Stephen had a chance meeting that led to an adventure in A.I. development. The recreation of the human brain?
Subscribe: Apple Podcasts | Google Podcasts | Amazon Music | Blubrry | RSS | More
Dave Young:
Welcome back to The Empire Builders Podcast. Dave Young here with Stephen Semple sharing stories about business building and empires, going from small to massive sizes. And Stephen, you said you’ve got a weird one for me today. I don’t know what to expect.
Stephen Semple:
I’ve got a weird story to share, and it’s a little bit outside of the norm of what we talk about because it’s not really about an empire, it’s about artificial intelligence, which is a topic that a lot of people are looking at and thinking about. And I was down in Austin, Texas a couple of weeks ago meeting with Roy Williams, our founding partner, writer of three New York Times bestselling books, and we got talking about artificial intelligence. And I shared with him this little artificial intelligence journey I went on, and he was like, “You have to tell that story.”
And I was like, really? He goes, “Yep, you got to tell it on your podcast and you got to tell it exactly the way you told me.” So I was like, okay, here we go. I got this opportunity. I was invited by Ellen K, who’s a big-time DJ in Los Angeles. To give you an idea of how big she is, Ryan Seacrest got his start with her. She has a star on the Hollywood Walk of Fame. So she gave me a call and invited me to this event that she’s hosting where she’s giving Marc Anthony his star on the Hollywood Walk of Fame.
So she goes, “Do you want to attend this event?” I’m like, hell yeah. So I hop on the plane. I fly out to Los Angeles. I’m at this event, and it was really wild because they closed down the street. I’m behind the ropes. There are all these screaming fans. Marc Anthony’s there. All sorts of other celebrities are there. And I’m like, you know what? I want to meet Marc Anthony. I figured out how they were leaving, so I positioned myself close to that door, so he was going to have to walk past me. And there’s no one at that spot yet.
No one’s figured this out yet. So it’s like, awesome. I’m in the prime location. And it’s working perfectly. He’s leaving. He’s walking past me. His back is to me. So how am I going to get this guy to turn around? Well, I’ve been working with a guy, a really amazing coach who’s been teaching me a lot about public speaking. But whenever you’re working with these coaches, they always throw these other little tidbits out. And one of the things he was saying is about status.
He says, if you ever want to establish yourself as being on the same status as somebody and get their attention, here’s what you do. You say the word hey, has to be hey, hey, your name, not theirs, your name, and one word. Now, if you can’t do one word, two, maybe three, but no more than three. So Marc Anthony’s back is to me, and I’m hearing the audience, “Hey, Marc Anthony, look this way. Hey, Marc Anthony, look this way.” And I said, “Hey, it’s Stephen Semple. Congratulations.”
He fricking stops, turns around, shakes my hand, thanks me for my support and how much it means to him. And we literally talked for 30 seconds. I was so unready for it, that I didn’t have my camera out to do a selfie. Because at that point, I could have done a selfie because we’re peers, right?
Dave Young:
Yeah, yeah. Okay.
Stephen Semple:
So I didn’t get the picture. That’s bad for me. But the cool thing is, if you’re ever in a situation where you want to do that, that’s how you do it. Now, why do I share this story? I’m sitting in the Hollywood W Hotel, which is right on the strip, and I’m sitting having a drink after this event. And this dude walks up to me and says, “How do you know Marc Anthony?” He was in the crowd’s front row. And so I had to confess, I don’t know Marc Anthony. I shared the story. I confessed my sins. And he was like, “Well, that’s pretty cool,” and we have a couple of drinks together, and we get talking.
He says, “Hey, what are you doing tonight?” I said, “I don’t have any plans.” He goes, “Well, so the W, there’s one tower that’s the hotel and there’s one tower that’s residence.” He goes, “Well, I’m having a bunch of guys over to watch the football game. You want to join us?” And I was like, sure. So I arrive at this penthouse suite and it’s all these technology guys hanging out, and it turns out the guy who I was talking to does a lot of work with Tesla and their AI. And here’s what I started learning about AI.
This is where the conversation gets weird because here I’m the marketing guy amongst all of these tech billionaires and angel fund investors and guys working so deeply on the bleeding edge. It’s incredible. I was listening to them talk, and they kept referring to AI as being a blob. I go…
Dave Young:
A blob?
Stephen Semple:
A blob. I go, “What do you mean blob?” Well, it’s a blob. I was like, blob. Right, yeah. It’s just a blob. I said, “Is it code?” No, it’s just this blob and you start to train it. I said, “You don’t program it?” No, no, no. It’s just a blob that you train because it’s already been created. It’s your job now to train it. The closest analogy I was able to get is to think about it as being an infant. Knows nothing, but has the capacity to learn. And that’s basically AI. We get talking about the training. So I was like, well, what do you mean to train and talk about the training and all this other stuff? So here’s where it gets even crazier.
Dave Young:
Okay.
Stephen Semple:
They said, “Yeah. One of the challenges with training AI is that you got to really watch for it, it makes stuff up, makes shit up. AI makes stuff up.” I was like, “It makes stuff up?” It’s like, “Yeah, it makes stuff up.” I said, “So it lies?” He goes, “Yeah, absolutely.”
Dave Young:
It’s got no conscience, no moral compass.
Stephen Semple:
We think because it’s computer code that looks for patterns that it will always be truthful in those patterns, but sometimes it makes shit up. The other thing is they deeply believe in daydreams. They deeply, deeply believe that AI daydreams, and they were going through why they have that belief system. So here we got this thing that’s a blob that is our job to train, that daydreams and makes stuff up and works on pattern recognition, does not have a moral compass, and cannot explain how it comes to its conclusion. Now, what side of the brain does that sound like? It has no language.
Dave Young:
It’s on the right side. It’s the right hemisphere.
Stephen Semple:
Right. So now here’s what they’re working on, because what they realize is there’s a bit of a problem on that when you’re developing these tools for self-driving cars and self-flying planes and all this other stuff where you need to be able to query it and get an answer as to why do you believe that. So now they’re creating a second series of AI. That sole job is going to be to query the other side of the AI, figure out why it’s come up with that conclusion, put it into words, and explain it to us. What does that sound like?
Dave Young:
That’s the left hemisphere.
Stephen Semple:
Right. Okay, are you ready for it to get even weirder?
Dave Young:
Sure.
Stephen Semple:
So here’s the other problem. They go, okay, so they’re working on that right now. This is not bullshit. This is being created as we speak. Here’s where it gets crazier. They also suddenly realize there’s a problem with that approach because the problem with that approach is if you’re querying a car in terms of why it is making this decision, there’s going to be times when the car goes, “Shut the hell up. I can’t talk to you right now. There’s an accident happening in front of me.”
There needs to be a gatekeeper, a police officer between the two sides of the brain to, A, reduce the number of queries and have these two sides get along. Well, the more we studied the human brain, guess what that is? That’s the hemispherical well and it’s the hemispherical lateralization. It’s that place between the two hemispheres.
Dave Young:
The corpus callosum. Sure.
Stephen Semple:
Yes.
Dave Young:
That’s the pathway.
Stephen Semple:
Right. What we’re basically creating is a god-damn human brain.
Dave Young:
Isn’t it a weird time to be alive, Stephen?
Stephen Semple:
The funny thing is one of the guys who I pointed that out to didn’t even realize that. It’s not happening because they’re mimicking the human brain. It’s happening because we have this problem, and this is the way you solve the problem, which then creates this other issue when you layer together this is what works. I’m like, dudes, dudes, that’s the brain. Millennia after millennia of evolution has made it that, wow, this functions pretty darn well.
Dave Young:
Best we’ve got.
Stephen Semple:
Best we have. But here’s the interesting thing I’ve learned from all of this that we need to understand, I’ve only just started on this path myself. AI is here to stay. It’s going to become more powerful, more prevalent, more all those things. So we need to figure out how to use it. And this is what I got talking to them about. I said, “Look, for dumdums like me, what should I be doing?” And they said, “Start becoming good as a trainer. Your job is to become an amazing coach.”
Dave Young:
An AI coach?
Stephen Semple:
An AI coach. Learn how to coach and teach AI. And if you really think about it from our perspective, Dave, if you’re going to coach it on creative, it’s about creating amazing creative briefs. It’s a creative brief. And here’s the reason why I don’t fear AI in terms of taking our jobs, is that what I know writing good creative briefs is really, really hard. But the better you get at writing a creative brief, the better the creative will be. The better you get writing an operator’s manual, the better the operation will be.
But then the other part is you have to recognize that the AI will periodically be wrong, so just like an employee. So in other words, you can’t just sit there and give it to the AI and never monitor it. Just like an employee who has bad days, the AI will have bad moments. And you need to go back, recognize that retrain it, and say, “That was a mistake. Don’t do that again,” because it’ll make things up and it will lie. And so the mistake businesses could make is oh, I got the AI in place.
I never have to look at it. Oh no. Oh no. Because the other thing is when it does something wrong if there’s no correction on that, it goes into its programming as correct. I never got any feedback, so therefore it’s fine. I can do that again in the future because there are no consequences to it.
Dave Young:
It’s frightening.
Stephen Semple:
Crazy. In one way it’s frightening, but on the other, like any technological change, whether we’re talking about the beginning of the Industrial Revolution or the beginning of computers, we have always had this same fear, that it’s going to replace us, going to replace us, going to replace us. We’ve always had that core fear, and those who have succeeded are the ones who’ve said, “How do I use this as a tool? And what is my role with this as a tool?” And our role is going to be being coached.
Dave Young:
I like that viewpoint of it. When I say it’s scary, it’s because the change is happening so quickly.
Stephen Semple:
It is.
Dave Young:
Right? It just wasn’t that long ago that we weren’t thinking about AI. I mean, other than just some nebulous concept that was out there that we knew was coming.
Stephen Semple:
Correct. Yes.
Dave Young:
And now that it’s arrived, it’s arrived quickly with a storm.
Stephen Semple:
And its ability to learn is unbelievable, and its ability to use huge amounts of data is incredible. But again, its weakness is it’s all pattern recognition, pattern recognition, pattern recognition. And I’ve been playing around with it on the creative side, and it’s throwing in the Crazy Ivans and the weird character diamonds and the odd constraint that actually all of a sudden creates some pretty incredible outputs. And so that’s what I’ve been sort of experimenting with.
This is really super basic, but here’s one of the first basic principles that makes it different. When you do a Google search, you do a search. And then if you want to refine that search, you start over again. You may add a longer tail, but it’s sort of a startover. When you’re working with an AI and you ask it to do something, and then you ask it to refine, it’s not forgetting the first set of instructions or what it did in the first place. That’s retained. It’s then building on that.
Dave Young:
All right.
Stephen Semple:
So that’s the reason why it’s much more of a coaching process. Because it’s like, okay, I want you to do this. All right, now add this voice to it. Okay, now I want you to build this constraint. Oh, and I want you to research this thing. Okay, add that to the original thing. Okay, cool. That’s a mistake. Remove that. Now make it longer now. But each time it’s building upon the instruction set from the previous time.
Dave Young:
Fascinating.
Stephen Semple:
Yeah. What was interesting was being in this environment where you’re just hanging out, watching a football game, and drinking drinks with these guys, led it down a very different conversation because it wasn’t a lecture hall. It was like, well, what do you mean blob? Because I had never heard that term before. That was an insider, but it explained a bunch of things. Because when they were trying to explain that to me, I was like, “Oh. Oh. Okay, so you don’t program it?” And they’re like, “No, no, no, no. You teach it. You coach it.”
That’s what you doing. You’re coaching it down a path that you want it to do. And then like any great coach, you still got to monitor and you still got a course correct. It’s not one-and-done. It’s not, you’re done with the AI. So what will end up actually, I believe, will evolve as a position in business will be the customer service AI coach, where your job is coaching the AI on how to do customer service correctly.
Dave Young:
You live an interesting life, Stephen. Just thought I’d throw that out there.
Stephen Semple:
And it was just this weird thing where this guy comes over to me and goes, “How do you know Marc Anthony?” It’s like, “Well, I don’t actually.”
Dave Young:
So you’re teaching that AI?
Stephen Semple:
Well, it was sort of like, no, I just did this little giggle, and here’s how it turned out, and here’s how I ended up here. And he’s like, hey, do you want to join us for drinks? We’re watching a football game. It’s like, sure, what the hell? I have nothing else going on. When I shared that with Roy, Roy was like, that’s such an interesting perspective. People need to hear it because it puts us mentally and even emotionally on a different path in terms of what can AI do for us.
I’m still not sure how I want to apply it, but it has changed how I’m playing around with it. I’m still very much on the early experimentation basis with it, but it has changed my thought process in terms of how I approach my relationship. And I think that’s almost how we have to think about it. What is my relationship with AI?
Dave Young:
Yeah, fascinating stuff.
Stephen Semple:
It is an interesting times that we live in.
Dave Young:
It certainly is. Thank you for sharing that, Stephen.
Stephen Semple:
All right. Thanks, David.
Dave Young:
Thanks for listening to the podcast. Please share us. Subscribe on your favorite favorite podcast app and leave us a big fat juicy five star rating and review. And if you have any questions about this or any other podcast episode, email to questions at theempirebuilderspodcast.com.
- The Secret Weapon of Trivial Pursuit: Celebrities - September 4, 2024
- What Made Tripadvisor a Leader? Customer Centricity. - August 27, 2024
- How To Recruit Good People - August 20, 2024