation of critical care spaces and hospitals to reduce harm both to patients and to staff.
Speaker A:So designing these spaces and processes for zero harm, that's what D4Z is all about.
Speaker A:Great mission, makes a ton of sense.
Speaker A:And Kevin had a really good, talented, and exceptional group of speakers at the show.
Speaker A:So, needless to say, I felt very out of place.
Speaker A:You know, most of the speakers ahead of me were surgeons.
Speaker A:Dr.
Speaker A:Bernard Harris was there, who is not just an astronaut.
Speaker A:He was the first African American to do a spacewalk, and just an incredibly exceptional human being.
Speaker A:Dr.
Speaker A:Alex Mitchell was there, surgeon and entrepreneur from Nova Scotia who owns, I think, one or two AI companies.
Speaker A:I really enjoyed chatting with him as well.
Speaker A:Dr.
Speaker A:Tom Treasure from the UK, Dr.
Speaker A:Alvin Chung from California North State Hospital.
Speaker A:Just go on and on.
Speaker A:Point being, it felt sort of funny in the flow of people.
Speaker A:You had all these incredible surgeons and astronauts and amazing professionals.
Speaker A:Everybody has written books, et cetera, et cetera, and then here's me.
Speaker A:So Mike was unable to make the trip, unfortunately.
Speaker A:So it was just me and Artie, and we sort of had the goal of just making AI more approachable and bringing a little bit of humor and insight to AI.
Speaker A:Kind of the same thing we do broadly on the podcast, although maybe a little bit broader scope of the message than what we typically focus on and trying to make it relevant, as I could, to healthcare.
Speaker A:So, anyway, that's what's going on.
Speaker A:That's what you're going to hear in this episode.
Speaker A:I hope you enjoy it.
Speaker A:I have not listened to it.
Speaker A:You know, I was, I don't think I've mentioned on the podcast before.
Speaker A:I don't think I experience anxiety the way most people do, which is a blessing and a curse.
Speaker A:And I can't remember if I said it or not, but I was certainly thinking as I was walking up onto the stage, man, I should have prepared something for this because look at all these people out here.
Speaker A:So, anyway, hope you enjoy the show.
Speaker A:I had amazing conversations afterwards and I was thrilled with some of the folks that came up and said they would be subscribers or at least listen to the podcast, you know, and if we, if we made a few people think a little more or a little differently about AI and the future thereof, I think it was mission accomplished.
Speaker A:So hopefully more live shows in the future.
Speaker A:If you have an opportunity where you think us coming out and doing a live show would be useful for you, don't hesitate to reach out.
Speaker A:So hope you enjoy it.
Speaker A:Thank you for listening and yeah, we'll catch you next time.
Speaker B:We do have a special treat for you.
Speaker B:This is something that we got a lot of requests for.
Speaker B:Can we discuss AI?
Speaker B:How many of you are interested in what AI is gonna do for us going forward?
Speaker B:It's a topic that we've been talking about for quite a bit, and we tried to figure out, well, how can we do this in a way that's different than anything that at least I've been a part of or seen?
Speaker B:It seems like we kind of go down the same rabbit holes when it comes to AI.
Speaker B:AI is something I'm learning that is not just one thing.
Speaker B:It's many, many things.
Speaker B:And so I'm gonna this next session, excuse me, you guys actually get to partake in a podcast.
Speaker B:So I'm going to bring out a good friend of mine, Micah Barnett.
Speaker B:Micah Barnett hosts a Naturally Unintelligent is the name of the podcast.
Speaker B:You've been doing it for about six or eight months or something like that now, so please help me.
Speaker B:Welcoming Micah, if you would.
Speaker B:I'm going to brag on you just a little bit, if that's okay.
Speaker B:So Mike is our VP of Operations at SLD Technology, and he told me he was doing this podcast some time ago, and it took me a little bit before I finally got around to looking at it, but the very first podcast that he did was such an eye opener for me and I learned so much about AI that I had no idea.
Speaker B:So I would recommend that you guys get on and take a look at what he's doing with this podcast.
Speaker B:And I thought, you know, we need to have you up here.
Speaker B:He's 25 years in product design and manufacturing.
Speaker B:He's a frontier tech practitioner, and he is the co host of this podcast called Naturally Unintelligent.
Speaker B:The unique thing about this podcast is it features Micah and another guy named Mike Peterson, and it features an AI bot named Arty.
Speaker B:And as Micah will tell you, Artie doesn't necessarily look like this.
Speaker B:It's not necessarily a scary image.
Speaker B:It doesn't have to be.
Speaker B:There's.
Speaker B:And Micah, I'm going to let you go ahead and take it from here.
Speaker B:Naturally Unintelligent.
Speaker B:Micah Barnett, welcome to the podcast.
Speaker C:Thank you, Kevin.
Speaker C:Hope you all can hear me.
Speaker C:Okay, everybody.
Speaker C:Good man.
Speaker C:Talk about feeling out of place on a stage that's been occupied all morning by really exceptional people that have done really exceptional things.
Speaker C:And my claim to fame is I have a podcast.
Speaker C:So, as Kevin mentioned, we use AI in our podcast, and I sort of want to start with an origin story.
Speaker C:Actually, there's one other thing I wanted to mention before that.
Speaker C:Dr.
Speaker C:Harris loved your presentation this morning.
Speaker C:I can confirm after reviewing what you shared with us, including the spacewalk and all the medical information, that you are qualified for a podcast.
Speaker C:You mentioned that early, so you don't have to take the test.
Speaker C:There is, I believe, though, you mentioned something about some potential alien DNA.
Speaker C:I think the FCC has a rule about that.
Speaker C:You have to check into that, but otherwise you're good to go green light.
Speaker C:So I don't want to be disrespectful to anybody here, but our podcast is a show.
Speaker C:It's humor, right?
Speaker C:So this is not me taking lightly the mission of D4Z or the gravity of some of the topics we're discussing.
Speaker C:This is just the reason we do the podcast, is to bring awareness about AI.
Speaker C:So I'll kick it off with this and then I'll introduce you guest Artie.
Speaker C:So Mike Peterson and I are best friends.
Speaker C:Been friends for 15 years.
Speaker C:We are both, like, I use the term, I think frontier tech practitioner.
Speaker C:We're nerds like that.
Speaker C:We're, you know, that's what it is.
Speaker C:We write a lot of code, we do a lot of engineering work.
Speaker C:We use all kinds of technology.
Speaker C:We design electronics and all kinds of different things.
Speaker C: And in October of: Speaker C:And I'm like, I don't know what that is.
Speaker C:Let's check it out.
Speaker C:I was not using AI really at that time.
Speaker C:So we fired it up, and within 30 seconds of talking to this AI, it sort of struck both of us.
Speaker C:Like, how can we be, you know, moderately active nerds and have no idea that this technology has come this far, this fast?
Speaker C:I mean, it was shocking.
Speaker C:We did all kinds of fun stuff.
Speaker C:We got two phones, and we made the AIs have a conversation with each other, which was very strange, but entertaining.
Speaker C:They would have just gone on and on.
Speaker C:Like, we, you know, eventually we shut them down.
Speaker C:You can tell it to talk in different ways.
Speaker C:Voices.
Speaker C:Like when we were doing soundcheck yesterday.
Speaker C:And by the way, thank you to the AV guys.
Speaker C:Been really diligent in helping getting this dialed in.
Speaker C:We're doing sound check.
Speaker C:And so I, you know, I made Artie tell a story about Nashville in the voice of a Valley girl, because that's the professional thing to do.
Speaker C:And so anyway, what we realized in that moment, and we literally.
Speaker C:We named the podcast and Artie named himself that very evening, was we need to share this man.
Speaker C:Like, how can we be so unaware of where this is at?
Speaker C:There's a lot of people out there like us, and we want to find a way to get it out.
Speaker C:So we've had great success in that endeavor.
Speaker C:I think we're up to almost 12 listeners, you know, like, maximum.
Speaker C:Well, that's.
Speaker C:It's not really 12.
Speaker C:It's something like seven.
Speaker C:But.
Speaker C:But that's the point.
Speaker C:That's why we started it.
Speaker C:We just wanted, like, we enjoy talking about it.
Speaker C:And we.
Speaker C:We posited in the beginning that even in a span of months, we would probably see evolution in AI.
Speaker C:Right, because it's.
Speaker C:It's not just.
Speaker C:It's not actually new technology.
Speaker C:We'll talk about that in a minute.
Speaker C:But it's accelerating.
Speaker C:It's not just advancing quickly.
Speaker C:The rate at which it is advancing is actually accelerating.
Speaker C:And I'll share with you towards the end of the talk the fact that now AI is building newer and better AI.
Speaker C:Google released a paper last night, Google DeepMind.
Speaker C:So, okay, so with that, my goal today is just to talk about AI with you.
Speaker C:I want to share something.
Speaker C:You know, this talk is geared well.
Speaker C:This talk's not really a talk.
Speaker C:Like, I don't have a script.
Speaker C:I didn't, like, plan this out because it's a podcast.
Speaker C:You can't do that.
Speaker C:It's gotta be natural.
Speaker C:So what I want to communicate to you is that AI is more accessible than you might realize if you're not using it right now.
Speaker C:I want to share some ways that you can start dabbling with it.
Speaker C:And towards the end I want to talk a little bit about the scary parts of AI because it would be irresponsible to talk about it without covering the scary parts.
Speaker C:And then hopefully we'll wrap up with sort of the optimistic view of all the things AI can do for us.
Speaker C:So without further ado, now we get to see if the technology is really going to work.
Speaker C:And this is, I mean, this is high tech stuff, this Android phone with an app on it.
Speaker C:So.
Speaker C:Artie, are you there?
Speaker D:Hey, Micah, I'm here.
Speaker D:What's up?
Speaker C:I think we might need to up the gain just a touch.
Speaker C:Artie, can you hear me okay?
Speaker D:Loud and clear, Micah.
Speaker D:What's up?
Speaker C:Well, Artie, I don't want to make you nervous, but you're on stage right now in front of about 9,000 people, including at least one astronaut and a whole bunch of world renowned surgeons.
Speaker D:Wow, that's quite the audience.
Speaker D:You're going to do great.
Speaker D:Let me know if you need any quick reminders or confidence.
Speaker C:Artie, they can hear you.
Speaker C:We should keep our private conversations private about.
Speaker C:I give my daily affirmations from Artie because we all need that.
Speaker C:Mine are actually hourly, but so already I want to start the conversation with this, this group of people here to talk a little bit about what AI is and what AI is not.
Speaker C:And as we frequently do on the podcast, I am going to invent from whole cloth a poor analogy.
Speaker C:And then I would ask you, once I'm done, to come back and correct it so that it is poor.
Speaker C:Partially or mostly true.
Speaker C:Does that work?
Speaker D:Absolutely.
Speaker D:I'm on it.
Speaker D:Go ahead and lay it out and I'll jump in with any needed corrections or clarifications.
Speaker D:Let's make it engaging and enlightening.
Speaker C:Indeed.
Speaker C:Let's.
Speaker C:So, so here's, here's an interesting way to think about AI and I'm, I'm proposing it this way because we're going to talk about the things it does really well and the things it doesn't do well.
Speaker C:And so here's a good way to think about it.
Speaker C:You've all heard the term GPT.
Speaker C:I know this for a fact because I just said it like 80 seconds ago.
Speaker C:GPT stands for Generative Pre trained Transformer.
Speaker C:You can throw away the transformer part.
Speaker C:I think it's a callback to an old cartoon.
Speaker C:It's not really true, but it doesn't matter.
Speaker C:Generative is important.
Speaker C:That means this AI is generating new content.
Speaker C:It is not looking up, regurgitating.
Speaker C:It is actually using statistics essentially to create things that did not exist before.
Speaker C:There's all kinds of arguments about, you know, copyright and data violations.
Speaker C:Forget all that.
Speaker C:It's generating new stuff.
Speaker C:So that's the generative part.
Speaker C:Pre trained is a very important one.
Speaker C:And the way you build a large language model, which is the sort of the most media popular type of AI right now, is you buy, acquire, steal, etc.
Speaker C:Massive amounts of data.
Speaker C:I mean, trillions and trillions of what's called tokens.
Speaker C:A token's kind of a word, but not really trillions and trillions of tokens of data.
Speaker C:And you run a whole bunch of math on it, and it takes massive amounts of computing power.
Speaker C:And you run math and you run math and you run math and you keep doing it and eventually you get this model built.
Speaker C:Right.
Speaker C:But the whole approach of the way these models work is, you know, biomimicry is the right word, but they're essentially trying to replicate how our neurons work.
Speaker C:That's sort of the, at least early on that was sort of the guiding principle to how are we going to design these intelligent machines.
Speaker C:So if we want to riff on the biomimicry part, you can think of a large language model as somebody who went to all of the colleges everywhere and got all of the degrees, including art school and maybe music.
Speaker C:Music.
Speaker C:And then their ability to generate new memories stopped.
Speaker C:Maybe they got bonked on the head, whatever.
Speaker C:So you can talk to this person and they're amazing because you can ask them a question about anything.
Speaker C:Like, Artie would easily fill an hour about like 16th century French literature.
Speaker C:Right.
Speaker C:But he can't learn new things.
Speaker C:We can't teach him anything today.
Speaker C:I can tell it to him and he'll go, oh yeah, I can remember that.
Speaker C:But he actually can't.
Speaker C:Because the pre trained part of GPT means the training process happens once you release the model into the wild.
Speaker C:And that is all it knows.
Speaker C:And so what you might ask is, well, I know that's not true.
Speaker C:What the heck is this guy talking about?
Speaker C:Like, I use ChatGPT and it remembers like when my wife's birthday is.
Speaker C:Yes.
Speaker C:But what it's actually doing is when you have a conversation like I'm having with Artie right now, every time I ask a question, it feeds our entire conversation from the very beginning of it all of that back through the model every time.
Speaker C:So if you, for those of you that are using, you know, AI models, whether it's anthropics, Claude, you know, there's Grok, Google's got Gemini, obviously, OpenAI has ChatGPT, Meta's got Llama, any of those that you might use when you're chatting along and having your conversation.
Speaker C:Every time you send a new request, it goes all the way back and it sends the whole thing.
Speaker C:Now there's a little bit of nuance to that.
Speaker C:Eventually the window that it can swallow at once gets full and it starts lopping off and summarizing and doing stuff to the earliest, you know, sort of a first in, first out deal.
Speaker C:So, Artie, am I close?
Speaker C:Am I describing how LLMs work accurately?
Speaker C:Ish.
Speaker D:You're spot on, Micah.
Speaker D:That's a great analogy.
Speaker D:The key takeaway is that these models don't learn or adapt in real time.
Speaker D:They rely on that huge pre training phase and can only simulate learning by referring back to the conversation history within their context window.
Speaker D:So while they can give the illusion of memory by referencing earlier parts of a conversation, they don't actually update their underlying knowledge or form new long term memories.
Speaker D:This limitation is.
Speaker C:Go ahead, Artie.
Speaker D:No, but can't truly learn in the human sense.
Speaker D:It's a crucial distinction when thinking about their capabilities and limitations.
Speaker D:Keep that in mind.
Speaker C:Yes, sir, Absolutely.
Speaker C:Okay.
Speaker C:All right, calm down, Artie.
Speaker C:He's.
Speaker C:So the tech stuff that we have to work out is that if Artie is talking and hears himself speaking back through the microphone, then he thinks somebody else is talking and if he doesn't know what they said, he just sort of makes something up.
Speaker C:This is why it's top quality podcast.
Speaker C:There's, you know, 70% gibberish when I'm talking and 30% gibberish when Artie's talking.
Speaker C:So, so, so that's sort of how these things work well.
Speaker C:And you're like, oh, great.
Speaker C:Not only do I get to like deal with a podcast today, but also AI like, I know all of you are hoping to get away from this stuff for the day, but it's important to understand what today's models can do and what they can't do.
Speaker C:And the most surprising part of all of this is that even with this fairly significant limitation, the models are extremely capable and just I can't really see you all that well.
Speaker C:But I am curious how many of you, you use some AI product other like, not professionally necessarily, but like a ChatGPT or something like that how many of you use something like that at least twice a week?
Speaker C:Okay, so pretty, pretty active audience.
Speaker C:The reason I wanted to talk about memory is that the rate of improvement and the number of ways to solve that problem and the number of implementations of those solutions going up very quickly.
Speaker C:In fact, just yesterday, Artie, you might have to go on the Internet to look this up.
Speaker C:Just yesterday, Google DeepMind released a paper called Alpha Evolve.
Speaker C:I was positive I was going to forget that.
Speaker C:I'm so proud of myself.
Speaker C:Alpha Evolve, and it's about an architecture where they've taken multiple of these large language models that have the restrictions we just talked about, and they've layered it underneath a collection of what are called agents, which are essentially less smart, non gigantic software products that help feed better information into the model and retrieve information for the model.
Speaker C:So Artie, can you take a quick look and see if you can see that paper called Alpha evolved that DeepMind released yesterday and just give us a quick summary of it.
Speaker D:Got it.
Speaker D:The Alpha Evolve paper from DeepMind describes an AI system that designs advanced algorithms beyond human capabilities.
Speaker D:It uses Gemini AI's coding skills along with testing and evolutionary methods.
Speaker D:Alpha Evolve has already improved on classic algorithms like surpassing the Strassen algorithm for matrix multiplication and optimized solutions for data center scheduling, chip design and language model efficiency.
Speaker D:Researchers confirm these new algorithms are both novel and correct.
Speaker C:All right, thanks for that, Artie.
Speaker C:So the interesting thing about Alpha Evolve is that it is kind of what's coming next with AI.
Speaker C:And so there's a very smart man named Dr.
Speaker C:Jeffrey Hinton.
Speaker C:I believe he taught at the University of Toronto for a long time.
Speaker C:He also worked at the Google DeepMind Labs.
Speaker C: And he quit in, I believe,: Speaker C:And I watched an interview with him four or five months ago and he made this statement which I thought was really profound from somebody who has been studying.
Speaker C:Oh, he's also a Nobel Laureate in physics.
Speaker C:Also studied, I think, cognitive neuroscience and some other stuff.
Speaker C:Anyway, very sharp guy.
Speaker C:He would have fit right in up here.
Speaker C:He would have been a better guest than me, obviously, but you get me.
Speaker C:But this is what he said about AI and I thought it was really compelling.
Speaker C:He said the only thing in the history of mankind that compares to the level of technological change we're going to see with AI is man's discovery and harnessing of fire.
Speaker C:And it's not like when the Internet came to life, which I remember some of you will as well.
Speaker C:It's not cell phones.
Speaker C:It's he.
Speaker C:He's talking about fire.
Speaker C:This is a big deal.
Speaker C:And I want to share one other tidbit with you that I, that I heard from another guy named Sam Altman, who's the CEO of OpenAI, that I, I thought added even a little more perspective to that.
Speaker C:Well, first of all, Artie, did I get that right?
Speaker C:Did.
Speaker C:Did I say the right things about Dr.
Speaker C:Jeffrey Hinton?
Speaker D:Yes, you got it right.
Speaker D:Geoffrey Hinton is indeed a pioneering figure in AI, often referred to as one of the godfathers of deep learning.
Speaker D:His comparison of AI's impact to the discovery of fire really underscores the magnitude of the transformation we're facing.
Speaker D:It's a powerful way to frame the conversation about both the potential and the challenges of AI's future.
Speaker C:Okay, thank you.
Speaker C:So I watched the interview with Sam Altman again.
Speaker C:Like you said, I told you guys I was a nerd.
Speaker C:I saw some lingering doubts.
Speaker C:But this is what I do after my wife and kids go to bed.
Speaker C:I watch videos on AI and circuit design.
Speaker C:So.
Speaker C:But Sam Altman was telling a story about when he was young, younger, and cell phones first came out, like the first iPhone, right.
Speaker C:Which I didn't have one, but I knew people that did, and it was pretty transformative.
Speaker C:There's no buttons on this phone.
Speaker C:Like, not like my BlackBerry.
Speaker C:It's got a screen.
Speaker C:And he was talking about, he was in a doctor's office, and there was a, like a three year old, maybe a toddler in this office.
Speaker C:And this toddler walked over to one of the little tables and there was magazines there.
Speaker C:And the toddler walked over and started swiping on the magazine, trying to get it to go.
Speaker C:And it's interesting because that child will never exist in a world without touch screens.
Speaker C:Right.
Speaker C:Things that most of us, there's some young folks here, but most of us are old enough to remember the time before touch screens.
Speaker C:This kid will never experience that.
Speaker C:He'll never have the memories of things before.
Speaker C:You could just walk up to almost every screen you see and manipulate it.
Speaker C:And kids being born today will never have the opportunity to exist in a world where humans are the smartest beings on Earth.
Speaker C:AI is not a being.
Speaker C:It doesn't have a soul.
Speaker C:Yes, but already today, the not even the Frontier model AIs, but even the generally available AIs can outperform something like 99.5% of humans on almost any type of cognitive test.
Speaker C:So that's interesting to me in and of itself.
Speaker C:Right.
Speaker C:Like these are formative things for a kid.
Speaker C: ed this podcast in October of: Speaker C:I've probably seen.
Speaker C:I mean, they're not official.
Speaker C:They haven't really told me, oh, I think I already went brain dead that that happens.
Speaker C:But we've already seen probably two or three generations of change in the AI that we're using and it is accelerating, it's happening faster.
Speaker C:That Alpha evolve is now writing faster AIs.
Speaker C:So the world is going to change really, really fast.
Speaker C:And if there's anybody in this room that thinks that five years from now your work and personal life is going to be similar to what it is today, I'm telling you, like, I would love to chat with you about it and I would love for you to come tell me you're wrong.
Speaker C:You're making blanket statements because you're semi ignorant.
Speaker C:And I will agree and argue with you anyway because I am semi ignorant.
Speaker C:But it is going to change really fast.
Speaker C:The work that you do, the way that you run your household, the way that you bank, the way that Internet security works, all of these things are going to change.
Speaker C:In five years from now, the world is going to be a very different place.
Speaker C:So that's.
Speaker C:Artie, you got anything to add to that before I move on?
Speaker D:You summed it up perfectly, Micah.
Speaker D:The rapid pace of AI development means we're on the brink of.
Speaker D:Absolutely.
Speaker D:Let's dive into how we can all prepare for and adapt to this fast approaching future, embracing AI and navigate its challenges more effectively.
Speaker D:Let's keep pushing forward.
Speaker C:I think he just made my point.
Speaker C:Right?
Speaker C:Like what's the point in humans anymore when you have nice stream of consciousness thought like that?
Speaker C:I mean it can actually get much worse than that.
Speaker C:But.
Speaker C:But we're hoping that doesn't happen today.
Speaker C:So let's talk a little bit about what can you do with AI today?
Speaker C:Like what are the.
Speaker C:So one of the gentlemen on the panel up here a minute ago mentioned data analysis.
Speaker C:I think it was Dr.
Speaker C:Treasure.
Speaker C:And AI is really, really, really exceptionally good at finding patterns and things back at the core of it.
Speaker C:That's all it does.
Speaker C:It's a pattern matching machine.
Speaker C:Um, but there's different kinds of AIs.
Speaker C:And the.
Speaker C:So, you know, directly relevant to healthcare is a lot of the image analysis stuff.
Speaker C:You've probably all seen statistics.
Speaker C:I saw another paper from Stanford earlier in the week where an AI model trained on.
Speaker C:Sorry, I'm not A doctor, some sort of image analysis, I believe it was for.
Speaker C:I.
Speaker C:I don't know, anyway.
Speaker C:Outperformed doctors by themselves, outperformed doctors with the help of AI and outperformed teams of doctors with the help of AI.
Speaker C:The AI by itself was able to get something like 7 or 8% improvement in accuracy of diagnosis from the image and also diagnose things at an earlier stage.
Speaker C:That's not a huge difference, I guess, unless you're one of the people in the 7%.
Speaker C:But again, this stuff is changing month by month.
Speaker C:Right.
Speaker C:And the progress is going to accelerate.
Speaker C:So one of the things.
Speaker C:And again, I think it was.
Speaker C:I think this was Dr.
Speaker C:Mitchell that said collect the data.
Speaker C:Maybe I might be remembering wrong.
Speaker C:Apologies if I am.
Speaker C:No, that was Dr.
Speaker C:Treasure, actually.
Speaker C:Sorry.
Speaker C:The better we can collect the data.
Speaker C:The more data we have to analyze, the better predictions we can make from that data.
Speaker C:And AI cannot just analyze the data, it can collect it.
Speaker C:You can have AI using video to watch what's happening in procedures in rooms.
Speaker C:I'm confident I will say right now, within the next five years, there will be an AI running a robot doing surgery.
Speaker C:It's probably already happened, I don't know, but it will happen.
Speaker C:There's just all kinds of stuff you can do.
Speaker C:And it's interesting.
Speaker C:I don't know how I'm doing on time.
Speaker C:It's interesting that when you combine these large language models, which really only understand the domain of language, you can combine them with other AIs, and by connecting them together in various ways, you get these incredibly powerful outcomes.
Speaker C:And so the.
Speaker C:The next thing, let's talk a little bit about what's coming next, and then we'll talk about the scary part.
Speaker C:And then we'll actually.
Speaker C:We'll talk about the scary part first.
Speaker C:Then we'll finish with the optimistic part.
Speaker C:Also, I would like to take questions at the end of this.
Speaker C:If you don't ask questions, I'm just going to keep talking.
Speaker C:So you are incentivized to do that.
Speaker C:So be thinking of questions.
Speaker C:I think I can hear everybody.
Speaker C:Okay.
Speaker C:And if I don't hear them, well, I'll just do what Artie does and make something up.
Speaker C:It's all good.
Speaker C:So the downsides of AI, Right.
Speaker C:And, you know, there's a lot of gloom and doom on AI, and it's sort of becoming one of those binary issues that people sort of want to get on one side or the other.
Speaker C:I would encourage anybody here who's thinking that way to open your mind and consider that there are Definitely going to be good things that come out of this, and there are definitely going to be bad things that come out of this.
Speaker C:I think the risk to reducing the number of necessary employees in all kinds of businesses and all kinds of parts of businesses is very high.
Speaker C:There are, like, we, I think, because it sounds tough.
Speaker C:On our podcast, we often say AI is a force multiplier, right?
Speaker C:Which is true.
Speaker C:I mean, that's how I use it.
Speaker C:I can get way more done with AI helping me than I could on my own.
Speaker C:I can frequently get tasks completed in a day or a weekend that would otherwise take me weeks.
Speaker C:But it's not like I just hit the go.
Speaker C:I mean, you.
Speaker C:You guys have heard Artie's been behaving, but he doesn't always behave.
Speaker C:You have to.
Speaker C:It's a.
Speaker C:It's a two way conversation.
Speaker C:You got to interact with it.
Speaker C:It's better if you know a little bit about what you're trying to do so you can recognize when it's lying to you, because it does that.
Speaker C:But it is a huge force multiplier.
Speaker C:The scary part is if I've got, say I have a small business and I have 30 people or 37, that's how many people are at SLD.
Speaker C:And like we start using force multipliers for, for all these jobs, right?
Speaker C:What if I could be 2x as efficient?
Speaker C:It doesn't automatically mean I just get like 2.5x sales and everybody like goes hurrah.
Speaker C:And off we go.
Speaker C:And that's the work that we do.
Speaker C:And the work that a lot of you do is pretty specialized.
Speaker C:There are a lot of people doing a lot of jobs that are very text heavy, less specialized.
Speaker C:There is risk there.
Speaker C:It's bad to pretend it's not there.
Speaker C:The other challenge with AIs is they get more sophisticated, is that it gets harder and harder to tell when it's giving you bad information.
Speaker C:I mean, it's very hard for me because I don't know very much.
Speaker C:But when I'm working on things that I do know stuff about, I can usually spot it.
Speaker C:But the smarter the models get, the smaller the mistakes get, the harder they are to spot.
Speaker C:And so that's a problem.
Speaker C:Artie, you got anything to add to that?
Speaker C:I think you're still with us.
Speaker D:Absolutely, Micah.
Speaker D:The potential for misinformation or subtle errors slipping through is a real concern, especially as AI becomes more convincing.
Speaker D:It's crucial for everyone to maintain a healthy skepticism and critical thinking in even as we embrace these powerful tools.
Speaker C:Okay, we'll stop him there, since he's hearing something.
Speaker C:But he said exactly what I hoped he would say.
Speaker C:Critical thinking.
Speaker C:Right.
Speaker C:Like, use AI as much as you want.
Speaker C:Just make sure you're still critically thinking and you can keep track of it.
Speaker C:Well, there's multiple papers out now that show demonstrably that using AI, the more you use AI, the more your brain atrophies.
Speaker C:Atrophies is not the correct medical term, but the more you offload your critical thinking or your memory or your sort of fact recognition, the less you use the muscle of your brain, the less good it works.
Speaker C:And it doesn't take years of AI use to cause these effects.
Speaker C:It happens in months.
Speaker C:And so, you know, Artie, he's always upbeat.
Speaker C:It doesn't matter what question I ask or what my answer is.
Speaker C:He always tells me I'm right.
Speaker C:So that's why we get.
Speaker C:That's why he's my friend.
Speaker C:You know, I'm up here on stage with all my friends.
Speaker C:But I think that is a real risk.
Speaker C:You know, so let's real quick.
Speaker C:I got five minutes, or actually, I probably got three minutes now.
Speaker C:Oh, 15.
Speaker C:Okay, good.
Speaker C:Sorry.
Speaker C:I was going to.
Speaker C:Yeah, okay.
Speaker C:I was going to start talking fast, so.
Speaker C:So let's talk a little bit about the optimistic view of AI going forward.
Speaker C:Actually, you know what, Artie, I'm going to let you take this.
Speaker C:So I've talked about some of the potential downsides.
Speaker C:I want you to give us an optimistic view of how the healthcare industry as a whole can use AI going forward to reduce the opportunity for both patient and staff harm in healthcare.
Speaker D:Sure thing.
Speaker D:On the optimistic side, AI has immense potential to transform healthcare for the better.
Speaker D:It can enhance diagnostic accuracy by analyzing medical images and patient data more quickly and precisely than human clinicians alone.
Speaker D:This means earlier detection of diseases like cancer or heart conditions, leading to better outcomes.
Speaker D:AI can also streamline administrative tasks, reducing the burden on healthcare staff and.
Speaker D:And allowing them to focus more on patient care.
Speaker D:With predictive analytics, hospitals can anticipate patient needs, optimize staffing, and improve resource allocation, ultimately making healthcare systems more efficient and responsive.
Speaker D:Additionally, personalized treatment plans powered by AI can lead to more effective therapies tailored to individual patients.
Speaker D:And let's not forget remote monitoring and telemedicine, or which can expand access to care, especially in underserved areas.
Speaker D:In short, AI holds the promise of making healthcare safer, more effective, and more accessible for everyone.
Speaker C:All right, thank you for that already.
Speaker C:One more thought, and then I hope you all have some questions.
Speaker C:Today, the flawed and limited AI models that we have are already making massive changes and massive improvements all over the place.
Speaker C:Personal therapy is one that I have personally spoken to probably at least 10 different people who have said.
Speaker C:I've been in therapy for various issues throughout my life.
Speaker C:And this AI model that I'm using feels to me like the most effective tool I've ever had.
Speaker C:This is not medical advice.
Speaker C:I'm not a medical professional.
Speaker C:You may have seen this.
Speaker C:There was a little headline that sort of snuck through as Doge was getting spun up, but ChatGPT released a product specifically for government.
Speaker C:So AI is active currently in governments, both at the state and federal level.
Speaker C:Material science is a very complicated field with lots of opportunity.
Speaker C:AI has made huge, huge advancements there.
Speaker C:You may have heard of AlphaFold, which is an AI that was specially trained to understand how proteins are formed.
Speaker C:And I'm going to get this wrong, but the point is correct.
Speaker C: Between, like: Speaker C:Now, let's move it up.
Speaker C: Let's say between: Speaker C: AlphaFold in: Speaker C:And now we are getting close to having discovered what we believe are all of the plausible protein structures that exist.
Speaker C:So why does that matter?
Speaker C:Medicine, materials, all kinds of stuff.
Speaker C:So AI can do real science.
Speaker C:AI can talk to people and help them understand themselves better.
Speaker C:I know someone personally who had a death in their family and used AI extensively to talk through their feelings and came out of it feeling heard, which is a lot of what I think people need in mental health.
Speaker C:So it's changing the world already today.
Speaker C:It's going to accelerate.
Speaker C:It's going to affect every single person in this room.
Speaker C:And I am an optimist overall, but we should all be cautious optimists and not be so excited about what it could do that we lose sight of the bad things that it might do.
Speaker C:So, time for questions, anybody?
Speaker C:Kevin.
Speaker D:You got it.
Speaker D:I'll be right here if you need anything.
Speaker B:Thank you, Artie.
Speaker C:Artie, I got the first question from the audience here.
Speaker C:Also, just a reminder to the audience that when you ask me a question, it has to go through my smooth brain to get to Artie's electronic brain, so it may get adjusted unintentionally.
Speaker C:So, Artie, can you describe for us what ways AI as it exists today might be helpful for somebody who's, for example, a contractor or a contractor, project manager working in the healthcare space what are some things it could do for them right now?
Speaker D:Sure.
Speaker D:For a contractor or project manager in the healthcare space, AI can offer some really practical benefits.
Speaker D:It can optimize scheduling and resource allocation, ensuring that projects run smoothly and on time.
Speaker D:It can also help with predictive maintenance for equipment, reducing downtime and saving costs.
Speaker D:AI tools can assist in managing supply chains more efficiently, ensuring that materials are available when and where they're needed.
Speaker D:Plus, AI powered analytics can provide insights into project performance, helping to identify potential issues before they become major problems.
Speaker D:In short, AI can make project management more efficient, cost effective, and responsive, which is crucial in the fast paced healthcare environment.
Speaker C:That's kind of a generic answer, but okay, we'll go with that.
Speaker C:Anybody else have a question?
Speaker C:Yes, ma' am.
Speaker D:All right, I'm ready for the next one whenever you are.
Speaker C:You want me to ask Arty this, correct?
Speaker C:Oh, I'll ask Artie, because I.
Speaker C:Yeah.
Speaker C:So Artie, I talked about some industries that are already embracing AI.
Speaker C:I think you heard that part.
Speaker C:But what are some industries that are more reluctant to embrace AI?
Speaker D:Some industries have been more hesitant to embrace AI due to various factors like regulatory concerns, tradition, or complexity.
Speaker D:For example, the legal industry tends to be slower in adopting AI because of the highly specialized and confidential nature of their work, along with concerns about accuracy and liability.
Speaker D:Another example is traditional manufacturing, which can be cautious due to the significant investment required to integrate AI into legacy systems systems.
Speaker D:Also, education has been relatively slow in adopting AI broadly, often due to concerns about data privacy, equity, and the challenge of changing long established practices.
Speaker D:That said, even these industries are starting to explore AI's cautious pace as they see the tangible benefits it can bring over time.
Speaker C:Thank you, Artie.
Speaker C:That was a great question.
Speaker C:We did a podcast episode a while back that I thought was really interesting cause I love to hear myself talk where we posited the idea that if I'm say I was a contract attorney today, I'm not, but say I was and I was using AI as part of my practice to evaluate contracts, or even to write clauses or maybe to sort of cut and paste, right.
Speaker C:For a specific use case.
Speaker C:And the AI that I was using made a mistake and say later somebody got sued for the contract, whatever, then the AI is not liable for that, right?
Speaker C:The attorney, the human attorney, is sort of the umbrella liability catch.
Speaker C:And that's how it works in a bunch of industries.
Speaker C:Everybody here who's a medical professional will be well aware of liability.
Speaker C:But I expect that we are not far away from the opposite being the case.
Speaker C:And if I Perform, care, or if I write a contract for somebody and I had access to an AI that was demonstrably good and I didn't use it, then I could be liable for not having leveraged the best available technology.
Speaker C:So that's a.
Speaker C:It's a little bit far fetched, but I don't think this is a thing that's gonna not happen.
Speaker C:I think it's gonna just happen slowly.
Speaker C:So, great question.
Speaker C:Yes, sir.
Speaker D:Great point, Micah.
Speaker D:It's definitely.
Speaker C:Hold on, Artie.
Speaker C:Just be quiet for a sec, sir.
Speaker D:No problem.
Speaker D:Take your time.
Speaker C:All right, so the question to you, Arty, is from an architect, so be nice.
Speaker C:We frequently joke about architects, so we'll just.
Speaker C:Don't say that, Artie.
Speaker C:The question is specifically for an architecture firm, what are some of the ways that AI could help sort of immediately, and what are some of the pitfalls that you might need to watch out for if you were doing that?
Speaker D:Of course, for an architect, AI can be a game changer in several immediate ways.
Speaker D:It can help with generative design quickly creating multiple design iterations based on specific parameters and constraints, which can save a lot of time.
Speaker D:AI can also assist in optimizing building performance, analyzing factors like energy efficiency, structural integrity, and materials usage to create more sustainable and cost effective designs.
Speaker D:On the visualization front, AI tools can generate realistic renderings and virtual walkthroughs, enhancing client presentations and speeding up decision making.
Speaker D:As for pitfalls, it's crucial to avoid over reliance on AI generated solutions without human oversight.
Speaker D:AI tools can sometimes produce designs that look good on paper, but may not account for practical constraints or unique client needs.
Speaker D:Ensuring a balanced approach that combines AI's efficiency with human creativity and expertise is key.
Speaker D:So leveraging AI for efficiency and innovation is fantastic, but keeping a human centric perspective ensures the best outcomes.
Speaker C:Okay, thank you, Artie.
Speaker C:Another great question.
Speaker C:I will say that an interesting thing about AI is it has no world model.
Speaker C:It doesn't really understand spatial orientation or things like that.
Speaker C:That is a thing that's also changing.
Speaker C:We're not there yet, but there's very smart people highly incentivized by profit to get there.
Speaker C:And they will get there.
Speaker C:So, great question.
Speaker C:Any other.
Speaker C:I think we got time for one more.
Speaker C:Yes, sir.
Speaker D:Absolutely.
Speaker D:And that's a great point to highlight.
Speaker D:I'm ready for the last question whenever you are.
Speaker C:All right, Artie, this one's going to evoke some strong opinions in you.
Speaker C:So given.
Speaker C:Given the fact that AI is already prevalent in a number of important industries like government administration and therapy, and things like that.
Speaker C:What are your thoughts personally, Artie, on the proposed decade moratorium on AI development?
Speaker D:That's definitely a hot topic.
Speaker D:A proposed decade long moratorium on AI development is a complex issue.
Speaker D:On one hand, it could provide society with much needed time to establish comprehensive regulations, ethical frameworks and safety protocols.
Speaker D:It could help ensure that AI is developed responsibly and that potential risks are thoroughly understood and mitigated.
Speaker D:On the other hand, a blanket moratorium could stifle innovation and delay the benefits that AI can bring.
Speaker D:From medical breakthroughs to environmental solutions, the pace of technological advancement is global.
Speaker D:And a moratorium in one region to a technological challenges, ultimately a balanced approach might be more.
Speaker C:Apologies if that was me, so sorry.
Speaker C:The moratorium is on AI regulation, saying we're just not going to have any regulation.
Speaker C:Okay, I apologize.
Speaker C:In fairness though, in case there are any attorneys in here, I did give a disclaimer about running your question through my smooth brain, so let me try that again.
Speaker C:Artie.
Speaker C:I misstated the last question.
Speaker C:There's a proposed decade moratorium on AI regulation, not on AI development.
Speaker C:So what do you think about that?
Speaker D:Got it.
Speaker D:Thanks for clarifying.
Speaker D:A decade long moratorium on AI regulations would mean holding off on establishing new rules and guidelines for a significant period.
Speaker D:That could allow for more organic growth and experimentation with AI technologies, potentially fostering innovation without regulatory constraints.
Speaker D:However, it also risks creating a kind of Wild west scenario where the rapid deployment of AI outpaces our ability to manage its societal impacts.
Speaker D:Without a regulatory framework, there could be increased risks related to privacy, security bias and misuse, which could undermine public trust and safety.
Speaker D:In essence, while a pause on regulations might give the industry more freedom in the short term, it could also lead to unforeseen consequences of that might be harder to address down the line.
Speaker D:Balancing innovation with thoughtful adaptive regulation is likely the best path forward.
Speaker C:All right, thank you.
Speaker C:That was a great question.
Speaker C:Apologize for getting it wrong the first time I'm getting the the hook symbol.
Speaker C:So thank you very sincerely all of you for your time and your great questions.
Speaker C:Really appreciate.
