1 00:00:07,133 --> 00:00:10,453 Speaker 1: You're listening to the Saturday Morning with Jack team podcast 2 00:00:10,573 --> 00:00:13,013 Speaker 1: from News Talks at b It. 3 00:00:13,053 --> 00:00:15,933 Speaker 2: Has been a stonking year for AI. So just look 4 00:00:15,973 --> 00:00:19,613 Speaker 2: at open ai, the makers of chat GPT. They are 5 00:00:19,653 --> 00:00:23,453 Speaker 2: on track now to thirteen billion US dollars in revenue 6 00:00:23,453 --> 00:00:26,093 Speaker 2: for this year, which is up from four billion US 7 00:00:26,333 --> 00:00:28,773 Speaker 2: in twenty twenty four. But they're looking at maybe hitting 8 00:00:28,773 --> 00:00:33,053 Speaker 2: annualized revenue of about nineteen billion dollars, which is extraordinary. 9 00:00:33,133 --> 00:00:36,493 Speaker 2: That's just from the paid user accounts on chat gpt, 10 00:00:36,613 --> 00:00:38,373 Speaker 2: and of course you can just use the free version. 11 00:00:38,773 --> 00:00:41,333 Speaker 2: Texbert Paul Stenhouse is here to look back at the 12 00:00:41,413 --> 00:00:45,493 Speaker 2: year in artificial intelligence. Paul, did they get as far 13 00:00:45,573 --> 00:00:47,413 Speaker 2: as you expected they would get this year? 14 00:00:48,453 --> 00:00:48,813 Speaker 3: Oh? 15 00:00:48,893 --> 00:00:51,373 Speaker 4: Good question. I think they did. 16 00:00:51,533 --> 00:00:54,933 Speaker 3: To be honest, Jack, I use a lot of I 17 00:00:55,013 --> 00:00:57,573 Speaker 3: say that as someone who uses AI a lot for 18 00:00:58,973 --> 00:01:01,733 Speaker 3: coding and technical things, and it seems to be one 19 00:01:01,733 --> 00:01:02,893 Speaker 3: of the areas where I've. 20 00:01:02,733 --> 00:01:03,933 Speaker 4: Really leaned into it. 21 00:01:04,893 --> 00:01:07,053 Speaker 3: There seems to be a war at the moment between 22 00:01:07,413 --> 00:01:11,533 Speaker 3: open ai and Google, Gemini and Anthropic to really kind 23 00:01:11,573 --> 00:01:15,133 Speaker 3: of nail the software developer experience. They're typically people who 24 00:01:15,173 --> 00:01:18,413 Speaker 3: want to try new things that are usually pretty. 25 00:01:18,413 --> 00:01:19,533 Speaker 4: Okay with change. 26 00:01:20,253 --> 00:01:22,533 Speaker 3: And what's interesting about code is there's a sort of 27 00:01:22,573 --> 00:01:26,453 Speaker 3: corpus of old projects you can reference and also kind 28 00:01:26,453 --> 00:01:29,373 Speaker 3: of documentation as well. Right, so it lends itself quite 29 00:01:29,413 --> 00:01:32,693 Speaker 3: well to AI. But it still isn't perfect. 30 00:01:32,773 --> 00:01:34,613 Speaker 4: But when I think about. 31 00:01:34,333 --> 00:01:38,133 Speaker 3: The things that it could largely could not do a 32 00:01:38,213 --> 00:01:42,093 Speaker 3: year ago or two years ago, Wow, it has come 33 00:01:42,293 --> 00:01:44,173 Speaker 3: a really really long way. It is not perfect, but 34 00:01:44,213 --> 00:01:46,933 Speaker 3: it's come a really really long way. I think it's 35 00:01:46,933 --> 00:01:50,613 Speaker 3: just interesting too that the people who probably have never touched, 36 00:01:50,653 --> 00:01:54,613 Speaker 3: of touched or thought of AI, maybe this year gave 37 00:01:54,653 --> 00:01:56,933 Speaker 3: it a go. I feel like it's reached a far 38 00:01:57,013 --> 00:02:00,253 Speaker 3: wider group of people. They've kind of started to dabble 39 00:02:00,373 --> 00:02:03,613 Speaker 3: as things have become, you know, a little more baked 40 00:02:03,613 --> 00:02:06,173 Speaker 3: into some of the services that they use every day. 41 00:02:06,813 --> 00:02:13,773 Speaker 2: So what is your preferred text generation service at the moment? 42 00:02:13,813 --> 00:02:17,053 Speaker 2: Do you still stick with chat GPT, because in recent 43 00:02:17,133 --> 00:02:19,733 Speaker 2: months there has been a bit of a sense that 44 00:02:19,813 --> 00:02:23,413 Speaker 2: Google is catching up, that Gemini is actually getting as good, 45 00:02:23,453 --> 00:02:25,333 Speaker 2: if not maybe better than chat GPT. 46 00:02:25,653 --> 00:02:27,413 Speaker 3: Yeah, And it really depends what you want to use 47 00:02:27,413 --> 00:02:29,253 Speaker 3: it for. And I know that sounds like a bit 48 00:02:29,253 --> 00:02:31,453 Speaker 3: of a cop out, but if, for example, in the 49 00:02:31,453 --> 00:02:32,133 Speaker 3: coding world. 50 00:02:32,333 --> 00:02:33,533 Speaker 4: It seems to change. 51 00:02:33,253 --> 00:02:36,413 Speaker 3: Every literally every few weeks, as these models get tweaked 52 00:02:36,733 --> 00:02:39,453 Speaker 3: and they're making changes and deployments. Now when it comes 53 00:02:39,493 --> 00:02:43,213 Speaker 3: to things like you know, some are slightly more creative, 54 00:02:43,333 --> 00:02:46,613 Speaker 3: some are better with academic some are better with more 55 00:02:46,693 --> 00:02:50,453 Speaker 3: formal type of writing, some are better at puzzing PDFs. 56 00:02:50,453 --> 00:02:54,213 Speaker 3: And I'm making understanding it really does change based on 57 00:02:54,293 --> 00:02:56,333 Speaker 3: the type of task that you're doing, which I know 58 00:02:56,613 --> 00:03:01,013 Speaker 3: sounds crazy, but they do sort of have this this 59 00:03:01,093 --> 00:03:03,293 Speaker 3: is going to sound even crazier. They almost have this 60 00:03:03,373 --> 00:03:05,813 Speaker 3: kind of like personality behind them jack or like a 61 00:03:05,933 --> 00:03:09,213 Speaker 3: skills because they are trained on things, and they do 62 00:03:09,293 --> 00:03:12,733 Speaker 3: have these kind of hidden instructions telling them how they 63 00:03:12,773 --> 00:03:15,213 Speaker 3: should be working and how they should be responding and 64 00:03:15,453 --> 00:03:19,693 Speaker 3: if they should be you know, responding more succinctly or 65 00:03:20,133 --> 00:03:22,213 Speaker 3: giving you more sources. And all of these things are 66 00:03:22,213 --> 00:03:24,813 Speaker 3: baked in the background, so things that you don't even 67 00:03:24,853 --> 00:03:28,013 Speaker 3: see about those agents are kind of controlling them. So 68 00:03:28,053 --> 00:03:30,573 Speaker 3: I would say try some of them. There's lots of 69 00:03:30,613 --> 00:03:33,333 Speaker 3: them that start out for free. I found things like 70 00:03:33,373 --> 00:03:36,253 Speaker 3: perplexity to be really good for kind of current from 71 00:03:36,293 --> 00:03:40,453 Speaker 3: trying to do anything about current information. I find you know, 72 00:03:40,733 --> 00:03:43,173 Speaker 3: the Geminis of the world. If you're a Google Drive 73 00:03:43,653 --> 00:03:46,613 Speaker 3: or Google kind of workspace user, it's really easy to 74 00:03:46,613 --> 00:03:48,613 Speaker 3: tap into that world and get it to understand some 75 00:03:48,693 --> 00:03:50,173 Speaker 3: things about the documents. 76 00:03:49,813 --> 00:03:52,893 Speaker 4: You already have in there. So yeah, it's just you 77 00:03:52,933 --> 00:03:54,853 Speaker 4: can just you can kind of pick and choose. 78 00:03:54,613 --> 00:03:57,853 Speaker 2: You can experiment, yeah, yeah, and try and different things. 79 00:03:58,173 --> 00:04:00,653 Speaker 3: Yeah, and then music dedicated apps too, right, So things 80 00:04:00,693 --> 00:04:03,333 Speaker 3: like if you've ever made a PowerPoint presentation and you 81 00:04:03,373 --> 00:04:04,453 Speaker 3: never want to have to do that ever. 82 00:04:04,533 --> 00:04:07,013 Speaker 4: Again, there's an app called Gamma, for example. 83 00:04:07,733 --> 00:04:11,573 Speaker 3: So we're starting to see really specialize AI tools that 84 00:04:11,613 --> 00:04:14,213 Speaker 3: you can use to do very specific the Yeah. 85 00:04:14,293 --> 00:04:16,813 Speaker 2: Right, okay, So can I put you on the spot 86 00:04:16,813 --> 00:04:18,493 Speaker 2: and ask you for a couple of predictions in twenty 87 00:04:18,533 --> 00:04:21,573 Speaker 2: twenty six do you think Okay, first of all, do 88 00:04:21,613 --> 00:04:25,053 Speaker 2: you think that we are going to achieve this agi 89 00:04:25,253 --> 00:04:29,133 Speaker 2: artificial general intelligence, which is the kind of you know, 90 00:04:29,173 --> 00:04:31,973 Speaker 2: the kind of guiding staff for many of these big 91 00:04:32,013 --> 00:04:35,173 Speaker 2: tech firms. They want to basically achieve artificial intelligence that 92 00:04:35,253 --> 00:04:37,133 Speaker 2: can do anything and more that a human can do. 93 00:04:38,853 --> 00:04:41,333 Speaker 4: Not in twenty twenty six, Yeah, I don't. 94 00:04:41,133 --> 00:04:43,253 Speaker 2: Think so either. I don't think so either. I think 95 00:04:43,373 --> 00:04:47,493 Speaker 2: if anything, even though the large language models, the text generators, 96 00:04:47,493 --> 00:04:51,533 Speaker 2: the chat GPTs of the world are improving, and certainly 97 00:04:51,573 --> 00:04:55,653 Speaker 2: that the models that are helping with coding have improved massively, 98 00:04:56,093 --> 00:04:59,453 Speaker 2: I think we're probably not. It feels like the kind 99 00:04:59,453 --> 00:05:03,693 Speaker 2: of incremental improvements now as opposed to really big lockstep improvements, 100 00:05:03,693 --> 00:05:04,173 Speaker 2: you know what I mean. 101 00:05:04,173 --> 00:05:07,213 Speaker 3: Here's what's interesting, right, it can produce a lot of stuff. 102 00:05:07,693 --> 00:05:10,973 Speaker 3: And that's why the Merriam Webster word of the Year 103 00:05:11,093 --> 00:05:13,853 Speaker 3: is slop because that's what it's known as AI slop. Right, 104 00:05:13,973 --> 00:05:17,493 Speaker 3: These things just create stuff, whatever that is. But when 105 00:05:17,493 --> 00:05:19,413 Speaker 3: you start to look at like coding is a really 106 00:05:19,453 --> 00:05:22,573 Speaker 3: specific example and a really good example because it can 107 00:05:22,613 --> 00:05:24,693 Speaker 3: produce code and you can test to see if. 108 00:05:24,573 --> 00:05:26,813 Speaker 4: It works very easily and very quickly. 109 00:05:27,013 --> 00:05:31,773 Speaker 3: When it spits out sentences and sentences about quantum physics 110 00:05:32,013 --> 00:05:35,613 Speaker 3: or how to you know, look after a dog. 111 00:05:35,493 --> 00:05:37,573 Speaker 4: You know, like, it will give you information. It will 112 00:05:37,573 --> 00:05:38,453 Speaker 4: give you words. 113 00:05:39,453 --> 00:05:42,213 Speaker 3: They may be sentences, they may be paragraphs, but is 114 00:05:42,253 --> 00:05:45,293 Speaker 3: the content actually good and is actually correct? 115 00:05:46,133 --> 00:05:47,293 Speaker 4: And it's sometimes when you. 116 00:05:47,373 --> 00:05:50,053 Speaker 3: Really if you start to use AI for things you 117 00:05:50,133 --> 00:05:52,453 Speaker 3: know a lot about, or if you give it very 118 00:05:52,493 --> 00:05:56,173 Speaker 3: specific documents for example, and you're asking very specific questions, 119 00:05:56,773 --> 00:05:58,933 Speaker 3: it's very easy for you to then start to see 120 00:05:58,813 --> 00:06:00,373 Speaker 3: where it goes wrong. 121 00:06:01,253 --> 00:06:02,173 Speaker 4: I heard someone. 122 00:06:01,973 --> 00:06:05,333 Speaker 3: Refer to it the other day as a veneer of intelligence, 123 00:06:06,053 --> 00:06:07,453 Speaker 3: which I thought was quite interesting. 124 00:06:07,533 --> 00:06:10,653 Speaker 2: Yeah. Nice, Okay, I like that, ive a here of intelligence. 125 00:06:11,053 --> 00:06:14,413 Speaker 2: My next My next question, repredictions. Do you think that 126 00:06:14,933 --> 00:06:17,533 Speaker 2: we are going to see an AI bubble burst. 127 00:06:18,773 --> 00:06:19,973 Speaker 4: My portfolio? Short? 128 00:06:20,173 --> 00:06:27,973 Speaker 3: Yeah, not yet. I don't think not yet. Okay, Look, 129 00:06:27,973 --> 00:06:29,693 Speaker 3: I think we still need a lot of things, right. 130 00:06:30,733 --> 00:06:32,333 Speaker 3: I just saw a news article today there's a lot 131 00:06:32,373 --> 00:06:36,413 Speaker 3: of pushback, for example, from locals on AI data centers 132 00:06:36,493 --> 00:06:38,493 Speaker 3: or data centers in general, that people don't want them 133 00:06:38,493 --> 00:06:41,533 Speaker 3: in their communities. They suck up utilities, drive up power prices, 134 00:06:41,573 --> 00:06:43,373 Speaker 3: et cetera, et cetera, kind of these types of things. 135 00:06:43,373 --> 00:06:45,413 Speaker 3: So we may, you know, we may see some really 136 00:06:45,413 --> 00:06:48,653 Speaker 3: interesting pushback that starts to kind of soften some of 137 00:06:48,653 --> 00:06:52,213 Speaker 3: this kind of AI hype. But look, isn't it the 138 00:06:52,293 --> 00:06:54,613 Speaker 3: art of isn't it the kind of a g I type, 139 00:06:55,093 --> 00:06:57,813 Speaker 3: you know, general intelligence level where it can really do 140 00:06:57,933 --> 00:07:01,493 Speaker 3: anything right, but it can do a lot of interesting stuff. 141 00:07:01,933 --> 00:07:04,093 Speaker 3: And I think where we will see is in this 142 00:07:04,213 --> 00:07:08,653 Speaker 3: next year when people really start to integrate AI into 143 00:07:08,653 --> 00:07:11,293 Speaker 3: their workflows, because it really does have kind of, you know, 144 00:07:11,333 --> 00:07:14,333 Speaker 3: in some fields, the intelligence quote unquote of like a 145 00:07:14,413 --> 00:07:16,453 Speaker 3: junior employee. 146 00:07:16,853 --> 00:07:18,573 Speaker 4: When we start to see that, I think that's when 147 00:07:18,613 --> 00:07:19,173 Speaker 4: we'll see. 148 00:07:18,973 --> 00:07:22,293 Speaker 3: What the real what the real impact is of the 149 00:07:22,333 --> 00:07:24,333 Speaker 3: economy and everything. 150 00:07:24,053 --> 00:07:25,373 Speaker 4: Else, everything else. 151 00:07:25,733 --> 00:07:28,173 Speaker 2: Okay, so it was a big year for AI. Next 152 00:07:28,253 --> 00:07:29,533 Speaker 2: year is set to be a big year for AI. 153 00:07:29,613 --> 00:07:31,733 Speaker 2: But you twenty twenty five was also a big year 154 00:07:31,773 --> 00:07:32,693 Speaker 2: for past keys. 155 00:07:33,293 --> 00:07:35,013 Speaker 4: I know, Jack, it's the last show of the year. 156 00:07:35,053 --> 00:07:38,493 Speaker 3: I usually always give my psa that says, if you've 157 00:07:38,493 --> 00:07:40,653 Speaker 3: got some time over the holidays, you should update your 158 00:07:40,653 --> 00:07:44,533 Speaker 3: past words. But twenty twenty five has been the year 159 00:07:44,573 --> 00:07:47,573 Speaker 3: of the past key, which it's this new kind of 160 00:07:47,613 --> 00:07:51,493 Speaker 3: alternative to passwords that's coming. And it's great because if 161 00:07:51,533 --> 00:07:53,573 Speaker 3: you think of a password, you kind of just yell 162 00:07:53,653 --> 00:07:55,973 Speaker 3: it at the website that you're going to, and if 163 00:07:56,053 --> 00:07:58,653 Speaker 3: you're the if it's the wrong website, you can hear 164 00:07:58,733 --> 00:08:02,053 Speaker 3: the password effectively. Right what past keys are? It goes, Hey, 165 00:08:02,573 --> 00:08:05,173 Speaker 3: who are you? And you go, I'm Jack, And I go, well, Jack, 166 00:08:05,413 --> 00:08:06,493 Speaker 3: I've got something for you. 167 00:08:06,773 --> 00:08:09,493 Speaker 4: But only because you're Jack. And that's the past key. 168 00:08:10,773 --> 00:08:13,733 Speaker 3: How is that the more secure they are, more they 169 00:08:13,733 --> 00:08:17,893 Speaker 3: are encrypted, they get stored significantly more securely on your device. 170 00:08:19,893 --> 00:08:22,333 Speaker 4: You want to find out a little bit more about them. 171 00:08:22,373 --> 00:08:24,333 Speaker 3: I do think the other way that everything is going. 172 00:08:24,573 --> 00:08:26,653 Speaker 3: They are a little tricky though, because they do kind 173 00:08:26,653 --> 00:08:30,133 Speaker 3: of get stuck on the device or the service at 174 00:08:30,173 --> 00:08:33,093 Speaker 3: which you generate them for right, So, because we need 175 00:08:33,133 --> 00:08:36,013 Speaker 3: to have that little handshaky thing. We do imagine if 176 00:08:36,013 --> 00:08:37,853 Speaker 3: you only have the paskey on your computer and then 177 00:08:37,893 --> 00:08:40,853 Speaker 3: you lose your computer. So you've got to put them 178 00:08:40,853 --> 00:08:43,613 Speaker 3: into things like password managers. You've got to make sure 179 00:08:43,613 --> 00:08:45,693 Speaker 3: you have things like a recovery email is set on 180 00:08:45,733 --> 00:08:46,693 Speaker 3: your account. 181 00:08:46,293 --> 00:08:49,973 Speaker 4: Things like that. But they are great. They stop phishing. 182 00:08:51,093 --> 00:08:54,613 Speaker 3: I don't see why every bank hasn't been adopting them because. 183 00:08:55,613 --> 00:08:57,773 Speaker 4: They make you more seems like a god makes it 184 00:08:57,773 --> 00:08:58,293 Speaker 4: more secure. 185 00:08:59,213 --> 00:09:02,133 Speaker 2: Hey, thank you so much, Paul, have a fantastic Christmas 186 00:09:02,133 --> 00:09:05,773 Speaker 2: and we will Catchiggy and very soon. Paul Stenhouse our textbook. 187 00:09:06,493 --> 00:09:09,493 Speaker 1: More from Saturday Morning with Jack Tame. Listen live to 188 00:09:09,573 --> 00:09:12,613 Speaker 1: News Talks at B from nine am Saturday, or follow 189 00:09:12,653 --> 00:09:14,213 Speaker 1: the podcast on iHeartRadio.