Ep 34: Critical Thinking, Empowerment, and the Untapped Potential of AI Adoption with Giancarlo Erra

[00:00:00] Welcome to Elevate Your AIQ, the podcast focused on the AI-powered yet human-centric future of work.

[00:00:05] Are you and your organization prepared? If not, let's get there together. The show is open to

[00:00:09] sponsorships from forward-thinking brands who are fellow advocates for responsible AI literacy

[00:00:13] and AI skills development to help ensure no individuals or organizations are left behind.

[00:00:18] I also facilitate expert panels, interviews, and offer advisory services to help shape your

[00:00:23] responsible AI journey. Go to ElevateYourAIQ.com to find out more.

[00:00:28] Hey everyone, it's Bob Pulver. Welcome to another episode of Elevate Your AIQ.

[00:00:43] In this episode, I sit down with Giancarlo Erra, a creative tech entrepreneur with a wealth of

[00:00:48] experience in AI. Giancarlo is constantly building novel AI solutions or teaching others how to do so.

[00:00:54] You can check out his website and some of his AI solutions using the links in the show notes for

[00:00:58] this episode. Giancarlo and I delve into the challenges faced by small businesses and adopting AI,

[00:01:04] exploring misconceptions about the technology, and emphasize the importance of education and

[00:01:09] critical thinking in navigating the AI landscape. Giancarlo also stresses the need for society to take

[00:01:14] responsibility for AI adoption more broadly and for the workforce to upskill in order to adapt to

[00:01:20] the evolving job market. Let's dive into this comprehensive and informative conversation.

[00:01:24] Hello everyone, welcome to another episode of Elevate Your AIQ. I'm your host, Bob Pulver. With me today,

[00:01:30] I have the pleasure of speaking with Giancarlo Erra. How are you this morning, Giancarlo?

[00:01:36] Everything is good. Thank you very much. Thank you for having me.

[00:01:38] I guess it's afternoon for you, I suppose, being in the UK. But thank you so much for joining me. I'm

[00:01:45] excited to get into the weeds with you because you're like an AI renaissance man, I would say. So by way of

[00:01:53] introduction, why don't you tell us a little bit about your background and some of the things you're

[00:01:59] working on? Yeah, I'm a creative tech entrepreneur. I've been doing that for now almost 25 years.

[00:02:07] And I'm a professional tuning musician. So you will just split my life between

[00:02:13] those two things. Staying on the tech side of things. Yeah, started as a developer and then,

[00:02:20] you know, business tech consultant, technical director, and then obviously entrepreneur,

[00:02:24] started my own startups. And I've always been involved with mentoring and tutoring. And now I do

[00:02:31] a lot of that around obviously AI, because it is what every business wants to know about. And

[00:02:38] yeah, I'm, you know, I run the AI summits in the UK. We've done, you know, Liverpool,

[00:02:44] Cambridge, we're doing, we're going to be London, we've done Norwich, Peterborough, and basically going

[00:02:49] all around the country. And I do some AI workshops for businesses. And I do some mentoring and tutoring

[00:02:59] as well. And I'm an advisor as well about all of that. And my obsession, and the reason why obviously

[00:03:06] is the AI summits in the summer obsession is the simplification. So my obsession is getting all the

[00:03:10] people that are not tech, and possibly they are business owners, and then helping them and explaining

[00:03:17] to them. In particular with AI, I always done it with everything, trying to find the simplest and more

[00:03:23] efficient solution to their problem. In particular, when we're talking about businesses not having,

[00:03:30] you know, engineers or big resources to obviously embrace technologies.

[00:03:36] Yeah, that's amazing. I'm not sure how you juggle all of that. That's, that's amazing. And when you

[00:03:43] said, you know, business owners, so we're talking about small businesses, maybe medium sized business,

[00:03:48] what's your sort of average?

[00:03:49] Basically, I'm very much, you know, and if there are startups even better, you know, as a founder

[00:03:54] myself of quite a few companies, I get very excited about the early stages of creativity and madness.

[00:04:03] I've never been a big fan of the enterprise way, you know, the bigger the companies are less,

[00:04:09] the least interesting it becomes for me. So yeah, usually the small, the small medium businesses. Yeah.

[00:04:16] No, I can totally appreciate that having spent quite a bit of time at large enterprises, but the last

[00:04:22] few years, you know, working with startups and, you know, talking regularly with, you know,

[00:04:27] entrepreneurs like yourself, as you help these businesses, I mean, what are some of the patterns that

[00:04:33] you see? I mean, are you seeing like specific things that are consistent or, or maybe differences across

[00:04:39] like industries?

[00:04:41] Understandably enough, the smaller the business usually, obviously the least amount of resources

[00:04:45] you have and probably didn't have many employees. So everything is a bit all over the place in a

[00:04:51] way, and you're always struggling for time and efficiency. They are faster than the obviously enterprise

[00:04:57] and corporate, but very often they are also struggling to understand how they can implement

[00:05:05] these things without the big investments. Because in particular with AI, basically everyone can use

[00:05:13] chat GPT or whatever available tools. The second level from it, it is all the big galaxy of apps built on

[00:05:22] LLMs, be it charge GPT or whatever it is. And then there is the third element that is, you know, your own custom,

[00:05:31] you know, I don't know, developing your own custom model, or even using an open custom model and then, you know,

[00:05:36] make it your own. That is what most businesses think they want. And probably I would say 50% of the time they're right.

[00:05:47] But that is pretty much out of their reach, because first of all, they have no idea how to do it. And once you go

[00:05:54] and you see consultants and businesses and stuff, the pricing and the complication that they have out of it, it is

[00:06:01] unbelievable. And as always happens, you know, it happened with the blockchain, Web3 is happening now. And lots of people,

[00:06:10] they're obviously riding this wave, and they're charging incredibly stupid amounts of money, just playing on the fact that no one

[00:06:18] knows, and everyone is in a hurry to just adopt AI. And so when I said that is the pattern they see most of the time is they don't

[00:06:24] know what to do. And they are in a hurry to just do something, whatever it is. And very often, it means that they just put the

[00:06:33] technology in front of the problem. That is the main, main thing I do every single time, you know, they say, Oh, we want to use AI for this,

[00:06:39] this, this and that. And I'm always telling them, let's, you know, let's swap the table, let's turn this table, because

[00:06:46] you're putting it in the wrong order. What are these processes that you want? Let's talk about these internal

[00:06:51] processes, let's see what you want to do. And then let's talk about if AI is a viable solution, maybe it's not, you know,

[00:07:00] happened to me, probably the same with you, or lots of times that people think that AI is there, and they can skip on, I don't know,

[00:07:07] developing, they can skip on coding stuff, they can skip on paying their developers, they can skip on paying

[00:07:12] their marketing. That is always happening, that happens all the time. So that is the main thing. And

[00:07:18] the other one, as I said before, it is helping them understanding what they need, because it is true

[00:07:24] that charge GPT, it is something everyone can use. But in my, in my personal opinion, in my personal case,

[00:07:31] no one is really using it as it could be used. You know, the custom GPTs, for example, people are always

[00:07:36] surprised, as I tell them, actually, if you spend some time, I can make for you a custom charge GPT that is done with

[00:07:43] your stuff. And actually, it might solve probably a good 50% of the things that you want to do. And especially fine being

[00:07:50] charge GPT or whatever it is, because you don't need that extra layer of security. Some other times they need the personal

[00:07:56] custom solutions. So that is the main pattern I see. Putting the problem, the technology in front of the problem, and then, you

[00:08:05] know, not really understanding what they need. And yeah, that is obviously, part of the problem is the media narrative. You

[00:08:14] know, we can see all of us, we see that we have the tech media, if you want, for the tech people that they're all into the

[00:08:23] details, the tech stuff that obviously, normal people don't understand. And then you have the general media, the UK here, but I see the

[00:08:30] same in the US, that, you know, is basically mostly fear is either fear mongering or trying to explain something and not

[00:08:37] being it not being either accurate or simple enough for people to get it. So basically, there's this huge gap in the

[00:08:44] middle of people scrambling around, knowing by feeling that they need to somehow understand what this is, and not knowing

[00:08:52] what to do. That is the whole reason that I've done the air summits. And yeah, and this is the whole reason why I

[00:08:59] usually do advise people and companies with AI, just because of all those reasons.

[00:09:04] Folks like you and I spend so much time, you know, trying to keep up with all of these advancements that

[00:09:11] it's going out and interacting with the people who have other things to do, right, who are trying to run their

[00:09:18] business, who are trying to, you know, meet, you know, deadlines and work on projects and things like

[00:09:23] that. They don't have time for all of this. And sometimes we, we have to maintain that balance so

[00:09:28] that we understand how to reframe, to your point, what this all means, because a lot of it is sort of

[00:09:35] noise and a lot of it is way too much in the technical, you know, weeds about, you know, sort of point

[00:09:42] release and every advancement in these, in these models and some of the technical details. It's,

[00:09:47] it's ridiculous. I mean, I'm not as technical as you are, but I mean, just seeing some of this,

[00:09:54] I mean, I think if we were to fast forward to, you know, a year or two from now, we'd look back and say,

[00:09:59] why was that even like newsworthy? But I love your, your perspective in those observations make

[00:10:06] complete sense that everyone is in the US, we would say putting the, putting the cart before the horse.

[00:10:11] They see AI as this broad, you know, maybe set of technologies and these, this sort of silver

[00:10:17] bullet, however you want to think about it. And they're not focusing on where they need help

[00:10:23] with a particular business, you know, challenge. Right. And I'm wondering if you think some of this

[00:10:30] is because people didn't adopt the last sort of wave of technological, you know, advancement.

[00:10:38] Yeah, no, no, it is true. It is true. I think that obviously the fact that now it is so easily

[00:10:43] available because before there was the, you know, the easy chat interface of chat GPT, all of that was

[00:10:49] quite an accessible that is as usual, you know, that's what made it very accessible. And this is

[00:10:55] also a very good point of what you said about the perception of people, because I mean, in a way,

[00:11:00] it's not a new tech, I mean, not without in a way, it is not a new technology. You know, we're just

[00:11:06] throwing power and money at it, you know, to be more powerful processing, obviously optimized algorithms

[00:11:14] and, you know, more data, but we are, you know, forcing it through it. So basically, you know, very often,

[00:11:22] it's very interesting because where question always get asked, oh, they always tell me, oh,

[00:11:26] you know, it's going so fast. What's going to happen in, I don't know, just in a few months or

[00:11:32] next year or whatever. And I always tell everyone, I said, actually, yeah, there was an advancement, but

[00:11:38] we're still talking about the same technology. And I always tell them until someone comes up with

[00:11:45] either the hardware or the software next idea for generative AI, we're not going to see that,

[00:11:54] you know, you are going to see a diminished returns. And, you know, this is what's happening even now,

[00:11:58] you know, before, you know, within, between a GPT-4, GPT-40, and then there was the GPT-4 or

[00:12:05] whatever preview that in theory is reasoning, and then in practice, it is not very good anyway,

[00:12:11] you know, basically the perception, oh, now it's reasoning. And obviously, the thing is,

[00:12:16] if you don't know in a very simple way how it works, you know, you're very,

[00:12:21] it's very well understandable that you think, oh, now I can reason, which obviously is not true.

[00:12:27] You know, it's not, it's not what's happening at all. So yeah, that perception, I think it is going

[00:12:32] fast. It is going fast, that's for sure. But it's not really going to be anything major until we will

[00:12:39] see the next shift in terms of hardware or, or software. So I always tell people, probably,

[00:12:45] you know, the hype phase, very steep. We are now more in a plateau, and then we will see what's

[00:12:51] going to happen, you know, depending on, you know, you know, if you keep just, you know,

[00:12:55] we can't keep throwing energy and money, just, just the same thing, you know, there would be GPT-5,

[00:13:00] six, seven, eight, whatever. I'm sure it will be better. But the difference between GPT-3 and

[00:13:07] four, and then the four and five, there will be always less and less. So I always try to keep

[00:13:13] people also to try to give them, get to give them a bit of a perspective over reality check,

[00:13:18] if you want. Because it's everywhere, and it's not always newsworthy. But that confirms the fact

[00:13:25] that even the general media, they don't know what they're talking about. That also doesn't really

[00:13:29] help a good constructive dialogue about it. Because, you know, AI, like the internet,

[00:13:34] like mobile phones, is going to be a disruptive technology. It is. So it's going to be disruptive

[00:13:38] with some jobs. It's going to create many other jobs. But the thing is, a responsibility is on us

[00:13:45] as a society in terms of what we do with it. And until we promote a constructive talk about it,

[00:13:52] we are not going to get out of it. We can't, you know, my main fear is when someone tells me,

[00:13:57] oh, I'm scared of losing my job. I'm scared of fake news, whatever. I'm telling them I'm not scared

[00:14:01] of any of that. I'm scared of us leaving the responsibility of a technology that works fairly,

[00:14:09] leaving this responsibility either to the tech giants or to the governments. And I believe that

[00:14:16] they're not going to do it because, you know, they have, you know, they're not the ones that are going

[00:14:20] to change things. I don't believe in a top-down approach. I think that education and us as a society,

[00:14:27] it is us who are going to drive how this is adopted and how this is going to be planned out.

[00:14:34] But that means that we need education, we need a lot of staff, you know, we need to solve the

[00:14:38] challenges that we have in front of us. I definitely agree with you. I spent a lot of time in the

[00:14:43] responsible AI space and I use that term just for the audience's sake as the sort of umbrella term over

[00:14:51] governance and ethics and fairness and transparency, explainability. And so there's a lot of facets to

[00:14:59] this. And, you know, in some ways, the small business probably has less risk exposure, I suppose,

[00:15:08] but it doesn't mean that, you know, something that goes awry won't have a sort of outsized, you know,

[00:15:14] impact to the way that they, you know, operate, right? Like if you put everything into like a,

[00:15:20] I don't know, some agentic AI, you know, workflow, and all of a sudden you went from a, you know,

[00:15:25] mom and pop, you know, old school paper, pencil, cash register thing. And all of a sudden now

[00:15:31] you've automated perhaps too much and something goes wrong. And now all your, you know, orders are

[00:15:39] screwed up and all that. Maybe they all shipped to the wrong place or whatever it is. Like, that's a

[00:15:45] huge, you know, problem. But I do agree with you that regulation, you know, relying on the government

[00:15:53] is, while it's important for them to have those conversations and to push forward, they won't be

[00:15:59] able to keep up. And we've certainly in the US here, we've seen that quite clearly that they can't keep

[00:16:05] up or they have a knee jerk reaction to certain things. I mean, even the California legislation

[00:16:11] that was vetoed by Governor Newsom was really putting too much onus on, to your point, the big

[00:16:18] technology vendors, the ones that can afford, you know, the compute costs and the energy costs to build

[00:16:25] these models in the first place. You've got to really look at the use case. I mean, I do like

[00:16:30] what the EU did in terms of looking at sort of risk pyramid and you've really got to look at the use

[00:16:36] case and how a technology is used, right? I mean, a car is, you know, a utility. It's, you know,

[00:16:43] a lot of things, but it could also be used as a weapon, right? It's also a dangerous thing. So you've

[00:16:48] got to put regulations around, you know, how a particular technology is used.

[00:16:55] It's very interesting to talk about the car because during my round tables or public speaking,

[00:17:01] the car is as an example. When I see that everyone is very concerned about safety and stuff,

[00:17:06] the car is actually an example I always tell them. I always tell everyone, I said, look,

[00:17:10] as an example, I always tell them when you go and buy a car, the retailer, it's not going to spend

[00:17:16] like five hours with you telling you, oh, be careful with the car. You can kill someone.

[00:17:23] Or even more, the car manufacturer, they're not, we don't have cars that can go only five miles per

[00:17:29] hour and are surrounded by airbags everywhere around it so that we are sure they don't go anything

[00:17:34] wrong. It is us as a society that we are educated to it and we are taking the responsibility of using a

[00:17:41] tool properly. You know, that is the main thing. Obviously cars have done all the security reasons,

[00:17:46] and all the things that are needed to enhance that. But I was thinking, why with AI? But it's not the

[00:17:54] only AI in the case, obviously. Why with AI is all about the technology and stuff. We are not actually

[00:18:00] taking any responsibility as a society. Or, you know, companies as companies or, you know, governments

[00:18:07] as the ones who pay for education. Because as you were saying before, one of the main problems is the

[00:18:14] lack of education. It is. I mean, schools, they have absolutely no idea. They are still,

[00:18:21] you know, they're still considering AI as cheating instead of teaching the students how to use it.

[00:18:26] Jobs, same thing. So, I mean, we're seeing it everywhere. I think, you know, that is the main

[00:18:29] problem. And I think that as long as we think that this can be solved by someone else and not us,

[00:18:35] then I think it's not going to change.

[00:18:37] This is why I've been pushing on organizations and individuals to really understand, just get,

[00:18:42] start to use it, start to see what it can do, test its limitations, see where it's hallucinating,

[00:18:51] as they say. Ask it a question that you already know the answer to and see what it says and see

[00:18:56] if you really think it's giving you 100%.

[00:19:00] You know what you should know? You should know the You Should Know podcast. That's what you should know.

[00:19:08] Because then you'd be in the know on all things that are timely and topical. Subscribe to the You

[00:19:13] Should Know podcast. Thanks.

[00:19:16] You know, accurate response. Because it happens to me all the time. I will, just in the course of my work,

[00:19:23] maybe I'm researching a particular company or technology. I mean, I've looked up companies that

[00:19:29] I know exactly what they do. And then I ask it, and then I have to, you know, nudge it a little bit,

[00:19:35] like, no, it's this type of company. And then it'll just completely make up a story

[00:19:40] about what this company does based on my initial and secondary prompts. I'm just trying to give you

[00:19:46] a nudge and give you a little bit of context so that you can go and find the right information.

[00:19:52] But I know that what you just returned to me is wrong. And it'll just leave it out there

[00:19:58] as if it was, you know, fact. And then until I call it out and challenge it, it'll just leave it that

[00:20:06] way. But if I challenge it, it's like, Oh, you know, I shouldn't have said that. Why did you say it?

[00:20:11] You know, this is, it's okay to not fully understand how everything works behind it. The

[00:20:17] point is to apply your own critical thinking. It's not a calculator, right? And so apply your own

[00:20:23] critical thinking and understand how you can use it as an aid to start to get you to the right place.

[00:20:31] But, you know, as the models evolve, I mean, hopefully hallucinations decrease. And if you build

[00:20:36] to your point before about the custom models, and as you sort of train it with your proprietary data,

[00:20:43] or the way that you write, if you're using it as a, as a writing assistant and writing,

[00:20:47] you know, co pilot or something like that, it'll start to get better and better, but you can't

[00:20:53] completely rely on it, which is where I think people get into trouble. Yeah, yeah. I mean,

[00:20:58] you use the word critical thinking. This is always something else I use a lot when I speak to parents,

[00:21:03] or I've done some tutoring for college here. And, you know, they're obviously concerned about,

[00:21:10] you know, what jobs will be available for my kids. And I always tell everyone, I said, look,

[00:21:13] I mean, knowledge, it is something that AI can help with. But that means that we humans are left with

[00:21:19] what we are best and what we should be best at. That is not remembering stuff. What you're best at

[00:21:25] is critical thinking. I always say critical thinking. I mean, we raised the generation,

[00:21:29] including my one in the last 40 years, probably more that everyone needs to be an engineer or a

[00:21:35] doctor or a lawyer or whatever. So we don't need any more thinkers. We don't need any more creatives.

[00:21:42] We don't need any more, I don't know, philosophers, whatever. And always telling people, actually,

[00:21:47] what they, I will likely do, it is that all this, I think we call humanistic thing,

[00:21:52] all this humanistic sides of you, of us, they are the things that we need. We need more people

[00:21:57] able to critical think. We need more creatives. We will need a lot of the soft skills, communication

[00:22:02] between humans, empathy, you know, all these skills, the ones that before they were considered

[00:22:09] secondary, because you mean to, you know, if you're an engineer, you're better. Actually,

[00:22:13] you know, this is finally being reverted. And, you know, we're going back into using our brain

[00:22:19] in our brain. You know, for me, it's the same when I see someone spending a whole day

[00:22:24] scanning items as a cashier in the supermarket, or I see someone spending their whole day on an Excel

[00:22:31] file doing this stuff. I'm really feeling pain because I'm really thinking you as a human being,

[00:22:38] I wouldn't say you're being not used, or I think you are honestly wasting your time into something

[00:22:44] that you don't like. And probably you have so much better that you can give. I don't know,

[00:22:50] the cashier in the supermarket, they are excellent people, people. They are lovely, you know, they talk

[00:22:54] to people. Maybe we can use them in something else. Maybe they could do, instead of scanning items on a

[00:23:00] bar with barcodes, they could just do something like customer support, or maybe not even dealing

[00:23:05] with customers. Maybe they can be therapists, whatever. But you know what I mean? I think that all

[00:23:10] of us, we have such a big potential once we start exploring it and once we have the time for it.

[00:23:15] And that is the opportunity, I think that there is with AI. The opportunity is that it might free us

[00:23:21] from all this repetitive and boring and brain-killing jobs, and actually everyone else

[00:23:31] having the time for the rest. But there is a big, big problem in there that it will work only if we

[00:23:37] embrace it as a society so we protect the weaker people. We make sure that people that are in the

[00:23:41] jobs that will be replaced are upskilled, and that is a responsibility of the businesses.

[00:23:47] And then we need to be sure that governments jump on with education. That is the key factor.

[00:23:54] And they are very different, they are very difficult challenges. You know, asking governments to spend

[00:23:58] more on education instead of cutting, and asking businesses to spend more on upskilling instead of

[00:24:03] firing, they're very difficult things to do. But we need to face it, it's the elephant in the room

[00:24:08] that we need to face it. If we do that, then AI might literally allow most of us to do what we like

[00:24:14] in terms of jobs, and to elevate all of us. I really believe in that.

[00:24:18] Yeah, I do too. There's so much to unpack there. I guess the first part I'll unpack is around education.

[00:24:26] I absolutely agree. But I don't understand why people don't see that starting to learn how to

[00:24:34] use AI and use AI properly is similar to, you know, having your first computer or, you know, all these

[00:24:41] Gen Z, to some degree, millennials, and now Gen Alpha, I think are calling it, but they were given

[00:24:50] digital devices before they could talk, right? So, I mean, they can handle this easily. They're already

[00:24:58] using it in various forms with the apps that they use on their smartphone, which is basically a

[00:25:05] computer. And so there's AI, there's all kinds of AI on those devices already. So why would you not

[00:25:13] teach them how to use it just like they might take computer science in high school now? But I mean,

[00:25:20] you've got to understand where things are going and how you're positioning this future workforce

[00:25:25] to be successful for the jobs that may not exist right now. So the other thing I wanted to just hit

[00:25:32] on was the skills piece, which to your point, it's those human, what we used to call soft skills.

[00:25:41] Yeah, I agree with you. It sort of switched. They're not soft at all. Those are the powerful

[00:25:45] skills. Those are the real human skills that are the hardest for AI to take on, including the reasoning

[00:25:53] that you mentioned before. But those are the durable skills. Those are the ones that are going to last,

[00:26:00] and those are the ones that are important to complement whatever AI can do today and whatever

[00:26:06] AI can do tomorrow. We will always need the critical thinking, the empathy, the creativity,

[00:26:11] ability, and the ability to take in all this context and all this other information to make decisions.

[00:26:21] That's why, especially in the talent space, I get concerned when people are overly reliant on the output of any AI,

[00:26:30] whether that's in text form or it's a recommendation or it's an actual score or stack rank of a person.

[00:26:38] Behind that application is a person, there's a human being, and we need to treat that with the right

[00:26:45] sensitivity in terms of fairness and potential adverse impact and things like that.

[00:26:52] And so there's just so much that we can do to position ourselves properly for this. And if we don't

[00:26:58] get ahead of it and push legislators to do the right things in the interest of us as human beings,

[00:27:07] this doesn't evolve in a positive direction.

[00:27:11] No, no, I totally agree. As I said, it's a tricky one. It's a difficult one. But what I see is that

[00:27:18] every single time that I'm asked to speak at an event is something, usually it is always something

[00:27:24] like explaining to people how AI works in very simple terms and stuff, and then explaining them

[00:27:29] a bit of the tools and some tips and tricks on how to use the tools that are available there.

[00:27:34] The thing is that once the end opens the room to questions, you can be sure that you always end up

[00:27:41] talking about these topics. So that is always something that actually comforts me, I think,

[00:27:47] because I can see that people, even before understanding, all right, okay, how can I use

[00:27:52] HGPT, whatever it is for, I don't know, automating my emails, whatever it is, the technical detail,

[00:27:57] they all want to talk first about what we're talking about it now. And that gives me hope that

[00:28:04] actually, society knows, we as a society, we know that we need to go deep into that as much as we like

[00:28:14] to go deep into the technology side of things. So I'm always hoping that as more people like me,

[00:28:19] like yourself, get out there and try to help people with education, basically trying to have people

[00:28:26] thinking about things. You know, all these summits that I do, people always tell me, oh, I arrived here

[00:28:31] with lots of questions. And I got all my questions answered, but actually now I have more questions,

[00:28:36] even more, and they are more deep. And most of the times they are about all these topics we're talking

[00:28:41] about, education, how it affects us. And I think that, you know, it's a fertile ground at the moment

[00:28:50] with society. And I think that it is responsibility of, as I said, people like yourself, like me,

[00:28:56] like many others trying to do this, to basically try and help everyone starting this sort of

[00:29:02] constructive discussions. Because I think this is what will allow us to understand, you know,

[00:29:08] what is hallucination? What is not? You know, how AI can actually, because it does enhance our

[00:29:13] productivity, can make our life easier. And that doesn't mean that it's solving all the problems,

[00:29:19] and doesn't mean that, you know, it is stealing jobs or, you know, it is going to disrupt, as I said

[00:29:24] before, yes. But once again, you know, we have cars now, it's not that when, when the first car

[00:29:29] arrived, all the people that were, I don't know, managing horses, that just stopped things to happen.

[00:29:35] It's normal, it always happened.

[00:29:37] I do think, you know, whether it's, you know, the actual education system, or it's,

[00:29:41] you know, upskilling that you were mentioning before, I do think it's important for people to,

[00:29:46] you know, be given examples that can really hit home. And maybe even just some basic analogies,

[00:29:52] like we were talking about the car, right? So here in New York, at 16, you get your driver's permit,

[00:29:58] which you have to take a test for. And then you have, at 16 and a half, you get your actual license.

[00:30:04] That means you don't have to have another person in the car. But in that six months, you've got to

[00:30:09] practice, you've got to take driver's education, you've got to take an actual, you know, literal

[00:30:16] exam, like with someone from the Department of Motor Vehicles, you know, takes you out,

[00:30:19] they make sure you actually know what you're doing or whatever. So point is, with AI, we haven't done

[00:30:25] that, right? Like, open AI basically just announced ChatGPT and said, here you go, have fun. And it's just

[00:30:30] like you just handed the keys to essentially like a 14 or 15 year old that maybe had no business

[00:30:38] getting behind the wheel. And you just said, you know, go nuts, right? Enjoy, do whatever you want.

[00:30:44] Wow, right? Like, you know, from a technology standpoint, I was just thinking back to my time

[00:30:50] at NBC Universal, when we were moving to cloud off of, you know, data centers and legacy infrastructure,

[00:30:57] you didn't just give all of those database administrators and all of those owners of servers

[00:31:03] or whatever, you didn't just say, Oh, forget that, you know, just go to AWS, go to Azure,

[00:31:08] and just log in and, you know, have fun, right? You had to go through a certain set of steps to

[00:31:15] understand what it means. How do I, you know, spin up a new, you know, instance? How do I tap into,

[00:31:21] you know, S3 and get more story? You had to go through these steps, not to mention

[00:31:26] all the steps around, you know, privacy and cybersecurity and things like that, right?

[00:31:32] You don't want to just basically inadvertently, you know, leave the back door open,

[00:31:36] right? And cause all kinds of security incidents and, you know, maybe legal risk as well. So I just

[00:31:42] feel like, you know, AI, now that it's in everyone's hands, right? I think the big difference now with

[00:31:47] generative AI is everyone has, you know, the AI is, is the UI, right? The user interface,

[00:31:53] you're interacting directly with it. Whereas, you know, predictive AI in the past and analytics

[00:31:58] or whatever, like not everybody had necessarily direct access to some of those technologies. So,

[00:32:04] so we've got to do some, some of those practical steps, but I wonder in some of the training that

[00:32:11] you've done in the workshops that, that you do, are there particular sort of exercises or,

[00:32:17] you know, scenarios that you describe that resonate, you know, better that help people say,

[00:32:22] oh, now I start to see why I can't just rely on it like a calculator.

[00:32:27] So I think that when it comes to tools, it can be, A, it can be very, very, very overwhelming.

[00:32:34] And B, if someone needs to understand the basics, I wouldn't really go into, oh, this, you know,

[00:32:42] you can use this tool for this tool for this, this tool for doing that, because I think that is too much.

[00:32:47] And these tools, you know, most of them are new, some of them that will close. So I'm always trying

[00:32:52] to go back to basics, basically, you know, is it for me, it's an analogy if, I don't know,

[00:32:57] back in the days when I was learning how to, you know, I was 14, 15, when I was learning how to

[00:33:01] program, I wasn't learning, I don't know, I didn't start learning visual basic, I was learning something

[00:33:08] that wasn't visual, I was learning the basics, the real basics to understand what's happened,

[00:33:13] what happens. So for me, what I do, I first of all, in the room, I ask everyone to think about,

[00:33:19] forget about what you know about the AI, I always tell them, what is one internal process,

[00:33:24] very important, because they all want their support chatbots and stuff and that. And just so

[00:33:29] and I just tell all of them, just forget about it, because you have absolutely no idea how complicated

[00:33:34] and risky that thing is. So just put this thing aside. They told you, you can have your chatbot and,

[00:33:39] you know, fire all your support people. And that is not true. So forget that.

[00:33:43] Let's go to internal processes, because they're easier and are less risky. So I always tell them,

[00:33:48] what is an in your group, in your team, in your company, depending on how big,

[00:33:52] what is the process, possibly repetitive, most likely is going to be repetitive,

[00:33:58] that you don't know how, but you have a hunch that there must be a way to automate it. There must be a

[00:34:05] way because it's so simple. And you can be sure that usually a good 80-90% of the times,

[00:34:10] they come up with something that you can immediately think, ah, yeah, obviously,

[00:34:14] ChargeGPT can help with that. And I'm going back to ChargeGPT because I mean, I've used a lot of,

[00:34:19] you know, OpenAI, Cloud, Gemini, Copilot. I mean, there is the same thing. But anyway,

[00:34:28] in my experience, I mean, OpenAI, they are quite far ahead. You know, Cloud sometimes is better in some

[00:34:34] things. But in my experience, in OpenAI, they are much, much ahead of everyone else in terms of,

[00:34:40] you know, in terms of text generation. So I always encourage everyone to literally stick with one

[00:34:47] tool. And I always tell them, if I was a business, I don't have time to try all the tools, I would tell

[00:34:51] them just stick with ChargeGPT. We're talking, it's the same difference between Google and Bing.

[00:34:57] Bing, it is used, it is good. It's not as good as Google. So I would just say, for now, because we

[00:35:03] don't have time to just go into the single nuances of the difference, just ChargeGPT. And

[00:35:08] I think, to be honest, I think that more than half of the times, I can have a business living with a

[00:35:13] working solution that automates some in some of their internal processes just using custom GPTs.

[00:35:19] I think custom GPTs, they are the most powerful thing that you can imagine, you know, the thing

[00:35:25] that you can upload documents. So you can do a RAG literally without knowing a single line of code

[00:35:31] is very good at setting its own prompt, because it will ask you, what is this bot about? It's quite

[00:35:37] good with that. And, you know, you can implement tools, actions, APIs. So, you know, if someone

[00:35:43] needs something quite specific, you can basically tell them, you can use ChargeGPT, make it your own,

[00:35:51] put in your own documents, explaining to them, obviously, what it means in terms of privacy,

[00:35:55] disabling the training of the model with the data, but, you know, all the basics. Most of the times,

[00:36:03] I mean, you know, I've done two incredible apps just literally using the custom GPT. And people are

[00:36:10] always saying, wow, they just don't understand. They didn't know that. They think that they need to hire

[00:36:15] a developer and always tell them, no, this is why it is good. This is why it is interesting. You need your

[00:36:22] left side of the brain. You need your language to make it work. You don't need code. You need to

[00:36:28] understand how to speak to someone else in a weird way, because, you know, prompt engineering and all

[00:36:32] that sort of stuff. And they're always incredibly surprised. And funnily enough, when they ask me,

[00:36:38] what is, you know, when I tell them that I didn't study, I'm not an engineer, you know,

[00:36:42] I studied philosophy, you know, I come from, you know, I studied Latin, Greek history and philosophy.

[00:36:47] And they always ask me, how do you end up doing this? I always tell them,

[00:36:49] I ended up doing this. And the reason I love AI, it is because it is the ideal bridge between being

[00:36:55] humanistic and then being technical. And I always tell them you can do it yourself.

[00:37:00] You just need to iterate. So I personally, I believe that custom GPTs are by far

[00:37:05] the most powerful tool that you can imagine out there. And once you have a developer that maybe

[00:37:10] can do an API for some external tool, I don't know, to have it interacting with your

[00:37:16] database, or you are a store and you want it to interact with your shipping. So whatever it is,

[00:37:21] it is incredibly, incredibly, incredibly powerful. So that is usually, that is the only thing I do.

[00:37:27] And the funny thing is that they all can do it for $20 a month. I tell them, you can have as many

[00:37:33] as you want. I think I have probably 50 or 60 custom GPTs just for me. It's incredible what they can do.

[00:37:40] Well, we'll have to, we'll have to have a separate conversation where I could take a look at some of

[00:37:45] what you've built. I've built a few myself, but not for myself. I built them when I was just playing

[00:37:52] around when that ability, you know, first came out and I created one around responsible AI. I created one

[00:37:59] to recommend whiskey, but I haven't done, as you described, some of the custom ones that I know

[00:38:08] can probably help me. And I think you made a point in there that I think.

[00:38:12] Hi there, I'm Peter Zollman. I'm a co-host of the Inside Job Boards and Recruitment Marketplaces

[00:38:19] podcast. And I'm Steven Rothberg, and I guess that makes me the other co-host. Every other week,

[00:38:23] we're joined by guests from the world's leading job sites. Together, we analyze news about general niche

[00:38:29] and aggregator job board and recruitment marketplaces sites. Make sure you sign up and subscribe today.

[00:38:37] This is where I got concerned myself, which was, can this actually go in and out of, you know, my

[00:38:44] my CRM or my Google docs or my whatever. And so, but to your point, just start asking it what it can do.

[00:38:52] It's not just the AI to help you, you know, through all these different tasks and to help you make

[00:38:58] decisions and research and things like that. It's also its own, you know, help section and user guide.

[00:39:05] And all of that, all at once. And so you've got to just, you know, spend the time to your point,

[00:39:13] iterating and pushing its limits and understanding what it can do.

[00:39:17] And it's available to everyone. That is the thing I tell everyone. It is literally out there. You don't

[00:39:21] need to hire anyone. I mean, you can start with it. Just do it. This is what I always tell people. Just

[00:39:28] start experimenting with it. I mean, there are people out there that literally charge thousands to

[00:39:33] deploy, to develop a custom GPD and they tell people, oh, you know, this is your model and stuff.

[00:39:39] And people are always impressed. And this is why I'm always very disruptive as an entrepreneur. So I

[00:39:45] literally go to businesses and say, just please ignore all these people, just ignore all of them

[00:39:50] and just do it yourself. Because half of the times you don't need anyone else. Then once you start

[00:39:57] needing, for example, they might need, oh, I don't know, I am a legal firm or I am an accountancy firm.

[00:40:03] So you have compliance, you have internal documents, you can't upload them on OpenAI. And then this is

[00:40:09] when I just come in with my, you know, with a, you know, one of my projects is called your vault,

[00:40:14] that it is literally like a private AI for businesses. And then I start telling them,

[00:40:19] you can do that. You can have the same power of charge GPT with your own model that runs on your

[00:40:26] own instance is only for you. And those data, they're not going out to anywhere. So, you know,

[00:40:31] it's completely a hundred percent private. And the problem is there are not many solutions that

[00:40:36] do that for businesses, for small businesses in particular, in simple terms. And, you know,

[00:40:41] this is why literally my MVP for your vault AI already got two people that wanted me to try it

[00:40:47] because the market is literally asking for such simple solutions. But it is always the numbers that,

[00:40:53] you know, I wouldn't tell anyone to go on my service, your vault, until they have done some

[00:40:56] time spending it on charge GPT and understanding how it works. It doesn't make any sense. It's just

[00:41:01] selling to the 16 years old that you felt that you told me, giving them a Ferrari.

[00:41:05] You wouldn't do that. You did that. You first need to learn to learn on your Ford Fiesta or whatever

[00:41:09] it is, the scrap car that goes slow and you can destroy and you can make lots of errors. Once you learn,

[00:41:16] then first of all, you will know if you really need a Ferrari, and then you will have a better

[00:41:20] ability to understand how to use it or what you need from your, for your, from your thing.

[00:41:25] I'm a big Ferrari fan. It's hard to come up with a scenario where someone needs

[00:41:29] a Ferrari, but I understand your point. One of the observations that I've noticed,

[00:41:35] I have spoken to a few folks like you who are doing, you know, training sessions. And one of the things

[00:41:43] that stood out to me was some of the exercises that, that they go through. Like one person has created

[00:41:48] like a card deck, right? Where people can say, okay, you're going to use this and it's really to help

[00:41:54] people understand the basics of prompting and as well as understanding some of the responsibility. I

[00:42:00] principles too, I will say, but you know, here's, here's your prompt, here's some context or whatever.

[00:42:05] And then you're working in small groups, right? So it's not just this, you know, fun exercise,

[00:42:11] but you're actually getting some different perspectives, right? Some cognitive diversity

[00:42:16] that is applied to some of these scenarios. And the response of the feedback that I saw was that

[00:42:23] people really were much more comfortable. It reduced some of the fear that you were talking about

[00:42:28] earlier and having drove some of that constructive dialogue that you were mentioning earlier that

[00:42:33] can help people get over some of that fear. The other exercise that was just actually in yesterday,

[00:42:39] there's an AI ethicist that I follow and she created this game. It's called the AI ethics game.

[00:42:45] And so I participated in her pilot group just yesterday and it was really interesting. She had

[00:42:51] created a custom GPT to basically play out this scenario where she had basically say, I've got this fictitious

[00:42:58] startup, this is what we're building. I've got four personas, we're going to ask these four people

[00:43:05] questions about how they've designed their solution. So the scenario was, this is a startup that built an AI

[00:43:10] powered learning tool for developing markets. So people that haven't necessarily been exposed to all of

[00:43:17] this, this technology, and won't necessarily apply, you know, critical thinking by default, because they'll just be so

[00:43:24] enamored with this fascinating new technology. But you got to ask questions of the CEO, the CTO,

[00:43:32] a junior data scientist, and the product manager. And everyone from all of these different, you know,

[00:43:38] the whole audience was basically coming from these different perspectives. There were at least 10

[00:43:42] different sort of roles that people were coming from across industries as well. So again, you had this

[00:43:48] cognitive diversity and people were basically asking probing questions. Well, how did you, where'd you get your

[00:43:53] data from? How are you accounting for, you know, personalization when, you know, you're supposed to

[00:43:59] be, you know, these are students, right? Like, so you need to keep that data private. And, you know,

[00:44:03] it was just a really interesting and engaging way to try to get people to think more deeply about not just

[00:44:09] the solutions that might be out there now, but how startups in this space who are building these solutions

[00:44:17] are thinking about it. And I think it goes back to one of the things we were talking about at the

[00:44:21] beginning of this conversation, which is, you know, this, these shiny objects and everyone just kind

[00:44:26] of latching on to this technology and then going to try to find a problem to solve it with. Like,

[00:44:33] this is my new, my new shiny hammer and let's go find some nails. I feel like it was an interesting

[00:44:40] example of where, again, maybe this is because I'm close to this and I talked to these startups and I

[00:44:46] see these startups in the space building these solutions and building more of these sort of shiny

[00:44:51] objects, right? They're, they're building solutions. They're building those hammers in the hopes that

[00:44:57] people find nails to, to use the hammer and buy the hammer. So it's interesting to see from the other

[00:45:03] side, like some of these startups that are building things that also don't use that mindset that said,

[00:45:09] what problem are we actually solving? What is the pain or creating a, you know, a vitamin for like,

[00:45:16] what, what are we doing? And so it was just interesting to see all these people who are not

[00:45:21] AI experts by any stretch, actually asking intelligent probing questions. Like, did you really put ethics

[00:45:29] first? Are you really being human centric or all you're worried about is, you know, you got VC money and now

[00:45:36] they want to see, they want to see you moving fast and breaking things, but that's not necessarily

[00:45:43] appropriate, right? You need to be thinking about the human factor. You need to think about the ethics

[00:45:48] involved. You need to be thinking about, you know, the, the fairness and that these are people who are

[00:45:54] not technologists, uh, that you're targeting and you need to take a step back, um, because there's

[00:46:00] like red flags all over the place. Yeah. I totally agree. I totally agree. As I said, you know,

[00:46:05] we do, we go back to what you were saying in terms of the responsibilities on us and education is going

[00:46:11] to be a key element in all of that, you know, to just tackle exactly to do what you're saying.

[00:46:16] Yeah. So I know you sort of advise people not to, you know, go out on product hunter. There's a bunch

[00:46:22] of sort of catalogs of all these AI solutions cropping up and, you know, a lot of them are half-baked or maybe

[00:46:29] quarter-baked at best. And it, you know, it can be both dangerous and time consuming to try to go and

[00:46:37] play with, with some of those. You don't know what it's doing with your data besides some of the main,

[00:46:42] you know, LLM vendors. Is there anything that you've seen aside from the things that you've built

[00:46:49] where you think there's promise, like there's something that could be extremely useful or, or on the

[00:46:56] other side, is there anything that you see that you think this can't do anything good? This is

[00:47:00] dangerous. I don't know. I mean, to be honest, I tried many tools and in the end, I always go back

[00:47:06] to using, to integrating the main players myself, but probably this is also because obviously as a tech

[00:47:13] person, you know, if I don't see something that I like, I build it. To be honest, I mean, apart from

[00:47:19] using tools like, I don't know, the Grammarly for, for, for things that is very useful. I mean,

[00:47:26] OpenAI, it is integrated everywhere in everything I do. Um, lots of good AI for coding, you know, MVP

[00:47:33] developing, it is incredibly useful. And, you know, there are, uh, you know, things like cursor that

[00:47:39] it is like a VS code integrated with AI, which is, can be incredibly powerful once you configure it.

[00:47:45] I imagine, uh, you've played around with some of the ones generating like multimedia content or,

[00:47:51] you know, music and stuff like that. I've used obviously, you know, mid journey for image generation,

[00:47:56] but you know, it's always the same advice that most of the images it generates contains,

[00:48:01] they are, they have problems, but it can be incredibly useful. So, you know, I definitely use mid

[00:48:05] journey in terms of videos. I mean, I, I didn't try Sora. I've seen some other of the video

[00:48:11] generation tools at the moment, but they still look very weird. I mean, they still look like a

[00:48:18] series, a series of images stitched together. So we're still not there, but Sora is very promising,

[00:48:23] but obviously we have seen only the demos and I want to see, I want to be able to put my hands on it.

[00:48:28] So in terms of generation, possibly not. I mean, tools like there are many tools that generate music.

[00:48:34] They are at best elevator music at best. So I'm not really bothered by it because,

[00:48:41] I mean, there are reasons behind it, obviously, you know, to, to, to, to make music, you can't

[00:48:45] expect an AI to be trained from a mixed music as in from a stereo file that contains all the

[00:48:52] instruments of a song, because there is no way for it to understand the single parts apart from the

[00:48:58] very simple cases. So you usually just get more obvious than with text. I think mishmash of

[00:49:05] something existing that sounds horrible. And it is a copy of something else.

[00:49:09] I mean, what I see very well AI used it is when it is integrated into the production process. So for

[00:49:15] example, I don't know Adobe Premiere, I do a lot of video editing. They have now Firefly that is

[00:49:21] literally an AI integrated into editing. Same thing with Photoshop, very, very useful, very powerful.

[00:49:27] So you don't use it for necessarily generate stuff, but you use it to edit, to correct things.

[00:49:35] Same thing in music, you know, there are lots of plugins for, you know, in the box,

[00:49:40] mixing and mastering, they all use AI. So I don't know things like polishing a voice vocal take,

[00:49:47] because I don't know, I was recording it at home, and there is an ambulance passing by doing my best

[00:49:52] take. Before it was going to be quite tricky. Now there are many plugins that they use neural networks,

[00:49:57] and they know how a voice sounds. So if there is a dog barking or an ambulance passing,

[00:50:02] they can literally filter it out intelligently. So that is where I can see the full power of it.

[00:50:09] But in terms of multimedia, I'm not thinking, I don't think I see anything really useful in terms

[00:50:14] of generation yet. That's really interesting. I hadn't thought about that

[00:50:18] angle, but it does tie to some of the other themes that we see, like AI should be augmenting,

[00:50:24] you know, human abilities. And to your point, a better use of AI is to clean up

[00:50:32] the audio and make you, the musician, the singer, etc. or the podcaster sound as good as possible,

[00:50:43] but not take that job away from that person.

[00:50:47] Yeah.

[00:50:47] Well, I mean, if you think about it, it's the same with text. You know, I always tell people

[00:50:51] never use an image from Chargipit, from Dali or from Midjourney as it is without literally

[00:50:57] tracking it. And even more, I always tell them never use a text as it is out of Chargipit or whatever.

[00:51:03] Just consider it a draft at best of the times. And most of the times, it's just going to be an

[00:51:10] inspiration for them. You know, that is where AI is useful as a resounding board, as a tool for

[00:51:17] firing up ideas and then for drafting. That is what it is. So that is where I can see, I see it being

[00:51:22] very useful.

[00:51:23] So any words of advice for people trying to get started who have not attended one of your workshops

[00:51:31] and just know that they need to carve out the time, but haven't yet?

[00:51:35] Yeah. I mean, as I said before, I think the best advice I can give them, it is

[00:51:40] personally speaking, just go on Chargipit, create an account, it's $20, pay the $20 because it is what

[00:51:45] you need. And then just start experimenting. There are, if you go on Google and you Google OpenAI

[00:51:51] custom Chargipit, you will see a million and a half tutorials where you don't need to be a developer.

[00:51:57] It's not a tech thing. It's just a tool to be used and just go and experiment with that. Because I

[00:52:03] think this is where you would see really the potential and also the pitfalls of, of, of AI.

[00:52:10] And I think you will have a much better idea of how it can be used for your business.

[00:52:14] Awesome. All right. Great advice. Giancarlo, thank you so much for, uh, for joining me. I think, uh,

[00:52:20] this was a fascinating conversation. I think, uh, listeners have a lot of

[00:52:24] takeaways and, uh, a lot of things that they can go and play with themselves. I'm going to

[00:52:29] make sure we include, uh, some links to some of your projects and your, your startups in the show

[00:52:35] notes so people can, can access those and know how to reach you. So thank you again.

[00:52:40] You know, I'm visible, obviously on LinkedIn and stuff. I'm always happy to have a chat with people.

[00:52:43] Perfect. Perfect. Thank you very much for having me.

[00:52:46] Absolutely. Absolutely. Well, thank you again, Giancarlo. And, uh, thank you everyone for listening.

[00:52:51] Uh, we'll see you next time.