Latest / KAJ Business & Money LIVE / The Promise of AI with Ian Harris | KAJ Masterclass
Transcript
- 0:23Welcome to this very special edition of the Cage a masterclass live the show which
- 0:28ensures that you profit from your time spent here with experts, either through the
- 0:33industry in sights, information or simply learning from them. And today we have
- 0:39Ian Harris. He is a technology expert and founder of Pulse podcast. Welcome
- 0:45to the show. Ian, Hi a Jay, Welcome, Welcome to the
- 0:49Showbody, It's great to be here. Thanks for having me on today.
- 0:52You are welcome to the show. Ian. You are welcome to India in
- 0:55this online forum, and I'm sure not just in India, but a lot
- 0:59of people as the globe will benefit from what we are going to talk about.
- 1:03We'll be talking about the promise of artificial intelligence and will get to understand
- 1:07all that is good, not so good, and all the bad things also
- 1:12that it can do to humanity in the future that people are fearful or we'll
- 1:17talk about that. But first to understand from you Ian exactly. I've told
- 1:22you you are a technology expert, the founder of False Podcasts, but you
- 1:25have done a lot of work with large companies in their product and technology strategy.
- 1:30So I want you to tell them a bit about your journey about the
- 1:34kind of work you have done so that they know exactly who Ian Harris is,
- 1:38what expertise he holds in technology, Oh for sure. So I've been
- 1:42building platforms, technology platforms for many years and helping companies devise their product and
- 1:51technology strategy, so taking looking at where a company is at the moment,
- 1:56trying to work out where they want to get to, and then using technology
- 2:00to help deliver those products into the marketplace. And I've done that based in
- 2:04the US, in Europe and now in Australia where I'm currently sitting and lovely
- 2:08to talk to everyone in India and other plants of the planet as well.
- 2:12So great to be here right right now, Let's talk about the promise of
- 2:19technology and what technology is doing, artificial intelligence is doing at the moment.
- 2:24How do you look at artificial intelligence? People are looking at artificial intelligence in
- 2:31their own way. As a technology person, you will obviously look at it
- 2:37from a different angle. But in terms of people, what are the things
- 2:42that are good? In two aspects. One is from the business aspect,
- 2:47the work people do either workwise. Second is the life part of it,
- 2:53so the good not the bad, and the ugly part that can happen which
- 3:00all these people, even the godfather of Internet artificial intelligence from an even to
- 3:07you know, the open Aico, he also has been talking about that it
- 3:12can really go bad. So help us understand what we should understand of all
- 3:16the voices that are coming up from Chat Deputy, we have moved on to
- 3:21gepty forte, and people are waiting for so many other versions to come in.
- 3:25So we want your expertise on this. Help us give us a peek
- 3:29into a bit of a future of artificial intelligence in our lives and business.
- 3:35Okay, so we've been we've been using artificial intelligence for many years, and
- 3:40back even ten years ago, we were looking at how can we train computers
- 3:45to learn things? How can we train them to understand what's going on around
- 3:49us? And a good way of understanding what artificial intelligence has been like in
- 3:53the past has been we can feed it. We can feed it a number
- 3:58of images that have, for example, a number of the number zero to
- 4:00nine, and we feed it thousands and thousands of someone having drawn the latter
- 4:05zero, the number zero, someone having drawn the number one, and then
- 4:09after having done that thousands of times, we can present it with another image
- 4:14of a number and they use that training to decide what number it is that
- 4:18it thinks it's been given. Now, what's happening really when we do that
- 4:24is we're actually feeding the AI engine a series of numbers that represent the pixels
- 4:28of the image, and then we're saying this represents the number seven, for
- 4:32example, and then using that to help to train the artificial intelligence to understand
- 4:39what a number is. It doesn't really understand what a number is. All
- 4:42it really understands is that these series of numbers that represent the image. It's
- 4:46told that this is the number seven, for example, and then if one
- 4:48forgive it another image, it can look at that image you go, well,
- 4:50you know what, it's pretty close to what I think of is the
- 4:54number seven as far as my training has been concerned. And so we have
- 4:58a set of data that's the training set, and we have a set of
- 5:00data that's the test set, and based on that test the results, we
- 5:04can decide that, yeah, it's doing pretty well at understanding what is the
- 5:08number seven when it sees an image. Now that's a pretty simple example.
- 5:11And back back in many years even we were looking at what we call optical
- 5:15character recognition, So being able to understand the text of a page, for
- 5:18example, by presenting it to a computer that all seemed fairly only inert,
- 5:26really didn't really seem to be much of a threat. But now what's happened
- 5:30is that we're actually able to start using our training data that's much bigger than
- 5:33just an image or a number or a letter. What we're doing now is
- 5:38actually using training data of the entire internet, so using all the publicly available
- 5:43data, everything that's on a Wikipedia for example, all the publicly available websites
- 5:47and feeding is texted and saying this is a bunch of texts that makes sense.
- 5:51Now, based on this information, I want you to create some text
- 5:54based on what you already know. And this is what we're calling a large
- 5:58language model. So using a model based on a large amount of data.
- 6:01It's based on language, so based on text itself, so not based on
- 6:05images, but based on the actual text, and indeed not just English,
- 6:11but lots of different languages as well. And so based on that information,
- 6:15now the large language model can start doing things like okay, let's create a
- 6:17story or create something from scratch, or based on this amount of information,
- 6:20this text, what is the answer based on what you understand. And this
- 6:25is why I think it's now gone from being something that's basically useful and inert
- 6:30to being something that people are considering, Oh, hang on a tick,
- 6:33Does this affect what I do as a job? Is this going to affect
- 6:38what I'm actually doing on a day to day basis? And I think a
- 6:41good example of that is when we have large language models now that understand about
- 6:46programming, they understand about code. Okay, So now we can say to
- 6:49a large language model, based on all the code that you've seen on for
- 6:53example, on GitHub, all the publicly available code, everything you've seen on
- 6:57stack exchange, now tell me give me a snippet of code in say typescript
- 7:02that can download an RSS feed of a podcast and download the episodes and now
- 7:09understand the text that's in it, and to give me a summary of that
- 7:12particular text itself. Okay, that sounds that something a programmer will take I
- 7:16don't know, maybe fifteen twenty minutes, half an hour, maybe an hour
- 7:19to down to do. And now the artificial intelligence can spit out something that
- 7:24can do that in a few seconds, maybe a minute. It can give
- 7:28us the answer to that sort of problem. So in some ways. It's
- 7:31sure it's going to change the way we do our work. For a programmer,
- 7:34instead of spending an hour doing something, we can now ask a tool
- 7:38to do it in a few seconds, in a few minutes. Maybe it
- 7:42still needs fixing, but then most code that comes from humans is the first
- 7:46draft and it needs updating as well. But now we can do something that
- 7:49is it's a bit a good way of thinking about it. It's a really
- 7:54good force multiplier. So instead of just having me doing programming, for example,
- 8:00I've got me based and the power of all the code that's ever been
- 8:03written, it's publicly available. I can now use that to accelerate my development.
- 8:07So instead of programming taking hours or days, I can reduce that amount
- 8:13of time, at least for small parts of the actual code itself, down
- 8:16to minutes. And so now I can get things done a lot faster than
- 8:18I could do previously. And so it is fundamentally going to change the way
- 8:22that we work. Right right until this point, it is good, But
- 8:28I want to understand now when people are saying it can go beyond that.
- 8:33So I want to understand what exactly is the purpose of artificial intelligence. Everything
- 8:39is made for some purpose and till when the purpose is to teach us.
- 8:46Though in my if I understand, this is all the data that has been
- 8:50collaated, collated with all the knowledge that already exists, and then with the
- 8:56use of all that data available that you feel into those machines, they'll learn
- 9:03out of it, and then they come out with things and give results to
- 9:09you as for your proms or search. So and it could be you know,
- 9:13others material, but you will never know because some words here, some
- 9:16words here, you know, pulled out of hundreds and hundreds of hours of
- 9:22data that is available. So what are we learning new here? Except for
- 9:28you know, a few things here? So help us in both that sense
- 9:31that what is the purpose of artificial intelligence? Will we stop at that purpose?
- 9:37Or there is a dmius purpose beyond that, So help us understand.
- 9:41And in terms of learning, also are we learning new or it is the
- 9:46same old stuff? Good? Good question, I mean some big questions,
- 9:50some big questions there. So firstly, there's probably three different types of artificial
- 9:56intelligences that it's worth exploring here. So versus large language model generating texts.
- 10:00So now we can create a large amount of text relatively easily. So I
- 10:05can say, write a blog post about how AJ's podcast is amazing and everyone
- 10:11should listen to it, and it can spit out a couple of paragraphs completely
- 10:15truthfully, very convincingly, and I can use that for my marking materials.
- 10:18Okay, So it can start accelerating my work. It can make things easier
- 10:22for me. You can do things that are perhaps a little mundane pretty easily.
- 10:28And now I need some Now I need some human intervention to try and
- 10:31sparkle it up a bit and add some links into it and make it a
- 10:35bit more interesting perhaps, But it's going to start to make my job easier
- 10:39if I'm if I'm doing text, for example, writing text, it's one
- 10:41example. Okay, but what now if I'm using artificial intelligence to generate voices,
- 10:48so I can take text and turn that into speech. So if I
- 10:52can train an AI to sound like I do, then I do I need
- 10:56me anymore? Can I just use this AI to generate text generate speech.
- 11:01Well, that's one of the things we're actually doing in Polse podcast is taking
- 11:05text that we've generated a script for a podcast and then turn it using AI
- 11:09voices to turn it into a very realistic sounding and persuasive and engaging podcast.
- 11:15So we see that as a good thing because it allows us to reach a
- 11:18niche audience for podcasts that wouldn't be able to do this previously. So,
- 11:22as you know, it takes it takes time and effort to make a podcast.
- 11:26You have to edit it, you have to write a script, you
- 11:28have to maybe license music, get music engaged. You can have to then
- 11:33spend time marketing and delivering it. Part of that effort is involved in the
- 11:39actual mechanical aspects of actually generating that that podcast, and if you have you're
- 11:43a small company, can be quite expensive to actually create a podcast that actually
- 11:46supports your marketing efforts. So at Polse Podcasts, we actually spend time doing
- 11:52this using artificial intelligence. So it means that this effort takes less time,
- 11:54less energy, and can be done at a lower cost. So we see
- 11:58that as a good thing. I think where it steps over that line though,
- 12:01that you're talking about AJA is okay. So what if I'm in the
- 12:03middle of an election year, for example, and I'm generating voices that sound
- 12:07very much like the politicians that I'm hearing about, and they say something that
- 12:13they didn't really say. That's that's a genuine concern, right, And so
- 12:18we're starting to see that a lot of these engines now are putting in place
- 12:20restrictions on what they can do in terms of the voices for example. For
- 12:26so the engines that we use can't be used, for example, to generate
- 12:30the voice of a well known politician or even a well known celebrity, because
- 12:33it's important that they actually enable they have not only the control over their content
- 12:39and what the actual sound of their voice, but also the messages that they're
- 12:43giving as well. And so it's important that people understand what is real and
- 12:46what is actually not real, and so extra there's a challenge there in terms
- 12:52of trying to make sure that these AI engines are used for good purposes.
- 12:56And finally, the last part of that is engines that are now being able
- 13:00to create images. So we can say to an engine, now you've seen
- 13:03all these images from the entire Internet. Now we want you to create an
- 13:07image of aj climbing up a mountain, putting a flag on the top and
- 13:13saying he's conquered Mount Everest. And absolutely, you can now create images that
- 13:18have people in them, that have images in them. You can create the
- 13:22design the images themselves, and they do look amazingly photorealistic. So again,
- 13:28do I need the graphic designer to do that now or a photographer? Well,
- 13:31if it's if it's an imaginary scenario, then maybe not. Maybe I
- 13:33don't need a photographer. Now, I can do that by myself at home.
- 13:37I can type some words in and get a beautiful image that represents my
- 13:41product or what I'm talking about. But again, we need to make sure
- 13:45that there's the right restrictions around these ais so that they are real images that
- 13:48are either coming out of a camera or they're an artificial image, which is
- 13:54not actually aj It's not something that I can say, actually happened in real
- 13:58life. It's actually an artificial image. So are there I think that the
- 14:01challenges and the exciting parts and the challenges of using artificial intelligence right now,
- 14:09right still challenges. It is fine. Now let's evaluate this. You talked
- 14:13about podcasts. It is good where you use the technology. It minimizes the
- 14:18type of work, the mundane work that you need to do. Now in
- 14:24terms of you see the misuse we all know about deep pics. Last year,
- 14:28people were creating just for example, that how things can go wrong.
- 14:33They were creating podcasts out of two historical figures and talking within themselves, and
- 14:37they were actually you know, people will find in due course if you put
- 14:43it in the internet on the Internet and they were living, they were contemporaries,
- 14:46people will think these people actually had this conversation because people are not declaring
- 14:52it. But if you look at it in terms of learning how these two
- 14:56people will, or how to presidents or how to prime ministers in different errands
- 15:01talk, that is good in terms of learning. But as long as you
- 15:05have the right thing. So that is one part we need to understand.
- 15:09Now in the present, let's look at this. One example is Sora in
- 15:13terms of images video, and then we talk about Cora. Two days back,
- 15:20I read an article which said that Korra is almost dead. Why because
- 15:24robots are answering. And then robots are answering those those questions raised put out
- 15:30by robots itself. And so what is happening is it's a chain reaction which
- 15:35is going on where humans are off the creation process and robots are talking among
- 15:41themselves, people are using them, so that creativity is gone from the text.
- 15:46Kora is almost that the version of the writer us that Kora is dead
- 15:50in that sense, whether yes or no. That I still sometimes read if
- 15:54I feel that this is something very important, but now I also have to
- 15:58think whether this has been this is correct or not correct. Perceptions can be
- 16:03different, but facts are different. When people are writing about now in terms
- 16:08of Sorah, the Sora images are very real. Where is this aiding creativity?
- 16:15In some aspects? It is if I need to go back into the
- 16:18past and I want to see those something created which I will not be able
- 16:22to or anyone will not be able to do that, that is good.
- 16:26But if it, you know, push us out, the creators, push
- 16:30us out, the artists from the whole system, you know, then it
- 16:34is the wrong thing. And who will give that guarantee? Initially people may
- 16:38talk all about the good things, but I would because you see you also
- 16:41talk about you know, artificial intelligence has come up aage where we can do
- 16:45things now that are quite not quite human but human enough. Yes, the
- 16:51fear is that this human will not be able to hold the thin red line
- 16:56for long. It will be put out. That is absolutely the challenge.
- 17:03So now we're seeing from open eye their lodest latest model which enables us to
- 17:07not just generate images, but a series of images so videos now, so
- 17:11we haven't really seen how long this takes to create. The theory is it's
- 17:15it's going to take some minutes, tens of minutes even to generate these little
- 17:18videos. But it's the first time we've seen someone saying that they can generate
- 17:23minutes worth of videos. So up until now, it's been maybe a few
- 17:27seconds of video, so really just frame by frame by frame by frame and
- 17:30maybe fifteen frames as a second, they can get to a second, maybe
- 17:34a few seconds. So we've seen that, we've seen that in the in
- 17:37the past, but now we're seeing generating a minutes worth of video, which
- 17:41is really almost kind of long form. You can if you can do a
- 17:45minute, you can do five minutes. You can string a bunch of five
- 17:47minutes together. You can make an entire movie. You can make a whole
- 17:48video out of it. What are the challenges at the moment with the image
- 17:53creation and even the video creation is that it's very hard to do reproducible images.
- 18:00So I can do an image of a magically invented person and they might
- 18:03be they might be sitting an armchair having a glass of wine. Great,
- 18:07it's great as a single image. What is really challenging is for us to
- 18:11be able to then say to the AI, I want another image of that
- 18:15that same person that you've invented, but I want them in a different setting.
- 18:18So this is really hard right now. It's really hard to create an
- 18:22image of a person that looks a certain way and reproduce them again somewhere else
- 18:26in a different setting. Because the engine doesn't really have it doesn't kind of
- 18:30understand what it's done to actually create that image. It just creates an image
- 18:34of a person out of its I'm going to say imagination, but really it's
- 18:37out of its data bank of what people look like. But it's very hard
- 18:41to do that again. And so we'll be seeing images of seeing videos now
- 18:47of people that are walking down a street or having a glass of wine at
- 18:49a seeing them having a cup of coffee in a coffee shop, and that'll
- 18:56be interesting in its own but and it's very hard to get that same per
- 19:00and then put them in another setting for another video and enable them to do
- 19:03something else. So there's still a gap there. Short form videos, short
- 19:07amounts of things, no problems. It's it's a big challenge now to tell
- 19:11it from reality. But still it's very hard to take that person and do
- 19:14them, make them do different things, and so even getting them to talk
- 19:17or be able to sync their lips with text is hard. So I don't
- 19:21think we're going to see movies in the next five minutes, in the next
- 19:25couple of maybe even six months, maybe even you in five years time,
- 19:30we'll probably be able to create an entire movie from scratch and be able to
- 19:33say, I'd like a movie with me in it, for example, and
- 19:36it's an action movie, and this is the things that I want to happen,
- 19:37and it'll be able to generate that whole thing. But we're just on
- 19:41that kind of beginning of that creative journey. But I think I do think
- 19:45it enables us to use our imagination in new ways. I don't know if
- 19:48a few years ago a Jay, when photoshop came out, everyone was like,
- 19:52well, we don't need photographers anymore and we can just make anything out
- 19:56of anything. It doesn't matter. That's that's not really what happened. What
- 20:00enabled It enabled the whole area of creativity to open up for people to use
- 20:03graphic design to create new images. It created an entire new industry that wasn't
- 20:08there before. And so what we are seeing is that we'll see people that
- 20:11are using artificial intelligence to generate images and videos. An entire industry will spring
- 20:18up that there wasn't there before. Now, maybe there'll be less of using
- 20:23humans for particular tasks. But we've always found with technology that it really does
- 20:29open up new opportunities for us to do new things, use our imagination in
- 20:33new ways, and these tools enable us to be creative and to create things
- 20:38we hadn't even imagined before. And so that's why I think it's an exciting
- 20:41time for us using AI, because it does open up new opportunities for us,
- 20:45both in terms of business and in terms of our creativity in our own
- 20:48personal lives. Right right in terms of our personal lives, how much we
- 20:55will have control over our lives in the going day coming days, because already
- 21:00the issue some one of the top guys told me that listen, we must
- 21:04understand that there is nothing called privacy in today's time. We may consider sometimes,
- 21:11but we must so we have to use everything that we are using either
- 21:15on Internet or on our phones with a lot of precaution. Do not consider
- 21:19yourself secure or very things very private. Private. Things are private. Do
- 21:25not bring anything, you know, all those things, any of those things
- 21:27online. How do you see this part going forward? Because everything is so
- 21:36interconnected. You know, all those barons who own these companies, they are
- 21:41doing farming, opening, creating bunkers for themselves, whereas we are using these
- 21:47phones and all these technology and giving our whole story, whole life to them
- 21:53in the form of data. How do we see see a lot of these
- 22:00companies. Earlier governments, democracies were powerful. They used to represent the will
- 22:07of the people. But now these companies have got Some of these companies have
- 22:11got so much of power that companies or even the you know, elected government
- 22:17officials, they are depending on those people too during elections, during other things,
- 22:22you know, and even President by then he was using all those images
- 22:27during this thing. You may not like TikTok, but again you are using
- 22:30TikTok so that you can reach people, and artificial intelligence injected into these things
- 22:37can create huge amount of impact for you, presumably, But you are also
- 22:41giving away the power to these companies in the same way. Very recent report,
- 22:48China has apparently hired many private companies with deep powers of hacking into individuals
- 22:55and organizations and countries. So that is also alarming in the say So,
- 23:00what I mean to say is that the big artificial intelligence that is we are
- 23:07talking about, not till the avail leble businesses can use it in the right
- 23:11manner. That can pose the big challenge for governments, for individuals who are
- 23:18electing a government, but later on that thing is actually not happening on the
- 23:23ground, that is the big challenge. Democracy might get impacted. It is
- 23:29almost like a new world order setting in with those who are having the power
- 23:33of the artificial intelligence, who are controlling artificial intelligence, and of those who
- 23:38do not have that capacity. You may be very good in one thing,
- 23:42but you are not good. And at how robots talk help us give us
- 23:47an understanding of this aspect. You are a technology expert. Yeah, that's
- 23:52a good that's a good question. And there are many things we should be
- 23:55worried about here. So if I can't control my own my own image,
- 23:57the use of my own image, the use of my voice, and some
- 24:02other company has access to my voice and has access to my image, then
- 24:06what control do I have at that point? What's going to stop these companies
- 24:08taking my image and taking my voice and making me say anything that they want.
- 24:12I mean, I don't think that anyone's too worried about what I'm going
- 24:15to say. But for some people that are in a position of power or
- 24:18position of influence in terms of our politics and our government, that can be
- 24:22very important and very disastrous in fact in many ways, because if they can
- 24:26be said, say something that's market affecting, and that gets out and people
- 24:30believe it for five minutes, that can definitely have effect on the markets and
- 24:33how people do financially as well. And so we are seeing governments now that
- 24:38are spending a fair amount of time and effort in order to get their hands
- 24:42around what this technology means and how we should how we should regulate it.
- 24:47So in the EU, for example, is just publish some legislation that tries
- 24:52to define what companies can and can't do with AI, both in terms of
- 24:56the generation of images and videos and texts and so on, but also in
- 25:00terms of where does this data come from? So what data can companies access
- 25:04that give them give them this ability to do these things in the first place,
- 25:08and what permissions do they need in order to be able to use this
- 25:12data for their business? And so I think that's important that governments get their
- 25:18hands around what this means. And in many ways, I agree that we
- 25:22have kind of run along in terms of our technology, but the legislation is
- 25:26only just starting to go. Hang on a minute, is this really what
- 25:29we want to happen? Is this really how we want to act as a
- 25:32society. So that's important that the governments make some lines in the sand about
- 25:37what we think is acceptable as a society in terms of what these artificial intelligences
- 25:41can do and what these companies have access to in terms of feeding the actual
- 25:47AI itself, in terms of what their capabilities are. So I think we're
- 25:49getting there. I think that we're not quite there yet, because even as
- 25:53governments are making this legislation, the technology is barreling down, and I think
- 25:56it's too late to stop this now. I don't think it's the thing that
- 25:59we want. I think it's a new tool, it's a new capability that's
- 26:02exciting things on the horizon for us. But with all technology, with all
- 26:06technology change, we need to make sure that the people that are involved in
- 26:11this, both in terms of actors and voice actors and content writers have rights
- 26:17that are appreciated and respected within our society, but also that we can have
- 26:22the opportunity to make use of this amazing technology to increase our creativity, give
- 26:26us extra powers in terms of our ability to do things that we couldn't do
- 26:30before. That's a great thing, but we need to make sure that we
- 26:33do that in a way that respects the capability, the content, and the
- 26:37people that are going to be using these tools themselves. Right, right,
- 26:41And so while we are talking of artificial intelligence, there is a new thing
- 26:45called artificial general intelligence. Now help us understand that. I've heard a couple
- 26:52of people talking about it. I once googled it, but then I forgot
- 26:56exactly what it is all about. So tell us exactly this article PICI general
- 27:00intelligence is all about? Again? Is it a threat? Is it a
- 27:06good thing? How should we look at it? Yes? So, as
- 27:08we've seen these large language models, which is probably the best example, where
- 27:12we used to be able to ask them a simple question like what is the
- 27:15capital of Paris? Sorry, what is the capital of France? And it
- 27:18would say it's Paris. So that's some sense of does it feel intelligent or
- 27:23does it just no facts? Okay, so maybe that's more like a fact.
- 27:26And then we'd say, okay, can you write a poem about Paris?
- 27:30Tell us about give us a four line poem about what it feels like
- 27:34to walk along the river in Paris. Okay, So now it's generating text.
- 27:38It seems a little more intelligent, a little more creative. Even the
- 27:44best of these large language models now can generate very plausible, very interesting,
- 27:48very capable summaries of text. It can generate information about other texts that it
- 27:56knows about. It can be reasonably creative about things. But it's really not
- 28:00not approach the level of writing an actual passionate novel that gets our heartstrings going,
- 28:07gets us excited. It's kind of just a little bit bland. It's
- 28:12kind of capable of writing sentences, but not really kind of I don't know,
- 28:17sort of not really approaching what humans can do in terms of our capabilities
- 28:21of writing and even creativity in terms of painting or capturing images and so on.
- 28:26And AGI is really about getting to the point where it is, to
- 28:30that point indistinguishable from what a very capable human could do, and in some
- 28:37ways even beyond human. In some ways, these large language models are already
- 28:41slightly beyond humans in that they have access to all the knowledge of the entire
- 28:45Internet at their disposal, You and I could look something up. But if
- 28:48we are having a discussion and we're not on the Internet, it's very hard
- 28:52for us remember now, now, when exactly was when exactly do this particular
- 28:59historical event happen? Well, large language models have access to all that information
- 29:03in their heads, if you like already, Okay, so we've got a
- 29:07whole kind of facts behind us. If we add the ability to be incredibly
- 29:10creative as well, if we have access to all of human knowledge, then
- 29:14we're at a point of beyond just being an expert in something, but being
- 29:18an expert in everything. Like a general intelligence generally accessible, they're generally capable
- 29:26of anything. At that point, we're reached a level where some people refer
- 29:30to it as almost like the singularity, like the point where the intelligences that
- 29:34we've created us smarter than us. And this is the thing that we're now
- 29:38talking about in terms of AGI. So what does that mean for humanity?
- 29:41What does it mean for us? I mean the worry that people have is
- 29:45that the general intelligences, these amazing machines, we'll look at humanity and go,
- 29:51well, I don't know. They're not so good after all, and
- 29:52maybe when we don't keep them around, is that the worry. In fact,
- 29:56these we have to remember that these intelligences themselves are really just computer programs.
- 30:02The software that is running it doesn't have any more capability to do anything
- 30:07other than what we give it to do ourselves or give it control over.
- 30:11And so if it's generating text, if it's generating images, maybe you can
- 30:15be amazingly persuasive, maybe you can create beautiful images, but really it's not
- 30:19going to take over the world and start doing things to humanity. It doesn't
- 30:23have that capability. It's just a computer program that's running. But what's exciting
- 30:27about that is that imagine if you had access to an intelligence that knew everything,
- 30:33was entirely capable of describing things to you in a way that you understood
- 30:37it, that could teach you and help you learn about things that you're interested
- 30:41in. That sounds amazing, That sounds great, And so that's the positive
- 30:44side of this idea of an ADI. It is an exciting concept, but
- 30:48we have to remember that in the real world, outside of science fiction,
- 30:53these intelligences are not really a thing that's biologically capable of doing anything. It's
- 31:00just a piece of software that we can turn off at the end of the
- 31:02day. And so really we don't have anything to worry about in terms of
- 31:04those Now, let's hope that no one watches this video in ten years time
- 31:07and realizes the entire world has been taken over by an ATI. But I
- 31:11don't think we have anything to be worried about. Like myst software, it's
- 31:15a tool. It can be used for good or it can be used for
- 31:17bad. But again that's like with all the tools that humanity's had all the
- 31:21way along. Absolutely, absolutely, I would also like to be very positive
- 31:26on this like you are, and let's look look at the situation added and
- 31:30ports. But we you once are smart, much more smarter than but we
- 31:34ourselves no be that as it may. Let us talk about your company,
- 31:41Yeah, Pulse Podcasts, So tell us a bit about that. What do
- 31:45you do and how people can make use of it? Absolutely, so we're
- 31:49excited about what we're doing because we were we were looking at creating podcasts for
- 31:56companies to help them expand their marketing efforts, and what we were finding is
- 32:01that to actually create a podcast, it's an expensive, time consuming effort.
- 32:06You need to get people into a studio, you need to be creating scripts,
- 32:09and we wanted to help companies do this, to create these scripts and
- 32:13to create these recordings and to add music and to mix it all together.
- 32:16And we have some companies that work with us doing this kind of thing.
- 32:21But what occurred to us is that there were many companies that really did want
- 32:23to get into podcasting because now with podcasting, we're seeing that there's hundreds of
- 32:29millions of users that love listening to podcasts every day across the world. That
- 32:34people drawn into podcasts because whatever particular thing you're interested in, you can find
- 32:39a podcast that is about that particular thing. The beauty of podcasts is that
- 32:45for me, I can find something that these exactly what I'm interested in.
- 32:47Fantastic, Okay, that's great. And if I'm interested in a particular brand,
- 32:52or a particular technology, or a particular hobby or whatever all a particular
- 32:55location, that's great. But the difficult is for the people that are providing
- 33:00this content into a podcasting form. If there's a small audience that's not many
- 33:05many, many, many thousands or hundreds or millions of listeners, it's actually
- 33:08really quite expensive to produce based on the number of people that are listening to
- 33:13it. Okay, so is there a way we can reduce that cost and
- 33:15make it accessible. So what we've done is to use help companies use the
- 33:21content they're already creating. So if you're creating an email newsletter or you're creating
- 33:25blog content on a regular basis, you're doing this already, then we can
- 33:29help you by taking exactly that existing content. We create scripts using artificial intelligence,
- 33:36we do several revisions of that. Then we take the scripts, we
- 33:38feed it into an artificial intelligence that creates a voice, and it's beautiful recording
- 33:43of the voice, and then we use our software to create a platform that
- 33:46actually enables people to mix music into it as well to actually edit the actual
- 33:52audio itself and upload it and create descriptions and titles and everything for the whole
- 33:55podcast itself. And so the exciting part is that companies can now create a
- 34:00beautiful, engaging podcast based on their existing content for a tenth of the cost
- 34:05that it would cost normally to actually get a voiceover artist to do the actual
- 34:08voice, to get us content created to create the script itself. So we
- 34:12can still do that, have those capabilities for companies that really need that sort
- 34:15of attention, but for a lot of companies or a lot of government agencies
- 34:19or cities or local government areas, it enables them to reach their local area.
- 34:23For example, for local councils that have maybe they have fifty thousand people
- 34:28in their constituency, and to create a podcast will be enormously expensive for the
- 34:32number of people in their constituency, but now they can do it cheaply and
- 34:35easily, and people can use that to get hold of the information that's happening
- 34:38in their local area on a regular basis and wherever they happen to be.
- 34:43And the great thing about podcasts is that you can get access to the same
- 34:47content that you might read on the web, but in an audio form.
- 34:51So now that I can if I'm commuting, I can now access that content
- 34:54while I'm on a train, rather than having to have a phone or having
- 34:58a laptop out to be able to access that same content. And we're seeing
- 35:00some great uptake with companies that we can now produce some beautiful content based on
- 35:06their existing content at a much lower cost. And that's I think one of
- 35:08the advantages of artificial intelligence is it gives us the capability of doing great things,
- 35:14reaching niche audience and you know, producing a quality product at a much
- 35:19much lower cost than we were doing previously. Wonderful. So what is the
- 35:22best way for people to learn more about you, connect with you, and
- 35:27connect with your company. We'd love you to come and visit us at Pulse
- 35:30Podcast www dot pulse podcast dot com. Check out some of our sample podcasts
- 35:36and we can show you from an existing newsletter and how we've turned it into
- 35:38a beautiful podcast and how and drop us a note and we'll show you how
- 35:43we can do exactly the same thing for you. So send us some content,
- 35:45we'll send it into a turn into a podcast for you and you can
- 35:47see what it sounds like. That would be great, wonderful, wonderful With
- 35:52this, it's set up on this very special edition of The Cage and Masterclass
- 35:57Life. Thank you so much indeed for joining us. Thanks for having me
- 36:00along today. It's been great.