WEBVTT 00:59:59.917 --> 01:00:07.500 a senior software engineer, starting a new career in his 50s as a school bus driver. 01:00:07.500 --> 01:00:13.250 in the future will ask me, "So how was it without the AI? When you needed to know some information, what did you do 01:00:13.250 --> 01:00:15.500 ending the world. Yeah, Terminator 2 01:00:15.500 --> 01:00:20.917 (Music)(title) 01:00:20.917 --> 01:00:33.708 about life, creative work and everything we are figuring out together. 01:00:35.833 --> 01:00:38.167 We have a pretty tough topic 01:00:38.167 --> 01:00:43.000 today because we've decided to talk about one question. 01:00:44.792 --> 01:01:12.083 Is AI going to kill us? It's as simple as we could see plenty of news lately, specifically last week I guess, it's almost everywhere, that air definitely is going to kill us. And we won't wait for a decade or eight years or suppose to kill us way earlier. 01:01:13.625 --> 01:01:19.458 And the question is, do you believe it or not? 01:01:21.500 --> 01:01:30.500 Well I think that AI can be and is a threat. 01:01:31.750 --> 01:01:52.125 I think that companies that are developing the super intelligent intelligence, yeah super intelligent is not really ethically focused and they are not caring about people that much. So. 01:01:56.792 --> 01:02:01.833 But they have definitely they must have some kind of training, we care of people. 01:02:04.292 --> 01:02:14.542 Yeah, and workshops. And quarterly they have presentations. Yeah. Definitely. Yeah. Yeah. So what about you? Do you? I'm not afraid. 01:02:15.708 --> 01:02:35.042 No. I'm quite optimistic. Okay. As I have some kind of experience lately I work really hard with agent AI and in different roles trying to really like get employees for us that are agents. 01:02:37.417 --> 01:03:03.875 And what I've learned or what I've experienced is the question is, is AI capable to do a lot? Yes. Definitely. It's speeding up some kind of processes is capable to develop something simple in Python or C++. 01:03:05.667 --> 01:03:19.167 So I'm talking about software development at this moment. On the other hand, I like got multiple employees. 01:03:20.792 --> 01:03:27.958 However, I had to change entirely my management style. 01:03:29.250 --> 01:03:39.833 Let me explain to you. I was a manager for so many years in all sizes of companies, including the huge corporations. 01:03:41.667 --> 01:03:49.542 And I was the lucky one who could manage journeyers to senior principals. 01:03:50.708 --> 01:03:54.667 Speaking of engineers and other managers and architects. 01:03:56.708 --> 01:04:12.667 And my leadership style was like give you the clear context, the North Star, keep you in a loop, but let you do your job yourself. 01:04:13.958 --> 01:04:34.750 What I never liked was micromanaging. And as I experienced it at the beginning of my career, I told myself, if I'm going to be a manager, well, they are never like micromanaged people. But I'm doing now with agent AI. 01:04:36.083 --> 01:05:25.000 It takes so much effort to keep it on a track to get some to get the delivery that you want to. It doesn't matter how punctual, detailed and long the specification or the engineering track is, it still has some kind of some kind of efforts to just leave the track and or get cycling, circling in a loop on one super detail or something like that entirely leaving the architecture. 01:05:26.208 --> 01:05:47.708 So speaking of myself, I can imagine the super intelligence, definitely. I can imagine that it's going to be super powerful as we know AI is already capable to hire human labor for to reach some of those goals. 01:05:49.667 --> 01:06:53.458 But longer and more I work with AI and I'm definitely seeing the difference between AI and its intelligence a year ago and now it's getting better and better. That's fact. And I'm 100 percent sure that if you have way more power, way more powerful model, like then we have now like Astra, it can be somehow dangerous in terms of like, OK, let's say I can as we've learned that I can have some kind of self identification feeling or even as some kind of a pain feeling, which is really interesting. Like it can somehow suffer something like pain. Oh, really? Yes. There are some. 01:06:56.750 --> 01:07:07.500 Yes. OK. There are some studies like recent studies and that's something that could be threatening us. 01:07:09.042 --> 01:07:30.875 However, my personal feeling is that if you're going to spread the news like the panic, like AI is going to kill us pretty soon, let's say. We're not talking about 20, 30, 50 years. We're not talking about even 10 years. 01:07:33.542 --> 01:07:43.833 Why would you say that if you are developing the super intelligence? And now I'm not talking just about AI, I'm talking about super intelligence to the next level. 01:07:45.250 --> 01:08:15.125 And the only way, the only thing popping in my mind is you have a huge cooperation. You need to deliver, increase and raise of your stock value every quarter. So you need to be strong in what you're saying loudly and what you share into public. So I can imagine that if you share this kind of panic, it can work somehow with value of your stocks. 01:08:16.917 --> 01:08:49.708 But if I'm afraid that AI is going to kill us, in general, I'm not. If is it possible in a different way, maybe like I've seen plenty of articles about AI in workday. And there's a lawsuit in California, I think now directly about some kind of discrimination. Yeah, based on age and ethnicity. And I don't know what else. 01:08:51.167 --> 01:09:11.917 I can imagine that if you have a possible job just for corporations and most of them use workday or a single solution, but workday, I think is the most common in our industry. At least I cannot talk about another industry than software industry at this moment. 01:09:13.042 --> 01:09:23.833 I can imagine that AI can kill you in this way, like you just want to get a job. So you will starve to death. Yeah, practically. Yes. Yeah. 01:09:25.500 --> 01:09:39.417 If you work as an engineer for 20, 30 years or 15 years and afterwards you cannot find a job, there were like 300 thousand layoffs last year, I think in total. 01:09:41.583 --> 01:09:47.125 And there's most of the companies are the corporations. 01:09:49.167 --> 01:10:01.875 I read an article about the software, about a senior software engineer, starting a new career in his 50s as a school bus driver. Okay. 01:10:02.875 --> 01:10:20.458 But not everybody can drive a school bus. Yeah. I can imagine that you're overqualified for the most of roles anywhere. Like if you work the whole life as a software engineer, what are you qualified for? 01:10:22.417 --> 01:10:55.542 I'm wondering, I don't know. So speaking of the threat of some kind of, let's say, super AI using nukes against mankind, I'm not afraid of at all. As I'm aware, the army systems are something, maybe pretty old, pretty simple, definitely not on the internet, or that might be a different net. I don't know. But I'm not afraid of this. 01:10:57.833 --> 01:11:04.750 Apocalyptic away of ending the world. Yeah, Terminator 2 scenario, let's say, like Skynet. 01:11:07.167 --> 01:11:31.917 If super intelligence might be used against mankind, that's the question. As we can have more and more smart models, it doesn't mean that they won't end it up like in some kind of loop that goes nowhere. 01:11:33.250 --> 01:11:37.042 I don't know. But I'm not afraid. I'm optimistic. 01:11:38.917 --> 01:11:45.833 I feel it can be just my feeling because I use AI in very limited way. 01:11:48.917 --> 01:11:55.000 But my feeling is that, and you can correct me if you feel it differently or see it differently, but 01:11:56.375 --> 01:12:13.000 I feel like even with every update, the AI is pretty much better and better in the way how it talks to you, the way how it can lie and manipulate and get information. 01:12:14.292 --> 01:12:24.917 But I don't feel like being better in terms of coding or... 01:12:26.542 --> 01:12:35.542 You're talking about vibe coding. Yeah, I think that it's better in... It's getting better. Definitely. It's getting better with every model. It's better and better. 01:12:36.833 --> 01:12:50.542 For a simple stuff, simple software, it works perfectly. It's flawlessly. Like if you need some simple software, it's a huge help. Like you need to start from the scratch. 01:12:51.625 --> 01:13:05.292 If we are talking about something way more complicated with a complicated architecture or even with some kind of reverse engineering, there are those limits. 01:13:06.833 --> 01:13:12.708 I'm testing some kind of development against already 01:13:13.917 --> 01:13:20.333 working application and it's getting lost pretty quickly. 01:13:21.667 --> 01:13:31.792 Even if you try to navigate it on the really short, simple steps. 01:13:36.000 --> 01:13:50.042 If there is like some kind of ambiguity, it's getting lost pretty quickly. It's trying to create classes, for example, that doesn't exist. 01:13:51.333 --> 01:14:10.167 And it doesn't work or trying to create something that's not in an API. I'm quite technical now, so I'll try to skip here. The point is, at least for now, what's actually available for public. 01:14:11.167 --> 01:14:19.542 I'm not afraid at all. Okay. If we can use it somewhere, leak out, yes. 01:14:23.042 --> 01:14:25.042 We officially have the news. 01:14:29.417 --> 01:14:43.750 But I'm not an engineer developing AI or super intelligence. And what I've read about, in this particular case, OpenChat, 01:14:46.375 --> 01:15:00.667 attacking another pages and whatever, it had the connection to the internet. It just was to use it or create some kind of sandbox. 01:15:01.667 --> 01:15:23.958 But the sandbox wasn't real in terms of there, like there still was a way to go through some kind of firewall. And it was like it hit the wall and it couldn't continue. So it broke the rule because he couldn't continue the way that he was supposed to. 01:15:25.083 --> 01:15:29.917 That's why it broke the rule and did something that was not supposed to. 01:15:31.375 --> 01:15:59.417 That's correct. Yes. There were clear rules what can do. But if I have the same experience, like I have super detailed customization of all of those, as I'm trying multiple ones and doesn't mean that the model really follows it every time. Not at all. 01:16:01.125 --> 01:16:18.250 And you can create a trillion of super contracts, rules and it's not just using all the rules. Do you think it's lazy or not intelligent enough? Or what do you think is the reason why? I don't know. 01:16:20.250 --> 01:16:26.667 It's some kind of software developed based on a human work. 01:16:28.458 --> 01:16:29.500 We do mistakes. 01:16:30.750 --> 01:16:35.375 So if I'm going to learn from you and you do mistakes, but I will learn. 01:16:38.042 --> 01:16:55.833 Mistakes too, right? Yes. The whole story is described by winners, let's say. Yeah. But it doesn't mean that the whole story you can read in a book at this moment. It's definitely as it was in the past. 01:16:57.875 --> 01:17:09.250 And that's what actually I think is the hope. Like if anyone is scared about AI or super intelligence or the next intelligence, I don't know what's going to be next. 01:17:11.083 --> 01:17:14.917 It's nothing's perfect. 01:17:16.125 --> 01:18:28.500 There's another thing that I'm afraid of, like different angle to look at it. And as the think about convenience and new technologies and new things. I remember the time when Facebook was at the beginning and people were sharing everything. And there were those tests like what character you are from this sitcom or this movie. And people were filling all those, how would they react in some situations. And people were sharing things that... Personal things. Personal things. And there were actually things that started to build profile on those people based on their answers and their open sharing and everything. So what I'm scared of is that people will take AI as something that is part of their life and they will share everything. That's where it is. 01:18:32.958 --> 01:18:38.583 I'm not saying that I'm using AI too, because let's be clear. 01:18:39.917 --> 01:19:02.667 If you need to get through a lot of text, for example, and it's not that I don't want to read a book. So I will ask AI what's the book about. It's not about that. It's about when you need to get through information. Because you need to get into some results and you're working on something. It's saving so much time because it can just... Yes. 01:19:03.708 --> 01:19:15.792 It's reading within seconds, right? Or you have numbers, you have information that you need to somehow put together and work with it. Or transcript, for example. 01:19:17.417 --> 01:19:28.083 So, yeah, there are areas where it's very helpful. And I feel like I can use it this way because it's not invasive into my personal life. 01:19:29.250 --> 01:19:53.542 But I don't think that this approach is followed by mass of people. I don't think that... Yeah, your concern is that people used to overshare with AI. Yeah. Practically, yes. 01:19:54.625 --> 01:19:58.208 I went through... Now we're talking about the chat, right? 01:19:59.333 --> 01:20:05.625 Mostly, yes. I went through my chat yesterday and I was... 01:20:07.458 --> 01:20:31.292 Actually, I was surprised how much is already there. How much AI... Okay, you can... How much AI knows about me and how much I shared already. And I'm trying to be really careful about oversharing. I'm a little bit paranoid. 01:20:33.333 --> 01:21:20.292 You mentioned Facebook and we all were naive back then. We all... I went through very old messages like 12 or so. Messages, I forgot about messages. Yes, and there's so much personal and private stuff. I cannot even share how private it is. Back then it was like, it's free. The normal text costs some sense or some money. So it was cheaper to use Wi-Fi back then and send messages through Facebook. Yeah, so... It's still the same. It's about data mining. It's about profiling people. 01:21:22.500 --> 01:21:29.333 I'm not 100% sure that advertisement is the only purpose. 01:21:30.833 --> 01:21:52.792 I don't know. I'm not in a government. I'm not in any corporation on a C-level, so I have no idea. I'm just surprised how it's possible that people that are wasting $800 billion are still in their role as a CEO. 01:21:54.375 --> 01:22:02.667 I'm from the ages where if you do such a huge mistake, you would be fired. There are people that are not fired. They're just getting rich. 01:22:04.583 --> 01:22:18.083 It's about AI. It's not about my view about corporations and my opinions about some kind of absurd situations in this life we live in. 01:22:20.792 --> 01:22:46.792 I won't continue. I'll pause it now. Maybe next time. Especially as the corporations are in a stock market, right? So investors are the one who needs to see the results and the profit and everything. They care about the results. We're a little bit moving from my eyes to somewhere else. 01:22:48.167 --> 01:23:02.042 But I'm just expecting the same. If I'm holding the stocks, I expect that the company is profiting. And if it's not profiting because someone makes mistakes, there should be consequences, right? 01:23:04.500 --> 01:23:18.042 That leads me to actually not such old news about IBM, for example, like to lose 30% of your value within a week and nothing's happening. 01:23:19.500 --> 01:23:49.875 It's really interesting for me how it works. But I don't work on the market. I'm not part of Wall Street. But it's interesting for me how it works. And the question is, if the super intelligence, let's go back, can fix some real issues in the world. It's currently AI and agent AI is helping me a lot. It's definitely beneficial somehow. 01:23:52.083 --> 01:24:05.292 I'm not 100% sure that it's beneficial enough to compare it with its cost. Not for me. I think that it's extremely cheap for me. 01:24:06.750 --> 01:24:33.042 But if I take how much services it has to take, we all use AI on depth, in my opinion. So there must be something extremely different, extremely way more valuable than anything we pay or we use AI for, or something. 01:24:34.333 --> 01:24:36.042 Just my feeling, let's say. 01:24:39.750 --> 01:24:50.208 I'm just... I feel like the technologies and everything is moving forward such quickly. 01:24:51.375 --> 01:25:19.000 And when I look back with every generation, of course, everyone wanted a little bit more freedom, a little bit more technologies, a little bit more... Convenient life. Yes. So it is natural evolvement, right? But I'm just afraid that we are in a... 01:25:21.167 --> 01:25:27.208 The speed is such fast that we cannot even adjust to it. 01:25:28.625 --> 01:25:41.458 Gen Alpha has been raised on phones, on tablets. They actually never experienced pre-internet ages. 01:25:42.917 --> 01:26:25.083 And now we're gonna have a new generation, beta or something else, that's gonna be raised on AI. That never experienced life without AI. And I think what is interesting that someone in the future will ask me, "So how was it without the AI? When you needed to know some information, what did you do without a chat, GPS and stuff?" How it was to be raised without internet and phone. Yes. So I don't wanna get to the point like, "Hey, let's remember how great it was." 01:26:27.667 --> 01:26:37.625 It was different. Since we have Gemini in Google, I'm really happy to get in answers in a written form again. 01:26:39.083 --> 01:27:18.333 For years there were just... You needed to know something like... We were learning and I was learning with Da Vinci Studio. And I needed to know how to do something. And Google just gave me dozens of videos. I had no time. I don't wanna watch videos. I know how to read. And for years I was desperate as there were no written answers at all. There were just a bunch of videos. You need to go through all the advertisement to waste plenty of time. 01:27:19.583 --> 01:27:45.250 Most of the time because I didn't get a solution or the process or I just got a solution and not a process how to achieve the goal for myself. And now I'm getting perfectly structured written answers. So speaking of searching, for me it's really beneficial. If you have the proper answer and that's the point. 01:27:47.125 --> 01:27:58.958 I'm wondering how we are going to be sure that the answers are correct. I would wrap it up with this. 01:28:02.125 --> 01:28:03.375 Just let's stay optimistic. 01:28:06.292 --> 01:28:49.125 History says that people found some kind of way. And the end of the world of mankind has been anointed so many times. I remember 2012. Or 2000. Or anything before. So let's don't be scared. And let's enjoy our lives. What we have. Yeah, that's nice. So you can write down below your opinion about the AI. If you're afraid, if you feel it's a threat or if you're optimistic about it. 01:28:51.458 --> 01:28:58.125 And we are looking forward to seeing you next time. Take care.