Legal Limbo: ChatGPT's Legal Entanglement and the Ethical Dilemma of AI Robots

0:00Today on the podcast we are going to be talking about robots, we're going to be talking about revolutionizing healthcare, we're going to be talking about a professor that failed all of their students for using chat chat chat chat. and a couple other interesting stories that are happening around right now. The first one I wanted to talk about as far as AI robots arriving. This is kind of interesting because in the past we've seen that Tesla's creating a humanoid robot and when they first debuted it, I'll be honest, I was rather unimpressed. The thing could hardly move. It didn't look super realistic.

0:30Not like some of the other images they kind of teased of this thing. And so I wasn't super impressed. Now, that being said, of course, any humanoid robot, it's impressive if it can move. But when you've seen things like the Boston Dynamics humanoid robot, which if you've seen their demo video recently that went viral, essentially the thing is on a construction site. Someone's like, hey, can you bring me a hammer? This thing looks like a person walking and it goes, picks up like a hammer bag, does a whole bunch of like flips, climbs up some scaffolding, gives the dude the hammer and then does a backflip off.

1:01It's pretty terrifying how powerful it is. In any case, Boston Dynamics has been like viewed as the most advanced like robot. Essentially, you've seen the robot dogs that are really hard to push over, whatever. You've probably seen the dystopian China mounting like speakers and other people mounting automatic rifles to those. So, you know, that's kind of sketchy. But Boston Dynamics is kind of seen as like the leader here. And what I do think is interesting is that there are other companies coming out. of course, Tesla had theirs. Tesla actually today released a video of their Optimus robot, and they have like five of them walking around.

1:41They're looking a lot more nimble. And it just reminds me of the advancements we saw with Boston Dynamics. Now, why do I bring up humanoid robots when talking on an AI show? Because like number one, inevitably, these things are going to have AI built into them. In fact, with Tesla's demo at the end of it, it literally said, come and work at Tesla AI to help us build things like this. So like they're the robotics and AI is going to be a combination. And I think that robotics is going to play a very massive phase in a really broad role of AI right now. The sentiment I've like heard from a lot of other founders that are working in AI right now, which is like how I've literally thought is like, they're like, Oh, man, things are evolving so rapidly, people are coming out with technology so rapidly, is this even worth developing?

2:26And is someone else just going to build something that's going to knock me out? Should I just go become a plumber or an electrician or do something like that? So that's a sentiment I've had. But like the thing is, you can do that. And I believe you have like three or four years on you if you want to go do that. But these humanoid robots, I believe mixed with AI are going to take, take those jobs as well, where essentially you'll have robot electricians and plumbers and everything else. So that's why I bring it up. And it's very evident in the way that even Tesla is pairing their robots with their AI arm.

2:59So recently, a new AI robot came was debuted by a by Sanctuary AI, right? See another AI company. And it's a Vancouver based robotics firm and they released something called Phoenix. So essentially, this is a humanoid robot. It's designed for the workforce is what they say. So literally, this thing is, you know, targeting electricians, plumbers and that area. It's about five, seven, and it weighs 155 pounds in their demo video it showed off some really complex hand movements it has a lot of different uh like its hands are very versatile they can do a lot but i will say you know i'm trying to be like very honest um in my skepticism with these things in their in their like little demo launch sizzle reel or whatever they didn't show this thing actually walking around or doing any work which is kind of what it's like touted as being able to do so i will put some healthy the skepticism there.

3:51And I said the exact same thing with Tesla when it came out. In Tesla's video, they actually had a whole bunch of them walking around and showed it doing a bunch of interesting things. So I would say Tesla's further ahead than them. And Tesla also has a really solid team just because of Tesla self driving AI and everything they're building there. And a lot of that team, I think overlaps and helps them. So I think Tesla has a lot a much stronger position. That being said, Boston Dynamics, as far as just a robot that can move and walk and do stuff, I think is even miles ahead of Tesla. But Boston Dynamics may not have the AI component that can really take those to the next level.

4:28So I'll be curious to see if they try to develop their own or if they end up partnering with someone as that inevitably is the next step in their company. So really interesting thing. We're going to stay up to date on what is happening in robotics. The other thing I wanted to talk about today is a company called Hippocrat AI. And essentially, they are building an AI powered tool that they say is focused on safety first. It's a large language model to replace WebMD. So if anyone knows WebMD, I definitely have some friends and family that are on there too much. You know, they're out searching all of the random symptoms they might have and are self-diagnosing themselves with all sorts of things.

5:05It's definitely a hypochondriac's worst nightmare because if you go on there, it will tell you you have any disease essentially okay not really but it actually is quite a useful tool because it does tell you a lot of different symptoms of things and while it is easy perhaps for some to over diagnose themselves it is a really good start to figure out what you have etc now this hippocrat ai is planning on replacing that with an ai medical model that essentially you can use now the reason i bring this up is because i do believe that uh we are going to start seeing a lot of these really specific AI models.

5:42And Sam Altman also said that he's he specifically said the transformer model is kind of tapped out as far as what we can get it to do. And he believes that the future is just going to be a lot more specific, a lot more hyper, kind of like hype, like niching down into very specific data sets. And so this would be the next step. That's where I believe we're going to start seeing this. We've already seen this happen with law. There's a very big law AI that came out. really received a lot of funding. And now this one, Hippocrat AI has received$50 million in seed funding from General Catalyst and Andresen Horwitz, which are some massive venture capital funds.

6:21So they have some really solid backers, they have$50 million. And I believe that they're going to be able to pull out some really impressive, some really impressive, impressive performances. So the CEO of the company, Moonjah Shaz, he specifically was talking about the fact that Hippocrat's AI outperforms most leading models, including GPT-4 and Claude, right? So Anthropics, Claude, and OpenAI's GPT-4. Those are kind of the big guys on the block right now. And he said that in over 100 healthcare certifications, Hippocrat AI beat them. So I do think this is important because this is a specific strategy that even Sam Altman, CEO of OpenAI, has touted as being very viable.

7:06And when you're getting$50 million and you have these really solid people backing you, I do believe this is going to be a company that makes some big strides. So I definitely keep an eye on them. They have the money and the manpower. And now I think they're just going to execute on a very revolutionary product. I think it's going to be extremely fun to watch this project as it continues to unfold. The third story I want to talk about today is kind of, this one's really trending right now in the news. essentially a Texas A &M professor failed every single student in his class for using chat GPT, but it's not what you think.

7:45So this story initially, I believe, kind of broke on Reddit, on our chat GPT subreddit, and someone said essentially that this, I'll read you the message that the professor sent out to the students, and then I'll tell you what the problems are with this. So he said, good afternoon, you shall be able to see your final grade in the course for this. At this point on D2L, before everyone re-emails me individually, read this carefully, in grading your last three assignments, I have opened my own account for ChatGPT. I copy and pasted your responses into this account, and ChatGPT will tell me if the program generated the content.

8:26will tell me if the program generated the content. I put everyone's last three assignments through two separate times, and if they were claimed by ChatGPT, you received a zero. The final grades for your course will be due today, blah, blah, blah, blah. He goes on to just tell him the rest of the thing for the class. Essentially, though, the student said every single person in the entire senior class, which you needed to pass this class in order to graduate, so everyone in his entire class is blocked from graduating, because he claims he put the last three assignments in ChatGPT and said, did you write this?

9:00And ChatGPT said it wrote every single one of them. Now, what's the problem with this? ChatGPT is notorious for, you know, sometimes hallucinating. So that's one thing. But like, definitely it would just claim the content. Like ChatGPT does not know if it wrote something or not. This is not how you, this is not an AI detection tool. Like this, this poor professor probably thinks he's super clever and, or he thinks ChatGPT is super clever and it's able to, you know, knows everything that's ever been written, or it somehow remembers everything it's ever written on any account. Anyways, this is not how you check.

9:34There's definitely tools for checking if content is AI generated, but pasting it into ChatGPT will inevitably give you a false positive. So, I feel bad. You know, a lot of people are saying they're hoping that the teacher, someone tells the teacher that that's not how ChatGPT works and allows the class to pass. I'm sure this thing's got a lot of news coverage. It was in Rolling Stones magazine, I think. They just wrote a story on it as well. So at this point, hopefully it's just an embarrassing experience for the teacher. He moves on and doesn't have, I don't know, any sort of pride or vendetta to try to actually sink the class.

10:08But yeah, anyways, just a note to the world, you need to use an actual AI detector. ChatGPT doesn't do that. Okay, one other very interesting story that came out recently. There was someone also on Reddit recently that was talking, essentially, they were going back and forth with a lawyer for someone that they were suing. And they said that they had to just recently tell the lawyer they'd been using ChatGPT the entire time because they assumed that all of his legal strategy and all of his petitions and documents and affidavits and everything else that he'd been sending them had been coming from an actual other lawyer.

10:48And so it's actually kind of funny the way that he had to tell them this. So essentially, the lawyer messaged him and said, I have a new counter offer that might help resolve your matter. Please let me know when you are available to speak. I am advised that you might have counsel. If so, I cannot speak with you without his or her consent. However, I would be happy to speak with your attorney. and uh you know he replies back and he says my counsel consents to us speaking uh can i call you now and they said i need to hear from him or her this is not my policy it's the new york rules of professional conduct an email would suffice and then they said it's chat gpt my counsel is literally chat gpt and he has a uh screenshot of like literally what he's been asking chat gpt which recently he said i need you to generate an agreement deal between two parties for the surrender of an apartment guarantee that the party formerly known as and then he blanked it out will cease all present and future legal proceedings give the party known as two thousand dollars on may 31st upon surrender of the apartment on that date then chad gpt goes and you know writes the entire agreement for surrender of apartment and release of claims and uh that these are the responses that he's been giving anyways it's kind of funny um the the lawyer then responded to him and said, then please call me.

12:07It's not considered a lawyer dot dot dot yet. Anyways, very interesting. You can use this as a lawyer. I was recently listening to a podcast today that was talking about using ChatGPT as as like essentially an advisor. They gave an example. It was on the it was on a podcast with Jason Calacanis, This Week in Startups, and he was saying, wouldn't it be interesting if you're sitting around a room and you said, hey, based off of you know, based off of the situation we're seeing geopolitically, let's ask the last six presidents or let's ask Thomas Jefferson or George Washington, what they would do about the situation in Taiwan, right?

12:47Like you can literally use this and based off its entire corpus of data and everything it knows about that person and responses and decisions that person has made, it'll be able to, you know, act as if it was that person. So obviously it can do it with lawyers. And I think that this is just scratching the surface for where we'll start to see this in the future so this is definitely an area we're going to want to keep an eye on and follow up with