The AI Coding Trap: Why 5,000 Lines of Code Might Be Worthless

People who have jobs and they go I'm not using AI in my job because the competitor will or their employer will be going nope you can be more efficient if you use AI for admin we need to learn from the past like when the linen industry was automated it didn't get rid of all the jobs it created more jobs

design a website and he used AI

and there were big issues with it and he's ringing him going there's an issue with this website can you fix But the guy couldn't fix it cuz he hadn't got a clue about program.

Exactly. Exactly. And AI is it hype or help? Good day and welcome to the show. I'm your host Pete Cooper and our topic is AI hype or help.

Today's guest is Patricia Mcccluren from Drumland's Analytics. She has a deep history in AI going back to 1997 when she did a PhD. So she's a veteran in the topic. We're going to be talking about sensors, AI, how it's used in industry, and lots of other topics. Let's get into it. Hi, Patricia, welcome to the show. Patricia McLaren from Drumland's Analytics. Patricia, you have a PhD in AI going back to 1997 when most people I think were probably not even aware of AI and its emergence. So, let's start off with a little bit of your background. Uh tell us about how you got into AI, what that PhD was about and we'll take it from there.

Okay.

Um I found myself in Birkbeck College, University of London doing a master's in analytical chemistry

and then I stayed on and I I started a PhD and the the title of the PhD was interpretation of sensor responses using statistical analysis and artificial intelligence. So the the basics of of the project was to build an artificial nose and electronic nose which consisted of eight sensors. So it was called a sensor array and to interpret the data coming from that electronic nose or artificial nose as it was called

we used uh conventional statistical analysis but also neural network neural networks.

So, so that was so nowadays people refer to it as artificial intelligence which is built on neural networks.

Yeah. Fascinating. So, is such a thing as an artificial nose available now? Is it something that's out there now? 1997 was a while ago. So, it's been over what 25 26 28 years.

Yes, indeed. Much more advanced now given you know advanced in technology and data processing. Um, artificial noses are available for various applications from monitoring the quality of say a perfume or the quality of a food or it's actually they're looking at using it to replace dogs sniffing out mines in countries

and also for medical applications.

Um like in 1997 I did an a study of a patients breaths to monitor the detection of Helicobacter pylori infection in the gut. So my nose in 1997 my artificial nose in 90 was able to distinguish between someone who hadn't HP highway pylori infection and someone who had but nowadays it's so much more advanced the technology the sensor technology has improved the detection limit has gone gone up so you can detect much more at levels of parts per billion

right so you you mentioned a few interesting applications there. You said that your nose, do you mean your physical nose on your face or do you mean your sensor that you created?

My the the artificial nose that I built.

Okay.

So, if we're talking about my nose, we're talking about your artificial nose.

Yeah.

All right.

Yeah.

So, I mean, that's a good point. I mean, can you compare your physical nose to the one that you created and say yes, it's doing better or worse than than a human nose?

Uh, an electronic or an artificial nose can do better because it you see if you're going in say for a qual in in a factory and you're doing quality control and you're going in every day to smell a product and determine the quality that is made based on your human nose. M

what happens if you go in and have a cold like I have a cold today so my nose is my smelling capacity and is impaired.

Yeah.

And and also when you smell something it depends on your mood like if you're smelling the quality of a perfume

you know it it depends on it whether you you feel oh I want to smell a lovely and but sometimes you you know you don't want to smell an overpowering per perfume. So the electronic or the artificial nose is much more powerful and it is not affected by a human's mood or the their their health at the time.

You know I had a very interesting experience in my 20s. I went back to visit an auntie. Um this is in Wales. Um we used to go there for Christmas.

Um, and I walked into the house and I there was hit by the smell of the house, which was pleasant, and I said, "Oh my goodness, it reminds me of Christmas. " I just had this overwhelming sense of of being back there when I was a child uh enjoying Christmas. Because I think a lot of the time we take for granted our sense of smell, don't we? Until uh until something like that happens and it just really alerts you. Or if you've got a cold and you can't experience certain things like the taste of food because you can't smell it.

Exactly. Yeah. Um yeah, definitely like your health affects your smell.

If you do lose your smell, apparently then your other senses are are adapted and improve a bit better like losing your sight. But I know there are there's research been carried out that people who have lost their smell,

they can actually get sensors embedded in their nose and it's connected to their

Wow.

Yeah. I mean, it's

amazing. My father lost his smell as he getting into his senior years.

He put it down to he was a carpenter and he used to work with these very powerful glues um that was used for laminating um kitchen tops and things and he felt that that was one of the causes but I don't know whether that's the case or it was just you know the factors of aging that but he found that he couldn't taste food you know so he wasn't enjoying life as much because he couldn't enjoy the taste of food. Yeah,

you can also there there's research now and you can embed sensors on your tongue if you've lost your taste. You can actually have sensors embedded in your tongue.

Wow.

And again, that's connected to your neural pathway. And isn't it?

So you your brain will be told um okay, you're tasting something acidic. So that signal will go and then your brain will go this is acidic. Maybe it's too acidic. You don't know because you're you've lost your taste, but but you know your your brain will will give you a signal.

That's amazing, isn't it? Like how

it is amazing. I mean, the idea of putting a sensor into your body to replace something that's that's is needed is is like science fiction, but you actually have some very real world applications. You mentioned perfumes.

So is this a way of keeping quality control so that you can just put this sensor across, you know, one in a 100 bottles as it's coming out of the factory and make sure that it's giving the right smells.

Yeah.

Yeah, that's that's certainly an application. In my PhD, I looked at distinguishing between real and fake perfumes.

So the electronic nose was able to detect because there's a lot of um fraud in making fake perfumes. It's like

I was studying in London and you'd go down Tottenham Court Road or Oxford Street and there'd be people selling fake perfumes and there's a a lot of millions of euros and dollars are lost through, you know, people making fake perfumes.

But the whole idea of um like the real and fake perfume and you should with a human nose you can tell the difference because if you put a fake perfume on your wrist

the the high notes will go very quickly. So you know that but with a real perfume the high notes are retained because it's it's developed not

a high note in a perfume is when when uh you've got the organic vapor coming off first.

Oh yeah.

So so your high notes come off first. So think of evaporation.

So so you get that initial smell. So that's the high notes. And then as you were sort of a real perfume or a fake perfume during the day then all the high notes go and you're left with low notes.

Oh,

but the low notes in a fake fake perfume may be totally not detectable.

So that's how you tell the difference. And yeah, the the array of eight sensors like in my system, we used what were called quartz crystal microbalances and we had eight of them and each of them was coated with what was called a gas chromatic stationary phase and it was able to pick up what was in the head space of any of the samples I and the the it was a called a quartz crystal microbalance a QCM device. So it would oscillate at a particular frequency. So my my system oscillated at 10 MHz. So when a vapor came in contact with it,

the frequency decreased because it couldn't oscillate as much.

So the shift in the frequency

on detecting the vapor

was indicative then of of what we were identifying. So the idea of having eight is that you have a fingerprint,

right?

And then and a finger so you're you're having uh data from eight sensors, not just one. So that each sensor doesn't have to be very selective.

M

so and you get a fingerprint and you get an overall fingerprint of the odor by the vapor and because you've s got so much data coming in

you need multivaried statistical analysis and that's where your principal component analysis comes in but more importantly where artificial neural networks are used you know so that was yeah back prop and soft organized mopping and yeah you know your the title of our talk today is help or hype. Like having done a PhD in 1997 and living through what's happening today, it's hype and they're making it like it's something, you know, really difficult to understand and it's so basic.

It's just just some algorithms and

there was a lot in there that I just wanted to unpack because I think there was some really you mentioned there were eight senses. I'm going to try and put that in a way that may be easier for the lay person or just general public to understand that the human eye does the same thing. It has three colors, red, green, and blue and it mixes those those amounts. You know, the brain, human brain mixes those amounts just the same way it does in a camera to create all of those colors. So when you look at something, you go, "Okay, I know it's that color because of the mixture of the red, green, and blue.

" And the you can do the same thing with your sensor, but you've got actually eight eight of these little sensors that's giving you what you call a fingerprint just the same way you can with colors um from the red, green, and blue.

Yeah. That's exactly. So the the idea is you see the the coating on each of the sensor could not be really specific enough.

M

so by having eight with nonspecific sensing materials on each of the eight

yes

you get more of a yeah identification of the the odor or the sample you're

right so I think this this comes down to sort of the AI component of this so you would have a known sample of a certain perfume aroma that you're trying to identify and then you would expose that to your eight sensors and you get all this data at the back and then it goes through the the AI neural networks and it says and you keep training it with all those different variations and it says out of this data I can predict this with a certain level of accuracy and then you give it something else and so you you're creating all

these classes we call them don't they so a certain perfume is a class

another perfume is another class and so you could have many many many is there a limit to how many you know different types of perfumes you could sense.

No, no. Given the computer power we have now, it's unbelievable. Like

back in 1997, computer power was very limited. And you know, I used to have data sets of only 10 samples because if I had sampled any more, my computer couldn't cope. The neural network software algorithm, it was called NeuroShell. Actually, they're still about. I I I looked looked them up and I said, "Oh, you still doing that product? " Oh, yeah.

We And it was called NeuroShell and it was, you know, very limited, but it it it carried out the math the mathematics.

Right. Right.

Yeah. It carried out the mathematics, you know, and so

So, what tools are you using now? I mean, something like TensorFlow or one of the Google Suite stuff or what are you using now for the training of the models? Um me personally, I don't use AI personally. No,

no, no.

I have retired from that game altogether. I Yeah.

All I do all I do at the moment now is write in the AngloEt and Irish newspaper which actually has a global audience really

and I have my colleague who's worked in the banking industry and he's the mathematician that has designed the AI model. So that's what Drumland's analytics does,

right?

So basically we focus now on using AI

to predict the the rate of a currency per in the forex.

Okay. So that's taking us to a whole other area. So can I

Yeah. Can I could I use um Drumland's analytics to you make money on on the forex exchanges?

Yes.

If you listen to my colleague called Kanti Prasad

and he's uh bringing out a book in June this year

and you'll be able to learn how to trade even if you don't understand anything just starting from scratch. You can go through the book. It's getting really good reviews at the moment. Um and then what we plan to do is actually set up what's called a a trading school online. So what we'll do is we'll use Kanty's book to go through and teach anyone who wants a second income or they don't like their job, fed up with their job and they want to give that up and work from home

trading in the forex market.

So let's um make sure everyone knows what forex trading is.

That's foreign currency trading. So you're trading off two pairs like US dollar and you know Australian dollar or something like that and you're making You're trying to make money on the shift, right? Back and forward.

Yes.

Yeah. You're trying to predict when to perhaps buy a product

or a service

when the currency exchange rate is more profitable, more attractive.

Yeah.

It sounds like magic. I mean, how does it work, Patricia?

How does it work? Basically, it's the AI model

has been built over it's nearly now two years been trained on data that has come from the internet and looking at how the currency purge has fluctuated and changed.

Yeah.

Even when even when Trump, you know, makes an announcement he's going to do something really unusual,

right?

And you have like a kick in the market with the currency pairs. The AI model tracks that and has been tracking that over the last year. So blips like that are in our package or in our model. So but it's hard to predict, you know, when someone with significant power is going to cause a shift in the rate of a currency per

you know. So

you're taking input from the news broadly. Um you're looking at I guess the history of that currency pair.

Yes indeed

over a twoyear period.

Actually the you looking at the inflation is there going to be an issue is issue in a country in terms of inflation if there's any um believe it or not the the time of the year affects traders. The energy and um people's people's mood affects the way the currency pairs go and shift.

Wow.

So why currency pairs? Why not just um you know trading on the stock market?

Trading with in stocks is more risky. You get a bigger shift. You can't really predict what's going to happen in a particular market. But when you come to the shift in the change in a currency per currency rate,

Mhm.

it's it's significantly less.

So you've less risk because there's less of a shift,

right?

But then I guess there's also not the gains. But if you're making steady gains over time,

well, that's where maybe Yeah. Well, that's maybe where if you're trading say with 30, 000 euro over say three months

making making 30 grand. So you're you're sitting in your account 60 grand.

So

you can't get rich overnight. But and you can't go and trade in currency pers or in the forex with $100 or $100 euro cuz it's you basically need a good amount of money

you know even trading with a thousand you see you can't what's the word you can't cushion your loss on a thousand but when you have a when you're somebody said gambling forex trading is definitely not gambling.

We need to get that out.

And Kanti is going to do a presentation tomorrow and he's going to focus in on that. It's definitely not

So, you said something in there about you can't do it with a small amount of money.

Do you mean you can't expect to make a living out of a small amount of money or you just need to have a certain amount of cash for the whole thing to work? You can't make a living out of trading with $500.

Yeah, exactly. Okay.

You need to Yeah.

You could step in and get familiar with it over a period of 3 months with a small amount.

Oh, yes, indeed. Yeah. You can you can um of course create a dummy account.

You can follow Kant's tips.

Yes.

On when to buy and sell.

Right. A dummy account. Is that like paper trading where it's just just numbers in an account and you see how it goes?

Yes.

Yeah.

And you're not it's it's called a dummy account and it's not you're not it's not real money.

But you can get you know a h 100, 000 in a dummy account. So for six months you could be playing around with a hund, 000

dollars.

Yeah. Yeah. And see how you're performing.

I'm very interested in myself and I'm sure our audience would be.

So

maybe I can put a link in the description to the drumlands analytics and uh they can I guess get the training, they can buy the book, they can or I can uh and do some paper trading and then ultimately if I like it enough I can do some real trading.

So you're not losing anything.

Right. Yes. But the most important thing is with um with trading, you need to remove passion and you need to remove your influence and your you know if you're one day you're thinking I'm going to go away from my plan my my trading plan.

Every morning you wake up you should have a plan and stick with it and don't start diver

because that's unfortunately that's when yeah when you lose so the advantage of this AI model

that Kanti has developed is that takes away all your personal

issues

right

you know not predictable you know you say oh I'm feeling lucky today that 100k I'm going to diverge from my planning. You know that you have a recipe to

do exactly this and and then that's where you lose. So

yes. Well, I think we've all impulse buy it at one point.

Impulse. That's a word. Impulse buying and impulse.

Yes. So that's not impulse buy when we put 100 grand down on Forex trading.

So what I'm hearing is a small investment in time. You'll learn something out of it. Um you don't even have to risk any money until you get comfortable. Well, and then there's perhaps a future for doing forex trading as a separate income and who would want all those things I think.

Exactly. Yeah.

Great. Really

and and Kanti is uh very much into ethics and he's um from India

and he wants to help people like he wants to help people learn how to trade and make a profit and feed their family. Right.

You know, um,

and there's plenty of very very wealthy people um, who who've made their livingings on trading. And so there's nothing, you know, the sky's is the limit.

And I think is it honorable or not? I think yeah, it is an honorable way to make a living. Some people would say, well, you're not adding value. But buying and selling is a business, you know, and I think it's as valid as any other. And and David. Yes.

So I got a question here from our list here about how AI has changed since you did your PhD in 1997.

Yes.

How has AI changed?

It was very straightforward in 1997. There was no hype and it wasn't, you know, people didn't expect much from it.

Now with all the hype, it's actually made people afraid of it.

I I you know I'm out every day and I'm I meet people and my target audience for teaching because I teach about a I introduce AI to the older adults and you know they're afraid of it. There. One of them said I I was chatting talking it was actually a lecture front to 30 elderly people

and one lady at the back put her hand up and went is AI an alien and she was serious and I was taken back and I went no and I didn't laugh at her because that that actually represents what every a lot of other people think that it's not human

and I say AI is not an alien because it's made by humans. So, it's not an alien and it's not to be afraid.

And I tell all my audience like and my readers, you know, my readers uh submit comments on on my social media when I post my article and they're like going, "But we're afraid of it. " And I says, "No, the worst thing people can do is bury their head in the sand and say, I'm not going to use AI. I'm going to hide from it. "

Right?

Bad idea. Especially people who have jobs and they go, I'm not using AI in my job because the competitor will or their employer will be going, no, you can be more efficient if you use AI for admin,

you know. So I war I do warn people not to ignore it

but to embrace it a little bit about those those fears.

I mean being an alien one is certainly straight out really out there but you know Elon Musk is a big proponent for AI and sometimes he's called an alien so maybe that's where the thing comes in. But where

where are the fears founded? What what are people saying to you when you teach them? Where are their real fears that they have?

Yeah. One is uh scams. They they're hearing about there's AI is used in phone scams,

right?

Email scams. So that that's that is one of the key ones. Second one they think

that sounds quite real though. That I think that could be.

Yeah. A real one. Yeah.

And then the the second one is when they they fear it.

They they think it's going to take their job. Or it'll take their the kids' jobs and there'll be no jobs in the future. There will be lots of jobs in the future but unfortunately routine and routine jobs in industry and factories will be replaced with AI and robot AI powered robotics

you know but but you know so

but they said that about the internet they said that about all sorts of other technologies didn't they computers jobs and

what happened was we adapted didn't we so we embraced the computers and and the other things and we moved on.

So, one of my

staff came to me uh actually it was only this week or last week and he said to me, "Pete, we're using these tools, these AI tools, um Claude and a couple of others, and he says, I'm getting 5, 000 lines of code done in a week. " Um, and that's about sorry, 5, 000 lines of code done in a day, and that's about what you could expect from a week. So he was telling me that he's running really five times faster by using the AI tools.

I did hear actually that they have to be careful about that because I was talking to a programmer and he said, "Oh, just be careful because it's not generating really good code and there's lots of errors in it and you need a good programmer to go through,

right?

You want to pick up the errors. " Um now I did I yeah he he gave me a bit of a negative view on it. He said no you still need the good programmers who are going to have to check all that.

Right. Well I've had that problem too. I had uh somebody seemed to be very good at creating out of the box a great user interface um for some of the projects that we were doing.

Um one of them interestingly was very similar to your forex trading. We were doing something on the tech stock side.

Uh and the user interface was fabulous but once we started hitting some problems they were became unsolvable so we literally within the first few weeks we had what I felt was 90% done but we could never close that gap the 90 to 100 was month after month six months you know I had to cancel it because there was just no way we could close the gap fundamentally the mistake that was made is exactly what you said he didn't understand the code that was being generated for him and it looked like we were making progress but actually it was mostly just a veneer on the surface and all underneath were all these problems which you just couldn't solve and in we had to

so we you know we have our own

ways of doing code which we've always done and we still do it the same way we we can mitigate that by dealing with the problems early on not we're leaving to the end right and that's general project management

um AI isn't a replacement for project management it's in It makes it more important to to get the project risks identified and work on the hard things first so that you can definitely solve them and then have people who are good problems solvers. I think that's the other thing. Using the tools makes us feel like we have a superpower but of course once all the problems start coming up in the code we don't feel like we're a superhero anymore. We feel the opposite. Yes.

And when you do have AI generating code, if something is wrong in the code, because you haven't written written it, it's hard to work out where the the the issue is. I remember talking to a guy and he got this person to design a website and he used AI

and there were big issues with it and he's ringing him going, "There's an issue with this website. Can you fix it? " But the guy couldn't fix it because he hadn't got a clue about program.

Exactly. Exactly. But, you know, that's that's not really a new problem. I mean, I've got I'm working with a client now who's been he built code the traditional way with a couple of I guess they were young graduates.

They built a whole load of code and there was a whole load of problems with with that as well. Now, that was not using AI. That was just that they weren't experienced coders

building to a specification. And they were sort of hacking away and creating all sorts of code just to get something to work and then it's created enormous amount of problems which we have to completely unpack.

Yeah.

And we're actually going to use II to help us do that actually to to unpack what's already been written so we can diagnose it

and then we can fix it.

Interesting.

Yeah.

And back to one of my articles and we were talking we touched on the textile industry automation you know and learning from the past

you know for example like an from Lur and county Arma and it was deemed um the center for linen production

right

and then in in Scotland it was the same and what came about was there were these uh ludites they were called ludites

which are people who are resisting change

because they're afraid of losing their job and their income. Of course they are. It's it's natural.

But so now the term is coming out. It's called AI ludites.

And I wrote about AI ludites in my So AI ludites are about people.

So remember ludite terminology came from the the textile industry in England. I didn't know that. So you've got AI ludites are people I meet every day. And they're they're they don't want to progress with AI. So, they're going, I'm not going to use AI. And then they take out their iPhone and I'm going, well, you're using AI, you know, you're Google. Um, but I think we need to learn from the past. Like when the linen industry was automated, it didn't get rid of all the jobs. That created more jobs,

better quality finish and you know so we need to learn from the past. It's a bit of a paradigm shift with AI technology.

We're going to be doing things very differently but learn from the past and embrace AI not to be

I mean it is very challenging because for the first time in history we now have an artificial intelligence which is in some cases better than human intelligence. I'm not saying in all cases but in some cases and I think that's where we get you know we're so used to the fact that being smart in our brain is what I you know distinguishes us from the animals and the machines.

Um now that's being challenged and maybe that's where the source of the fe theory is that you know if we can't

you know if we can't stand out on our intelligence on our brain then how do we stand out

but you know AI will always rely on human input.

It can't live on no data.

The problem is there's a load of rubbish data been put in even the ChatGPT. That's why you're getting a load of hallucinations and errors coming out. Yeah. But

so what you feed in is what you get out. So if you're feeding in garbage, you're getting garbage out. But AI will never replace the human because they are so creative. You touched on analytical. They're the they're their problem solving.

M

you see AI needs supervised and you'll always have the human more intelligent in terms of creating new ideas because that new idea is not in the AI model because it hasn't it doesn't have that idea yet.

So if the human is more creative and that's why

we you know humans will never be replaced with AI

because we're we're clever we're more intelligent than it.

Yeah. I think they're on to something there. There's a there and you said it in the middle there about it the AI can only respond to what's available in the digital world on the internet.

Now there's an enormous amount of data there but now we're finding out that there's more data being put in there by AI which is kind of

reinforcing what's already there not really adding to it right and we live in the real world. We have

infinitely more inputs to our brain as we grow up than the AI can even in the digital world.

I mean it can be very good at that but we can bring new things in like

I can have a face tof face meeting with you Patricia which the AI would never know anything about right and I might learn things about you

physically you know auditory and visually you know whatever but the AI can't possibly understand I always think that element of

human interaction is going to become more important right because the AI cannot do the human interaction Only humans can do the human interaction.

That's so true. Yeah.

That's where your spiritual and getting the balance between your natural intelligence, spiritual intelligence and artificial and get getting that right because if you use too much AI you know it'll affect your spiritual intelligence

you know engaging with you know chat bots and deciding it was too difficult to make a friend a real human friend so let's let's make friends through a chatbot Actually on Sky News this morning, very interesting. They're talking about young boys are using more chat bots more regularly than young girls. And young girls are meant to mature, you know, more quickly than the boys.

But the boys are using this to build relationships

and so they want to actually improve.

Yes. You know, relation, but they're going to chat chat bots for it.

Wow.

So, that was amazing. I I I thought, oh, Sky News is great. I love

They talk about AI every morning. It's really

I learned so much.

Maybe I should tune in to Sky News.

I do. They talked about the AI and Yeah. Kids.

This morning and I thought yeah

well you know um we are probably developing relationships with AI as we don't even realize I made a commitment to using AI more and more like I felt that I wasn't using it enough that I could um get the leverage that I was that I think I could get out of it and of course the more you feed into these tools like ChatGPT the more it knows about you and so it's it's starting to pull things up from chats that I had weeks or months ago and making them relevant.

And so it's a little bit like I'm having a conversation. The word is a partner like treating it as a partner. And I try to take that mindset.

I if I got a problem and I would perhaps talk to a human

or I can use ChatGPT. I'm not saying it's a replacement, but I think

in some ways I can open that doorway to having a discussion which I can add to it knows my history and can help guide me to some new ideas or

some ways of framing it or a bit of a sanity check with the types of things that I would do with a business partner.

Right now the yeah I did write one article about actually was two articles AI companionship. So it's a mental health app. I tried I actually tried it myself. I thought it was brilliant.

Um but but you know when I wrote my first article oh the was it went viral.

You know there was a lot of people saying you know you shouldn't be talking to you know a chatbot. You should go and see a psychiatrist. I said people don't have money to go. Not everybody can go and pay 20030 to speak to somebody a doctor. And in the first place you might get a doctor.

So the advantages of AI you know for mental health and chatbots

y

it's amazing you know and like we are working at the moment

with um Sebastian No and his AI companionship

and the mental health app and it's to actually

yeah it's it's actually through an organization I'm a member of and I'm a representative in Ireland here for woman for women OSMT and we're looking not getting this mental health uh to women who are in abusive relationships

who have no one to talk to,

right?

Um and it's a private conversation with with chatbot and it actually records um the every conversation that you have with a chatbot is recorded in a journal. So you're you're journaling.

You know,

a lot of people who are suffering from mental health don't feel like getting a pencil and paper, a pen and paper and writing down their feelings.

But the phone they talk to it and then all these are all then

um these journals are all

produced and you can print them out if you want and

wow

have a diary of you know a year's worth of conversation with this chatbot and and the

and the advantage of that AI chatbot is

it actually analyzes everything you say you know in the morning you're feeling a bit low little thing

and then it will go from your chats with me over a week

you know I feel that there is an issue here and maybe they don't tell you to do anything maybe you could go and go to the library and research this or go and meet somebody that can help you with this oh that's amazing

that's amazing you know it

made me think of uh something I observed I'm from Australia but we used to watch American shows and they were always bringing in the psychiatrists. Um, and there was shows about psychiatrists and it was sound to us that

that the person didn't need a psychiatrist, they needed a friend, right? They just needed someone to talk to.

Yes.

Um, and I think American society uh struggles with that a little bit that you don't people don't feel like they have this strong friendship bond. We've been living here for seven years.

Better in Australia where I come from, but I think this has improved. I I honestly think that that there's sort of a shift away from the need for a psychiatrist.

Yes.

Back towards some other forms of of connection and needing to know. I mean, so it's so human to know that you've got a connection with another person. If you if you're isolated, you're going to go and pay for it through a psychiatrist or you're going to work on having friendships that you can talk to. Now there's kind of another opportunity which I think you've opened the door to which is that you can talk to AI and feel that you have that connection and that satisfies us and can guide us a little bit towards perhaps a healthier life.

Oh, it's amazing. I highly re highly recommend it.

We're bringing it, we're planning to bring it into women's prisons um in Ireland here because you can imagine a woman in prison for a few years and she's hearing in the outside world all about AI. So, so and having having like to be able to converse with somebody,

an AI bot, you know, and

she can't talk to her fellow prisoners and, you know, and like the the platform that I'm involved in with Sebastian No, you know, there's no security breaches like it's well regulated,

right? Um now one of the questions you you did um suggest to think about yes

was about um is the safety around ethics around AI oh is the regulation moving quick enough?

No, I actually thought it's like a horse is already bolted from the stable, right?

And and all the regular because I've been to Trin and Italy. I've been to France um and I've been to India

and you have hundreds of AI experts from all the big companies who use including Amazon, Google

and they're all sitting around tables, right? Talking about regulations and I'm going when we put in a regulation this week there'll be a software out there that's already pushed through to the market.

Yeah.

We can't we can't keep up and there

Yeah.

It's it's um

Are there any particular regulations you'd like to see that that um that are not there?

Well, I think overall the effect of AI on society and you know individuals you know and losing your privacy rights and

you know like when I was there I was really the only independent person there.

The people who were there were all paid big wages by big AI companies and I I wasn't. So I was there observing and they don't care about the P Joe blog in the street and

old people been left behind and so they really do need they need representatives for the older adult, you know, to to put their views there.

Now, the National Standards Authority of Ireland are meant to do that,

you know, they're meant to be talking to the older people and and they've, you know, had campaigns out there where they had a postcard and they put this postcard in all the libraries in Ireland and on that postcard it had questions to the older adult,

what are your main concerns about AI?

M

so and and they were able to then the general public were able to write down their concerns and post it off

and then the idea is when the National Standards Authority of Ireland go around the world to all these meetings they're meant to say yeah we have a we have an issue here with older people but

it's not very well represented

I think there's a temptation to use AI for ga what I call gating functions most obvious one that comes to mind is job interviews. A friend of mine called me. He said, "Yeah, you know, I've got to get through all the AI interviews first and you don't know what they're doing, right? What is an AI interview doing?

I mean, is it screening on some something that is unethical like race or gender or something like that? I mean, is it what is AI screening on? " There's not like a human who has accountability. And this is I think where the biggest problem one of the biggest problems with AI is there's no accountability. It can do what it wants whenever it wants and everyone just says well okay let's just kill the instance or change the software like when a human takes responsibility like driving a car for example. You can be prosecuted if you get in a car drunk and you go and hurt someone. I mean you you are liable and you're accountable and you will pay a penance for what you've done.

That keeps things um I guess equitable, right? You you're driving a ton vehicle, it's dangerous. You have to be responsible for it. If you hand that over to AI, who's responsible, you know, and there was a case just recently with a fellow, you know, who uh couldn't get out of a Tesla car, you know, and died. And there's been many multiple cases where the AI vision system hasn't worked correctly. Who's responsible? Well, you see the they are setting up guard rails and companies will be heavily fined if they put out a product based around AI and it it harms someone, you know, they will be heavily prosecuted. It's

Yeah.

You know, it's Yeah, but it's hard to control, you know, and then you've got all this um fake news and, you know, images being put out, you know. It it's do you know what it is? It's hype help and a mind field. We're in a mind field. And um I've written an article about um when when the bubble bursts, AI slop, another term, AI slop. There's a lot of AI slop out there. And it means it's useless information taking up space.

Yeah. Taking up space.

Yeah. And it's so so when you've got and then you have the AI bubble which is going to burst. I I could actually uh trade on that.

Yeah, it will burst and um because there's been a lot of financial investment in companies and

the AI is is unfortunately not sorry unfortunately not delivering um what is promised and there's a lot of money been spent and they're going oh dear it doesn't suit

Patricia we're getting to sort of the end of the

I of our time for another hour.

We could we could maybe we should have you back um to do some more. Is there anything else that you wanted to add before we uh wrap up today? Questions for me or comments you want to make?

No, there there's loads more to talk about. You know,

there is.

Yeah.

So, let me give you the final question.

Let me give you the final question. AI, is it hype or help?

A bit of both. But

more hype than health.

Mhm. Okay.

You touched on it a little bit. What's the scoop? Why do you think it's hype?

Because I used it 1997. I think that's 30 years ago or something. 36 years.

And it's because they're thinking they're presenting this really sophisticated technology and it's you can't understand it. But it's really basic. They're they're thinking it's a bit of hype, you know, like it's hype. They're trying to make it into something that it's it's God, you know,

right?

So many mass, so many algorithms,

right? Yeah.

Elon Musk.

Elon Musk was asked about what an AI model was and he hadn't a clue.

He said it's a bunch of ones and zeros, I think.

Yeah. It's only in it's only input a hidden layer and an output layer and in the hidden layer you get all the mass

and you can follow the mass really easy.

All right,

that's it.

Well, Patricia, this has been absolute pleasure.

Oh, do you know what has been

on the show?

Thank you.

Okay.

Thank you.

We'll have you back again soon. Thanks.

Yeah. Okay. Thank you. Good evening.

Well, I hope you enjoyed that conversation with Patricia as much as I did.

Patricia has a deep history and a lot of experience in AI and it came from early work she did in creating an artificial nose to smell vapors. Now AI can certainly help that because multiple sensors go in and can be summated if you like or synthesized using AI. She also said that AI is not that complicated. People think it's complicated but it's actually not. It's just inputs a neural network block and outputs. And because of that, she felt that AI was perhaps overhyped and that maybe there's a bubble, an AI bubble that's coming. Well, if you want to know more about Skillion, please follow the links in the description.

If you want to know more about our guest, Patricia, or any of the products and tools that we talked about, follow those links, too. I hope that helps. Thanks for watching.