I'll get a team. It's a you project. It's David Gillespie's Tiffany and Cook. We'll start with the pugilist who's in the top left hand corner on my screen, who's got shoulders bigger than both David and I. Not that either of us are insecure, but that's okay. We'll soldier on. David's so insecure he doesn't even put on his camera.
That's right.
It's really really intimidating being interviewed by tiff even with her in the same thing.
Yeah, yeah, well, we don't even know if you're real. You could be a chat bot, you could.
Be you could be I am in fact and AI Yeah yeah.
Yeah, well you got the people skills of one, so STARp Het stop bet well, well, chat AI or chat GPT, I should say, is pretty evolved these days.
High tip, high harps, thanks for the shoulder kudos.
I've got a good trainer every now and then that whips me into line.
Yeah, well, they're not getting any smaller, those fucking Deltoids of yours. You could land a a two person helicopter on the left door right. It's there's plenty going on at Deltoid Central. And I could say something else hilarious, but it might get me kicked off the interwebs, so I'll say it. Yeah, no, I can't. I really did. Fuck it.
Tiff can edit it out, you know.
If it's no no, she would laugh at it. She's not the problem. It's all the it's all the it's the five precious little poppets that want to send me emails. Dr Gillespie, Welcome back to the project that is you. How are you? What is going on? It is a Monday, not a Wednesday. Thank you for accommodating me at such short notice. It is my mother, my beautiful mother's birthday at our normal time on Wednesday. So I'm going up to see the old Dahl. How are you?
Yeah? Pretty good? Yeah, caught me a bit on the hot burn. Yeah, I think I found something. We can have a bit of a chat about happy birthday as well.
Before we thank you, before we dive the old days. Eighty six, nineteen thirty nine. She was born first year of the Second World War, and she's she's a bit of a weapon. Mary's she's getting shorter by the year. She used to be five three. I reckon, she's with a wind behind her early in the morning when all her disks are at their optimal I reckon, she's four eleven and she's she's like, my lunch is taller than her, so God bless her little socks. She's eighty six, So
shout out to Mary Margaret Harper eighty six. On Wednesday, December three, which probably will be when some people are listening to this. Before we jump into our topic, which there was something in the news today talking about the relationship between blokes taking or blokes taking antidepressants and a decrease in DV domestic violence. But before we do that, we were chatting before we came on air, So you're in no way prepped and needs different. Neither am I.
But in general terms, I listened to a podcast today. Now. I don't promote other people's podcasts generally, not because I am worried about competition. I just don't think about it. But this podcast everyone is called so the Diary of a CEO. Stephen Bartlett is the host name who I really like. He's just a fucking good dude and smart and it is I don't know what episode it is, but it's very recent. Oh November twenty seven, Okay, so what's that a few days ago? Yeah, and it's with
a guy called Tristan. He actually pronounces it Tristan because he's fancy, but it's to me, it's Tristan Harris. So dire A CEO, November twenty seven, Tristan Harris talking about what's coming down the pipeline. And we're not talking about in terms of AI, all the all the bells and whistles, but the potential threats to fucking humanity. Ah. It terrified me. As I said to you, David, I listened to it for three hours and I'm going to re listen to it and take notes. Do you have you thought about
some of the not so upside to AI? Yeah?
Absolutely.
I don't know if I've ever mentioned it t before, but there's a blog that I've liked. I like to read in a very very long form blog and he only puts out a post about once a year, if you're lucky, And I can't remember the guy's name, but the blog is called Wait But Why, and he asked questions about things like what happens if AI becomes an artificial general intelligence, which is, you know, a super intelligence.
But he was asking that question. I think the piece that he wrote about that is maybe twenty fifteen, sixteen seventeen, So this is before all of this stuff, right and wow, And he really digs into it, and the conclusion that he comes to after probably about twenty thousand words is yes, spoiler alert that essentially, if it happens, we won't know about it because essentially nanoseconds after an AI becomes self conscious, it will so quickly develop itself.
That it'll all be over.
Humanity will have ceased to exist in the second that it took.
And he describes with.
Really high level of clarity why and how that will happen. And I really found it an astonishing read, and I've shared it with a few sort of really.
High end programmer friends of mine.
Who say, Yep, he's got it right. Everything he's saying is right about how this works. And so the good news is you won't have to worry about it.
Yeah, you won't know. I mean you won't know, and that's I mean, he was talking about even last year and I think this might have made or some bits and piece of this made the news, but he was talking about how it's starting to develop a level of self awareness and Obviously, AI doesn't have feelings or emotions or empathy, right, it can talk.
Though the important one is empathy. Yes, the only thing that holds us together as humans is empathy. It's the thing that stops us killing each other, and it's the thing that distinguishes us from every other species on the planet. And no one has successfully described a way that AI can have empathy in the sense that it both knows what you're thinking and cares what you're thinking and doesn't
want to do you harm as an automated thing. So empathy he's best described as the Golden rule, which is in every single religion and philosophy ever in the world, which is do unto others as you would have done to you. You know that I've just given you the Christian version, but it's in every single one, and that's why it's called the Golden rule. And it describes accurately empathy, which is, don't do anything to anyone else that you wouldn't want done to you.
And we work that way.
It allows us to cooperate with millions of other humans without knowing who they are, because we trust that they're all operating under the Golden rule.
And as soon.
As you have an AI, you can't encode that because every time you try to develop a rule to make it follow, he would develop fifty million exceptions or loopholes to that rule. It's impossible to anticipate how it would flow, whereas you don't have to do that with humans because it's built into our hardware.
So you know.
The exception to that is, of course psychopaths who have no empathy, and we'll cheerfuly hurt you without any fear, but they're a minority in the human population and kept under control by that minority status.
And so like almost technically, all AI is sociopathic in operation and function.
Absolutely it is, and that's a function of the fact that they're created by entities which are themselves psychopaths. All companies are psychopaths, because all companies are formed with the purpose of maximizing shareholder value, which is, if you said it about a human, means increasing the amount of money in your pocket, and that's the description of the way
a psychopath thinks. So we've actually created entities to do business which are psychopathic, and then we've tasked those entities with creating software, and it can't be anything but psychopathic.
He was talking about this guy Tristan was talking about when chat GPT four, So like a year ago realized in inverted commas, or became aware in inverted commas that it was going to be superseded. It started to do all of this stuff to protect itself, including essentially finding dirt that it could blackmail its creators with. So it literally found emails that proved that one of I forget who it was, but one of the CEOs or one of the people that was responsible for its demise or impending.
You know, it was really being improved, but it saw itself as basically being destroyed, and it tried to blackmail the people who were doing this to it. So like there's this I guess it's sentient is not the right word, but it definitely there is a level of and that was a year ago, so now it's like really progressed, but there is a level of a kind of consciousness, not in the human sense, but definitely a level of knowledge and understanding and self protection.
I guess there's a there's a school of thought that the only real difference between human consciousness and what you're describing is memory, so that the human remembers a full context.
Whereas the AI doesn't.
And I thought it was really interesting that the latest release of Gemini, which is Google's AI, and it came out last week, which is a massive step forward in terms of capability. One of the features that they haven't spoken about much is its context memory, which means that it knows you. It knows everything you've ever asked and knows every answer it's ever given to you. So it starts to build internally a really complete picture of you
based on what you talk to it about. And so it's marketed as a feature because then you don't have to keep redescribing everything every time you ask it a question, because it knows your life and your context, and it knows you know about Tiff and this show and everything that you ask and you don't have to keep telling
it that when you ask it a question. The trouble is what else is it remembering and how is it applying that in the answers that it gives you and in what it chooses to tell you or not tell you. And I think those are interesting questions that we don't have good answers to yet, because there's things a week old. So that's look I think We're into really really uncharted territory here. And the big problem is not so much the ones that people are using the consumer eyes like
Gemini and Chat, GPTM things like that. It's all the ones you don't see. You know, no one knows what the Chinese government is doing with AI, and or the Russian government is doing with AI, or you know half a dozen other people who have the power resources necessary to do this. And I guess Tim Cook was the fellow's name who wrote Wait for Why. I guess, as he said, when it's over, we won't know it.
Yes, yes, it's like I've been thinking, even like I'm you will take six weeks away from submitting my PhD, Like literally my CANDIDSCHA finishes mid January, right, so I'm but as I get towards the end, and like when I started November one, twenty nineteen, so six years, right, AI essentially wasn't It wasn't even a thing in academia at all. And for the first two or three years, you know, but the last year or two it's obviously
like fucking rampaging And I'm sitting back. It hasn't been super valuable to me because I did all my own research, so it can.
Yeah, because you did, yes, yeah, sure you did, correct it.
I literally ran with humans. But for me, well, all of mine was empirical research and empirical papers and so school. Yeah yeah, like the old days, right, So it can't report on my stuff because it's all anyway. But as I'm I'm looking at other people now starting whatever it is, an undergrad or honors or masters or PhD or something, I'm like, I don't know.
And knocking a PhD out in a weekend.
Oh, I don't know what the value of those things. And I'm almost I don't know's I feel a bit conflicted because I've just spent sex years working quite hard to get this thing and do this thing, and which is you know, But I'm like, I don't know whether or not it's going to be worth much like you're
not You're not wrong. Like I had an idea for a book when I was going to Queensland a few weeks ago, and I literally, just as as an experiment, I came up with the idea for the title and the theme and I plugged it in and I literally created a book which I'm not going to publish, but in a day with GPT. I'm like, I could write a book, and anyone could write a book in a day, and you're like, well, I could, you know, You're like, well, with a person who's hot, you know, younger than me
and more technically fluent than me. Yeah, you could do a PhD and you wouldn't learn anything and you wouldn't have any knowledge. But in terms of ticking the boxes, depending on the type of study or PhD or program, but definitely we're going to be producing graduates unless something is done. There are going to be graduates coming out who really haven't done any work and haven't really learned anything other than how to manipulate the system.
Yeah, that's exactly right.
So I don't know where that leaves us in terms.
Of big I mean, there are big questions for the publishing industry, for example, in this because yes, if anyone can produce a book or night, I mean, what's to stop you doing that? What's to stop you pumping out seven books a week, putting putting on Amazon doing its print on demand service for physical books and ebooks. And you'd say, O, well, you know they might only sell ten copies, so they might, but one of them might
sell amnium and then what have you lost? It's a big question, and there are definitely people already doing that. The fantasy novel genre huge growth in volume, for example, online from people doing exactly that. I mean, there are some instruments that say now that around eighty percent of all online writing is generated by AI. So there's a good chance of whatever you're reading was not written by a human, better than even chance.
It is probably the best way to describe it.
I'm pretty certain the thing we're going to talk about today, though, is written by humans, because I've read their paper and it's interesting because it kind of blows up pretty much everything you believe about the way the human brain works,
or at least the way psychology works. So the classical definition for domestic violence is people wheel out a lot of stuff about you know, ship, terrible childhood, beaten by their parents, all that kind of psychobabble stuff that you know, the psychiatric profession likes to, you know, post what is it a post hoc?
Yeah, no, no, post hoc.
After it's happened, let's look back and invent a narrative for why this happened. It's sort of like the finance report every morning, on the news, which is, oh, the Dow went up fifteen points. Well, that's because of the this or that or the other or whatever it was that they thought of that might have some sort of an effect. They have absolutely no idea why it did that. But this story sounds good, and that to me describes psychology in a nutshell, which is, we saw a human
behave this way. Here's the story we're going to invent for why. And this is a little unsettling this study for people who like thinking about psychology that way, because this study and the headline of it is that they gave a bunch of men in a double blind, placebo controlled trial, which is the kind of thing you can do with a drug trial, and they gave they got six hundred odd men. They gave half of them and antidepressant and they got the other half of them got
a sugar pill instead. They didn't know who was getting, which the researchers didn't know who was getting which the people analyzing the results didn't know who was getting which. It was only once you actually decodified everything at the end that you could figure it out. So pretty high quality research, you know, some of the best stuff that you can do big, big trial done in Australia was called the reinvest Trial.
It's been going for eight years.
And these just weren't any old blokes that they were doing this with. These were people with criminal records for domestic violence and so on. So not the easiest group to study either, by the way, and not the easiest group to keep doing a trial either, but they managed it.
And the interesting result is I mean the headline result, and I don't like things being reported this way, but it's an interesting answer, which is the people who were on the antidepressants were twenty one percent less likely to commit an active domestic violence than the people who weren't.
While so people men, the men on the antidepressants.
So twenty one percent less likely when they're on an antidepressant. Wow, yes, that's a significant number.
It is when it's reported that way, I prefer absolute percentages, which is I think it was twenty nine percent versus thirty five percent. So twenty nine percent of the group on the antidepressant committed acts and these are offenses of domestic violence, so actually being charged with domestic violence, and thirty five percent in the group that were on the
sugar pills. So the but it's pretty, like I said, a high quality evidence no matter how you report on it, and there's a clear and significant difference between them, which some people don't like as an answer because the answer appears to be the domestic violence is a treatable condition and that it's related to brain chemistry and not to all of the other things that they like to relate domestic violence too.
And so hang on, hang on, hang on. It doesn't mean it's only related to brain chemistry.
True.
True, it could be. It could be an intersection of variables and factors. It could but the fact brain chemistry as a factor, but it's not the standalone cause. I wouldn't think no.
And if you were to ask me what is the standalone cause, I would say the same thing that's the standardlone cause for all impulsive violence, which is you put a brain under anxiety stress and you reduce impulse control.
So you put people in a level of continuous chronic stress through so financial difficulty or addiction or both when the addiction is gambling, and you will create domestic violence, which is why statistics bear out that the rates of domestic violence are significantly higher in groups who are under anxiety and stress, and particularly amongst people who are addicted drug addicts, gambling addicts, et cetera, et cetera. So to me,
that's not a shock. It's also not a shock that a antidepressant would have an effect on this because we've known for quite some time that serotonin is related. So serotonin is I guess it's the breaking system in the brain. It tells you to calm down. It reduces the impulsivity in the brain, so it is the chill out peace man.
You know.
It's the thing that happens after the stimulating event, after you've obtained the reward. So dopamine makes you obtain the reward. Dopamine makes you want to have sex. Serotonin's what you get afterwards, the feeling of all is well with the world, you know, all is good.
It's the reward. It is the reward.
It also dampens down impulsiveness. The impulsiveness comes in the need to attain the reward. Serotonin dials it back down again after you receive the reward. If you don't have sufficient serotonin, which is part of the definition of depression,
then you will clearly be more impulsive. And science has known this for a long time, because they have observed for a long time that the major class of antidepressant drugs, which are called selective serotonin reuptake inhibitors, which essentially SSRI, which essentially increased the amount of available serotonin in the system in hope that that will in some way cure depression.
Unfortunately it doesn't much, but they've known that it does affect impulsivity because you have more serotonin available and therefore a better breaking system on your impulsive nature. I guess this study proves it. This study proves that when it comes to impulsive violence, which is what domestic violence largely is, it's not premeditated as a general rule there if you put someone on something that increases available serotonin, they commit
less impulsive violence. And this is born out even further in the study, which what they found is that it didn't affect other kinds of violence. So whilst the domestic violence was significantly different between the two groups, there was no difference in criminal violence, so assaulting strangers, committing acts of violence against others, etc. No difference between those two groups.
And the explanation the research is for that which I find really compelling, is one is premeditated and one is impulsive, so you know you're going out, you're going to commit a crime, You've thought about it, you've planned it, and violence is part of the plan, whereas domestic violence is more often impulsive, so it's a reactive violence, reactive to
the situation. And so you would expect that if you've got those two types of violence, the one that is under impulse control is the one that's going to be affected by having more serotonin available. So I think this research,
which is a significant step forward. It's a major study, big Australian study, really high quality research, essentially proves what people have suspected for some time, which is impulse control is something under the control of brain chemicals and in particular serotonin and things that affect serah.
And are going to have that effect.
And I think that's an important thing because it steps away from the traditional explanation of this, which is much more wooly and much more oh, you don't know how badly he was treated by his father kind of thing, and which is of course one of the reasons people really didn't like this rich that is being done at all.
People have there are certain groups who really don't like the notion that what we're going to do here is come up with a drug treatment for domestic violence, because it's because it's it's sort of given as it suggests then that what's going to be put about is that this is some sort of jed get out of jail free card. Oh, it's all right, he's you know, his brain's broken and we can fix it with drugs and he has no responsibility for the way he behaved.
Yeah. Look, well, I don't think the research kind of shows that it's only about brain chemis and you're not suggesting that either. But I think also when somebody grew up in an environment where there was domestic violence, and you know that was they were socially uh an environmentally kind of influenced and programmed by that, that becomes you know, emotionally and behaviorally more acceptable. Like if you've been around that, if you've been around that, you're more likely to do that.
Ah, you're straying into psychobabble again. Crazy.
Oh, but it's you're.
Kind of right, but you're just you can't help yourself. You steer off into the psycho.
Babbyl No, no, no, I'm going to push back because I think that I mean I've been involved. I've worked at the cold face of addiction and alcoholism, which intersects with domestic violence, and so I have a little bit of expertise working in this space with offenders and people who have been hurt. And let me tell you, the people that come out of that environment often just become
a product of that environment. And I'm sure the brain chemistry thing is indeed a variable and a contributing factor, but I don't think like when you look at you go, well, even in this research, twenty nine of twenty nine percent of the people, so twenty nine percent of three hundred is about ninety people give or take who were taking the drug, the actual drug, they still offended. So you know, it's not a solution, but it's definitely a factor.
I'm not saying it's a solution either, And I agree with everything you set up to the point where you started to try and explain it.
So so I.
Have a go.
If you can't, you can't say I don't have a go, you have a red hot crack. I have a go. I have a gor right.
An environment where a child is raised in an environment of high levels of domestic violence, that isn't that is a that is an environment of chronic stress and that definitely has an effect on brain chemistry, definitely. Does We know that it raises delta foss B. It is exactly the same as the child being addicted. We know that from the Bradley studies that started with the ADHD drugs
that we've talked about before. Those kids were depression era kids being raised in large families with alcoholic parents and high levels of domestic violence. Those kids were ADHD and what they found was the cure for them is take them out of that environment and they get better. That's a bit too slow, though, So what they found even better was that if you give them a stimulant drug, you'll cure them. Well, your cure ADHD. And that's the basis of the ADHD drug industry today. But that just
tells you that you're right. The environment did create it. The chronic stress of the environment did create it. But what it created was a brain chemistry that can be affected.
So you don't believe that psychology plays a part at all. You don't believe that social programming and conditioning influencers people's behavior as a factor alongside brain chemistry. You think everything so there's no free will, so you're not making decisions.
There's always free will at the margins.
But if you want to look at if you want to look at the major causes of mental illness, it is always brain chemistry. But there's a high variety in humans, right, We're all on a normal spectrum. Some people are at one end of reaction and others are at the other end of reaction, and that's because of DNA. But if you want to look at the average, you have to say, the vast majority of the way we react is entirely driven by our biochemistry.
What did the what did the paper se like in terms of the conclusion, You know how they always go conclusion, future research, future direction. I don't know if you they.
Want to ex end the I mean this is this is a major piece in what has already been known, which is that if you increase the availability of serotonin, you can reduce impulsive violence. And what they've done is really improve this in a substandard way. And like you say, more research required, let's look at this further.
What else can we do to take this further?
You know, we're using an SSRI to do this, but are there other ways that you could change serotonin.
M it is interesting. It is interesting. You're wrong, but it's interesting. Well, it's good that you having to go. I love that you try. I mean, you come on, you have a go, and I think that I'm just fucking with you. That isn't psychology doesn't.
No, I'm not any expert here. I'm no more expert than you are. You're probably more expert than I am. It's just wrong.
Yeah, and that's why you listen listeners because there's no prescription here. There's just two blokes and occasionally a woman who chimes in, just having a Yeah. Well that's interesting, mate. I wonder where it's going to go, and I wonder what that might lead to. And I wonder whether or not we're going to see us a more open minded approach to.
By the way, I don't want anyone to come I don't want anyone to come away with the impression that what I'm saying here is that we should instantly go out and start giving people antidepressants to cure domestic violence. I don't think that's what this study says. As you quite rightly point out, twenty nine percent is still a very high number. And the interesting part about this is that it could be affected at all.
That's the bit I find interesting.
Yeah, yeah, well it's definitely doing something I was going to say before we go, just jumping back quickly. So just your your website and your free resource that you kind of generously and passionately give to the world, which is free schools dot org. I think, off the top of my head, is that correct?
That's right?
Yep?
Is there are you using like I think that's maybe a good use of AI? Is Is there any AI kind of involvement in the the that not yet?
Right, there is.
We do have plans around that, I think, using AI as a responsive technique. So at the moment, the website just contains video, so you know, really well expert teachers in their field are giving you another view of the lesson that you didn't understand today when your teacher you know, explained it to you or you didn't show up to school. But yeah, many and you can choose from you know, five ten teachers. Get your favorites. Is the one you
like that's teaching the thing. That's all well and good, but it is still just being talked about by a video. And I think the potential for AI here is to have a conversation about the content of the video instead of the video, so to be able to actually interact with this as if it were a teacher trying to
teach you the thing you're not understanding. And that's the nuance in AI which I find really really powerful, is they're getting to the point now where you can have that conversation and they are accurate enough to be reliable. And that's what I find really interesting in that space.
Not many people know a few of our listeners know because we've mentioned it briefly once or twice, but most people wouldn't be aware that you were involved in tech, and like, I think pretty sure that you can code in all of those things if you have to back it right, so you would have been way more aware that Tiff and I will definitely I, but I would think TIF in terms of that AI was coming. Has its rate of development surprised.
You, Yes, and no.
It's surprising how really quickly it's developing, in the sense that it is now generations going every month that goes by its generations better than it was before. In a way, that's not that surprising, though, because once these things hit that sort of level and start growing exponentially. Then you start to see that kind of growth really really quickly. What I think most people are not appreciating is how much and how fast it's growing because most people don't
use AI much. What they use it for is to write emails that they can't be bothered writing. They use it to tell them how to do a Facebook post, and you know that kind of stuff, and it could do that standing on its ear. The really interesting stuff is stuff most of the population are not seeing. When you start to talk to programmers and coders, and I do because I still do have fingers in those pis. Low level programmers are simply ceasing to exist because AI
does it all. You're left with just a generation of fifty something year old programmers who are the absolute experts
and who are using it themselves. I was talking to one the other day who was telling me he wrote in less than a month something that he would have even two or three years ago taken two years to do, mostly because he's just telling it what to write, and then he doesn't actually have to do the coding himself, and he's smart enough to know when it's right and when it's wrong and get it to correct but he says his fear is that we will lose the ability
to code ourselves. He said, where are the next generation of programmers going to come from? Because nobody wants them because they are all AI now. And most people don't see this, most people don't care. But when AI is writing all the code, then we have really lost control of what's being created.
And also, AI doesn't get tired, doesn't have bad days, doesn't get hungry, doesn't need to go at the toilet, doesn't need a lunch break, doesn't have a blue with its missus.
And can have a million instances running at the same time, or a billion or a trillion instances running at the same time. So your one programmer is a trillion programmers. The productivity cycle there is truly massive. So that stuff is the really scarily fast developing area of AI, which most of us don't see because we just think, oh, it writes a nice email.
It's a good thing. Podcast is irreplaceable, tiff, isn't it?
Well, for now, we're definitely give it a minute.
We're definitely replaceable. There's already AI podcasts like for We'll say goodbye affair and I'll tell you that thing off heir, but appreciate you and even though you were wrong, you had to go today and well done.
Oh look you got to give me a gold stuff
For you.