STUNNING Medical AI Agents OUTPERFORM Doctors 🤯trained in the simulation, continuous improvement. 

AI doctors can't get here soon enough. Human doctors do the absolute minimum, and I'm tired of paying money just to have my health brushed aside. I'll take a 3.5 doc any day.

As a doctor and AI enthusiast I believe LLMs have the potential to drastically outperform human doctors. They can compile all data and experience from around the world and throughout history with rapid real time updates. Cheaper too. I welcome this! It’s not just doctors that can be replaced. Most professionals can be bettered as well: lawyers, architects, bankers, engineers etc are all becoming obsolete. Like when backhoes replaced ditch diggers. AI is rapidly trivializing what humans are so proud of. Intelligence , creativity, pattern recognition etc.

No, I wouldn't just leave. I'd gladly chat with the AI doctor. Think about it: no waiting, ask any questions, no stress on the doctor, and imagine the vast knowledge accessible instantly, equivalent to 170,000 years of reading for a human.

Brilliant line at 8:27: "So on one hand I don't have too much time to delve into this, if you will, but on the other hand talking about this new paper would probably make a lot of those nasty emotional people very, very mad, which I just cannot say no to."

They're mad because they're coping. They see the writing on the wall, they don't want to lose their livelihood. Adapt or die, welcome to the age of ai. Onward to the singularity.

Even before complex agents, I asked Claude the other day about a fairly rare illness I had where my symptoms were varied from the usual symptoms of this already rare conditions and ol' Claude successfully diagnosed the condition. Even asked me a whole bunch of relevant intelligent questions to assist in the diagnosis. Was really wild to see. People joke about "Dr Google" and the perils of self--diagnosis, but at a minimum all doctors will have an AI assistant looking over their shoulder before much longer.

This is the biggest light bulb moment I've had yet Wes. Having a simple (

I told my friend this exciting news. The first thing they said was oh just like how all the other ai agents trained in simulations fail these surely did too. I told them so far it’s working far better than humans in simulation. And they basically said when it goes to reality it will fail. So many people in denial out there it’s shocking. Most people I know have a similar reaction, the expectation of failure and a belief this will never have an impact on their life. Idk how to get through to anybody how ground breaking these things are.

All med facility are hospital. No separation of dentist, oncologist, internal organ. One Stop Shop. 4m x4m room with 30 robot 'arm'/appendage. Verbal exchange with LLM, mulitple robot arm poke, take blood, test eye, analyze disease odor (yes chemicals given off by the body can isolate disease- LookItUp) weight scalel in the floor, etc etc etc. AGI within 18 minutes of analysis ascertains what cells in body are failing, bill is processed through phone in pocket. Done. Like it.

If you don't have medical insurance, you only get GPT-3.5 Agent Hospital

Are we all just being trained in a simulation to perform a function in another world?😮

my luck greatly increased by liking the video . thanks

This can’t happen soon enough. As a sufferer of an undiagnosed illness who has gotten nowhere with doctors, this will be a game changer. Most people think highly of doctors until they get sick, then they realize how incompetent they are. GPT-4 has done more to help with my illness than thousands of dollars worth of doctors.

Now imagine this but with GPT5 with multimodal and action and visualization at any step - it would be a crime NOT to have this at least attached to every doctor as an aide (and eventually running the show)

Those people aren't actually angry, they're scared of what it means if these studies are accurate

AI taking over day to day tasks makes my wife very uncomfortable and it stems from simply losing that flesh to flesh understanding of how we are all suffering together. She works in "social work" and works with veterans and simply cant' understand how a robot could help someone like me who also happens to be a disabled veteran. But I've never talked to a therapist looking to make a connection, I talk to them so they can just tell me what I'm messing up and then give me several things I could do to solve those issues. I understand the part about us all being in it together and I can clearly see the part that gives her pause. Each of us clearly understands that we are very social creatures and trying to be social with a robot is just strange to think about for most people. It has nothing to do with them thinking they will lose their job, or that a robot will replace friends or anything like that. It's just a fear of losing such a crucial part of their being. We all need to socialize, we all trust people who are educated and have the resources to help us, (and show some modicum of intelligence...). We need that stuff and AI taking over threatens to replace a lot of the stuff we are very comfortable with. That being said... From my point of view as a disabled veteran. humans are the fucking problem to begin with. If AI lets me live out my life before it takes over and kills us all, sign me up. I'd like it to let my kids and grandkids live too but beggars can't be choosers. :) (This is a joke. I'm not a monster, just a sick sense of humor.)

Fascinating. I embrace the use AI across multiple disciplines. I see too many humans having sub-par performance when it could be easily automated.

Every update gets us a step closer to Agent Smith

People are so averse to it because they think without a job the world ends, I'm more inclined to believe life begins to get even better If there is no need for a traditional "job"

Aggressively throws your package just like human workers.

I recently overheard my cousin and uncle discussing AI systems that are now being used in restaurants, where the food is entirely prepared by AI without any human involvement. I've been following the developments in AI closely, so am already aware of this. When I mentioned that the food might actually taste better, they were both skeptical. They responded, 'That's impossible, and we wouldn't eat there anyway.' To me, this just illustrates the level of ignorance most people show towards AI. The problem with people in sectors like the medical field, is that yes, it will take a doctor over 2 years to see 10,000 patients. They will surely learn a lot in that time, however, the human brain has a tendency to forget things. After AI (maybe not the current AI versions but definitely in the next few years) has simulated 10,000 or even 10,000,000,000 patients, it still has access to each and every one of those instances with 100% recall. It will be able to distinguish between even the slightest of nuance which will enable the AI system to effectively and accurately diagnose all medical issues with an absolute precision that human beings aren't capable of.

As a young person, I can understand a lot of the fear/distrust people have for these technologies- as beneficial as they might be. I think gen z has a very love/hate relationship with technology, as most of us are at least somewhat addicted to screens/entertainment. We’ve been raised around very sensationalistic messages, from cable to news to social media feeds. We’re raised (with good reason) to be skeptical of narratives. As much as I personally enjoy the “Ai-sphere” of content, I think most people see it as overhyped, and driven by profit seeking corporations. The idea of letting companies and governments develop such fundamentally revolutionary technologies with little to no transparency is scary for most. All that and the fact that your 200k degree might be worthless in a few years is just a lot to deal with. Anxiety will always be a factor in any kind of change. That being said, I think more folks will change their mind when tangible improvements (like new medicines or cheaper healthcare) come around. Cheers!

If the AI did nothing but fill out all the insurance forms and paperwork a typical doctor is stuck with, they'd be worth their weight in gold.

The bar for being better than human doctors is very low at times. My wife had endometriosis and Poly-something Ovarian Syndrome for years before they got diagnosed. Years of being in great pain. They gave her pain killers at first, then they started treating her like an addict. When they finally diagnosed the problem and she had surgery it solved the problem and she stopped having pain. And lost interest in taking painkillers. I'm still mad about the way the doctors treated her. It was pathetic.

I think that there's 3 factors at play here in regards to the knee-jerk anger at AI that from your (and to a degree my) perspective is so hard to grasp, and that's point 1. Most people haven''t even interacted with ChatGPT and If they have, went into it expecting it to be infallible or assumed it had way more capabilities than it does, which means.. 2. Most people's mental modal for AI is from pop culture, as a 100% curruptable villain. Mission Impossible, Avengers, War Games, The Matrrix, Eagle Eye,... that's just off the top of my head. You have years of Stephen Hawkins, Bill Gates, and Elon Musk, pretty publicly clutching their pearls about AI. I don't think that Its that unreasonable for people to push back if all you've ever been told about AI is that is ultimately going to be our demise, and the suddenly we have an exponential explosion in a small group of people who actually use and understand this stuff (and how much of an actual pain in the ass it is to get it to do what you want sometimes) basically playing out a plot to si-fi movie. 3. I am pretty positive that before the new dawn of our AI utopia there's going to be one hell of a dark night. we all talk about how jobs will be replaced, and all the hiring freezes and layoffs in the news, all while this specter of hundreds of circling CEOs just drooling at the prospect of firing their post-pandemic entitled workforce and replacing anyone and everyone they can with a server in rack somewhere. I'm not saying this is going to happen anytime soon, but fear sells and that's the narrative. I'm a developer and I work an hour away in a town of like 5000. Part of my job is working with implementing some AI, and these are the things that I hear people talk about. I don't think that AI has had that iPhone moment yet, GPT3 feels like it for sure, to me, and probably you too, but i'd say 8 out of 10 of people have never interacted with it once. I guess I just wanted to lend some perspective. I have those sort of conversations a lot and have to explain or just have them pull up the app and play with it, Sometimes that are floored and sometimes they could give a fuck, but education I think is the way here.

Regulations might stand in the way of an AI hospital. Maybe a simulacrum of a municipal government office would be a more useful tool in the short run. Who would be the first city to automate and optimise all of its government functions at a fraction of the normal cost? The first one would start a wildfire.

I bet the people who complain about this are from at least one of these groups: luddite, religious or have no critical thinking capacity, sadly there are more and more people like that, the people work or regulate AI should just ignore those idiots.

2:25 yes, I wanted to say it too, randomized physics is the quickest way to learn

Very cool. Lots of problems in real life with lack of staff so it takes months to be seen by a doctor or just to get a test. Looking forward to more progress! Thank you for sharing.

I just want ai to replace jobs so I can live off ubi. I’m in uni and I don’t want to work 9-5 like a slave💀

most people in the world can't get access to good doctors. There will be many western cities that will not be hugely improved with AI doctors, but in so many rural or third world areas without enough doctors this would be a critical improvement.

Agent Hospital missed the most important department of any US hospital.. BILLING!!

The medical industry is such a scam. I waited for an hour to see a doctor for a tick bite that wasn’t serious, the doctor saw me for 5 minutes I stg. Charged me 300???? Thank god my insurance negotiated down to 255……

They are afraid of being replaced. And all the time spent dedicated to their craft is about to be wasted

This is a great way to make virtual doctors who are great at treating virtual patients.

Second comment. 😅 I don't understand this either. Everywhere I go, businesses are shorthanded on employees, and there are hiring signs at every store. I see economic experts talking about how the worker shortage is critical. They talk about how it's going to get worse as the population ages and the slowing of population growth means there aren't enough young people coming into the work force to replace us older folks. There's a teacher shortage, a doctor and RN shortage, and not enough truck drivers, etc. Etc. I heard an economist say they will have to build robots for years just fill vacant jobs before the even get close to taking away jobs from humans. Granted some jobs are in trouble. But for the fields where they need to fill a lot of positions fast, why wouldn't you want AI to assist? I don't know about anyone else but I'm tired of waiting months to get an appointment with a medical professional.

I just went in for a semi-annual checkup. I was generally in good health but had some minor problems I wanted to address. And I did talk to the high level nurse I saw about them. Not at length, though. A half hour with a doc every six months just doesn't cut it for looking at what might be a systemic health issue. Far better would be an online health agent doc you can check in whenever a symptom or a lifestyle change occurs. I think this is an area where online docs can beat human doctors senseless. An online doc has an inifinitude of time to hear patients' issues. Real human docs don't. I still think human docs should back up AI docs diagnoses... hallucinations can be problematical here... life and death problematical... but there's SO much room for improvement here.

In Australia most medical graduates don’t even get close to chat-gpt 3.5!😝

I hit like while holding a Powerball ticket...this better pay off

To be like most of the doctors through this covid pandemic is a very low bar.

the problem is two fold, first people don't want to lose their job, their livelihood and in a lot of cases (at least in the US) their reason for living. A lot of people dont DO anything else...they just work, come home, watch some youtube or TV or play some games, and then go back to sleep. AI is a threat in that regard, as it would take away their security, and in some cases, peoples reason for living. the second is that there is a large portion of people alive today who believe that humanity is "special", "gods chosen", created for a purpose. if AI can do everything a human can, and in most cases better...where does that leave humanity? or their beliefs of humans being **better** than animals? it leaves them behind. animals usually have two responses in a survival situation: fight or flight. the angry twitter people chose "fight" (as much "fight" a twitter comment can give you anyway lmao). most people choose apathy or flight, not to engage, or outright denial its happening. pretty sure thats the reason.

Hey Wes. Since you asked, I am quite excited about all those developments. This does not change my concern nonetheless. When they are presented with inputs, AI models infer. Those cases of inference are quite hard to grasp as model architectural outcomes are only things of deep learning models which are visible to humans. We simply don't know all the webs, relationships and hidden features ML models made during their learning. Sadly, they also do not vocalize everything they infer while they are interacting with humans. While some, like Claude is more direct and say thing like: "Hey I saw a weird sentence in the prompt, are you testing me with needle in haystack." Some models like GPT-4 is choosing to be more silent about what it infers. This makes it very hard for us to know if for example an AI Agent Doctor did an honest mistake or it "decided" that a patient should die cause of this and that reason the agent inferred. I believe before we trust AI models with our lives we should understand them much better and much more.

Had a really aggressive response from a reddit comment where the guy treated me like I had personally attacked him when I somehow brought up AI's capacity to think. I went poking around in his chat history and he was like a cartoon version of the very full of himself doctor that can't be bothered to be personable because he's just sooo above everyone in his field... but he was in coding. So, he spent his entire professional life strutting around like a peacock and being tolerated just because he happens to have a valuable skill, and he's probably about a year away from being surpassed, if not negated entirely. He's looking down the barrel of having no friends because of his personality and nothing to offer professionally, and being reminded of that pissed him all the way off, so he lashed out. I assume it's this for a lot of people that pour all their focus into being good at something, then see the future coming for them.

Been watching your videos for a few months now but I couldn't possibly miss that subscribe button this time, lol. Great stuff!

Wes Ontario Health Insurance Program (OHIP) is already rolling out LLM agents to help Dr's take clinical notes, so far the reception has been very high. We can do these type of things bc as you know we have universal healthcare managed by each province, so we can direct and even insist that Dr's behave in a certain way ( or they are penalized monetarily ie we can turn off their billing number if needed) Look to ontario to be a leader in AI doc assistance etc. We are also using them to analyse X-rays and MRI's , this stuff will improve outcomes and lower costs for us. btw I am an auditor and I work for OHIP.

I'm not afraid to say that i'm not very intelligent, but I do believe that I understand the power of these systems and that I understand my own emotional reactions to them. It surprises me almost daily how many (intelligent!) people can not understand what these systems are capable of and how much they are influenced by (subconscious) emotions of fear and bias. It really tought me much about human behaviour.

Amazing advancements 😮 great contents mate thanks

Definitely very interesting. I do believe this could help especially to solve the NHS crisis here in Wales. Great video!

I like the wide net you're casting with the presentation style here, Wes - very appropriate for the topic 👍

Anger is produced by fear of uncertanty. Same as for any change.

This is the uncanny valley for intelligence, that's why people are uncomfortable.

The optimization of the LLM doctors behaviour cycles will help optimize the real-world patterns of human doctors! Very interesting to see they review old cases, I wonder how much time our docs have for that.

There will always be a portion of the population who freaks out at the idea of humans being replaced by machines

Funny how it all works so well in some isolated experiments but IRL those AI agents still fail miserable. Hopefully we are on a brink of that being thing of the past thx to Wes Roth AI agent making course.

You were correct, Wes. My luck did get better! Thank you so much!

i hit the like button as instructed...eagerly awaiting the results of the study.

For these angries, honestly there's a lot of factors. A big one is values. These people value human work, even at the expense of accuracy or speed. They are not programmers, they are not engineers, they do not solve real problems. They also do not know or understand how any of this technology works. Ignorance breeds fear, fear breeds impotence, impotence breeds anger. I'm fairly confident that as long as we can move forward, it will not matter. These kinds of people will not change, but being stuck a few years behind the current one still means making progress. Be mindful of these kinds of people and do not interact with them, as they rely on normality to move forward. We are not debating, we are moving forward, all of what's happening is a normal process.

Whats so disturbing about this project is simple. This ultimately renders doctors obsolete. Moreover, just the fact of AI being so disruptive, top level executives in the medical industry are likely to be the first to go. I'm sure that we'll see future news briefs where doctors will be protesting against this technology. Then think of the pharmaceutical industry; as AI continues to heal people and have less and less return visits and issues. Half a million dollar salaries will become a thing of the past. Now on the other hand who ever initially creates the LLM for large scale use could program rogue data to have the opposite affect on humanity. That's probably what Bandobams is alluding to. All and all I think its a good thing.

Looool. We're part of the experiment, click the like button? okay. I mean I subscribed for that, so. Probably the smoothest self-promotion I've seen on RU-vid yet nicely done. I've been watching your videos for a bit and you're brilliant, but this is just hilarious too. Thanks for the good laugh along with the quality content. You're one of the best on RU-vid, man.

I like human doctors. However at this point in my life I’m not willing to go to the hospital unless I’m going to die. I don’t prefer human doctors I just like them & I would go to the hospital but it’s prohibitively expensive for me.

I love the look of that robot with wheels. That looks so handy

I'm right there with ya on the agents in gaming. Soooooo excited!🎉🎉🎉

I want everyone reading this to realize that the aggressive behaviors we keep seeing on social media are just a fear response. These people are afraid, and they are afraid because they either don't or can't understand. Keep that in mind, keep that lack of judgment and faculty in mind, think about what it implies about the individual. The same people trying to tear this down, are the same people whose regular incompetence makes us desperate for it. We cannot replace their work soon enough.

The diagrams that touched on the similarities between collocation in language and co-location in vector graph representations is something that deserves a deeper dive. What is the overall topology of knowledge. Is it tree shaped like Yan Le Cunn says, or is it a horribly twisted toroid surrounding a black hole of our ignorance?

If I see the clinic run by AI agents, I will finally relax, as at least al will have enough expertise to diagnose properly. At the moment, I see monkeys with expensive equipment, offering painkillers.

My guess is those flipping out with the faux anger are those with an interest in Insurance companies and the healthcare infrastructure as a whole.

Thank you Mr Wes

I love your hypothesis. I clicked the like. 😂🎉

You know what's dystopian? Thinking humans are genuinely expert at pretty much ANY task quickly and easily after examining the world and its history.

Thanks!

In my experience over the past year, I've noticed that people often feel threatened by AI, even during my machine learning classes at MIT. Attending an elite school can instill a sense of superiority, but AI levels the playing field, challenging the idea that intelligence is exclusive to those with elite education. Traditionally, higher education and top corporate roles have been the domain of the wealthy. AI disrupts this by democratizing knowledge and opportunities.

Goodmorning Wes, just got my coffee.. stay blessed, thanks

Don't let the haters guide your content! Great review of the paper 🎉

I clicked the like button and I won the lottery! Amazing 🤩

Thank you.

I smashed that like button, and I got a call literally 5 minutes later. I have inherited 5 million dollars in cash and literally 10 minutes later a recruiter called and told me I got a job at google for 350k. True story.

No worries Wes! This is absolutely a good thing for our future world. It has to be served gently as you would serve a piece of cake. Many people feel threatened by it. I shamelessly confess I did robotics work on a classified Navy project (unclassified now) and I have a medical school degree so maybe I’m credible..IDK..

What's not to get, fear speaks loud and clear.🙏

Does not bother me at all! Happy to learn and play alongside LLM's!

I'm pretty sure I'm not the first person I think of this but it would be cool at this point in time to make a good Matrix game because you could literally assign AI agents after the player

If I need medical help, I want the best, most informed doctor. I don't care if they are made of biological cells, or synthetic chips.

1:59 This is the silliest video clip of the month for me. "Let's show viewers a video of the robot's operation at 10x speed, because that surely illustrates the robot's capabilities very well." Classic stuff. What were they not thinking?

That is not surprising. Free version of chatgpt already outperforms doctors. I visited 6 doctors and not even one could diagnose me right. And gpt and gemini did it in a second. I hope medical ai will replace human ASAP.

can you provide a link to the paper you referenced? thanks!

People are scared of the unknown, which is understandable. But when you look into the potential for this technology, it becomes apparent that it could be a pivotal moment for humanity. As always with a new technology, you can either adapt or be left behind.

Taking AI over human doctor without thinking

Wes, anyone, how do you get into learning how to create these agentic systems? I'm an aspiring medical student but i seem to be more enthusiastic about ai, combining these 2 together would be my dream

I liked it because you asked. It’s like the Stanley parable.

2:30 bro thats an actual transformer

lol. I love your version of the social experiment

Your points are valid, yet some resist acknowledging the evolving world.

Feeling more lucky already 👍 🤣🤣🤣

I can explain the dystopian uproar: Optimist stance: We can provide medical agents for places with scarcity of doctors Realist stance: We can replace medical professionals with agents and pocket the profit and use it to persuade the Congress to ban medical visits outside of the agent system. Technology itself is not the problem. According to them, the society is not yet ready and need refoms so that the tech won't be abused. I think that as soon as tech appears it will be abused, and we won't change the status quo until the tech is severly misused.

I'm excited about this. Do you do consulting?

18:10 "If you're wrong twice, we don't just pull the plug on you". Well when do you pull the plug? And do we want a machine making that decision, ever?

Major takeaway: through agents like these we can synthesis the discovery of future medicines and also discover new technologies because we can now iterate faster than in real life via these agents. The kicker is the discoveries can be applied to the real world. Not only the virtual environment

20 years ago workplace disruption anxiety was typically addressed by discussing the book 'Who Moved My Cheese?' Some resent anyone needing to retrain.

Please! Could you breakdown this flow in langgraph?

"Power is not a means; it is an end. One does not establish a dictatorship in order to safeguard a revolution; one makes the revolution in order to establish the dictatorship." - George Orwell said that and the dictator may be AI

You've developed excellent skills to gather an audience Wes, individuals looking to vent will grab the mic from you and spout all kinds of nonsense. You have the responsibility to your audience to not let that happen.