Episode transcript The transcript is generated automatically by Podscribe, Sonix, Otter and other electronic transcription services.
Hi everyone. Welcome to the 5 Minutes Podcast. Today I'd like to discuss and make my comments about the Gemini, the new large language model, or the new AI generative AI of Google that was released last week. And first, it was announced as something like, you know, we are now putting in the marketplace a real competitor of OpenAI and ChatGPT. Just to remember, Gemini works like the GPT, so Gemini will be operating in Bard. The ChatGPT of Google, like GPT four, works on ChatGPT from OpenAI. Okay, so Gemini is the model and all the AI approach and the AI model that will support Bard. They released in three versions, but the top one that everybody wants to see is the ultra version, which has not been released yet but is to be released. And Google released several very powerful benchmarks comparing Gemini with ChatGPT. And then there were a lot of polemics on that comparison because nobody could, I would say, independently review that, and they compared with GPT four and not GPT four turbo. So there were a lot of polemics on that, and this was not a surprise for me. It's part of the game. What was a surprise for me was when I saw the demo video of Gemini, and trust me, I was happy that I was sitting in a chair because otherwise, I would fall to the floor with the amount of surprise. There was a yellow post-it, and then, like a life AI at the Bard, was talking and answering and saying, what is this? Oh, this is a paper that you are taking notes on, and then it starts drawing. All this is a line.
This is a curve, a line, and this up to the blue duck. And then, the user in the demo took the blue duck and said, oh, this looks like rubber or wood. And then, you know, it squeezed it, and then it made a sound and said, oh, it's rubber. And this was stunning. At that time, I thought to myself and said, imagine this in the project environment. Imagine you doing scheduling and Bart answering your schedule as you draw it in your hands. Or, for example, you are making a Kanban board, and Bart is saying no. Split this work into two post-its because this will be more than the number of resources you have, and then you have a lot of work in progress on that. So something like this insight or, for example, you are listing in a bullet point a series of stakeholders, and Bart could answer to you, oh, this stakeholder is more important than this. Shift their position, and this is as you draw. And this was wonderful. But then two days later, a little bit less than two days later, I received a Bloomberg article someone shared with me. Then, it says that that video was not so true. And I thought, what is this article talking about? And it's a Bloomberg article, and it's not something like an obscure website. And it says that Google admitted that they had edited that video, that they have reduced the latency and this and, you know, when I saw that, I said, come on, what's going on here? And I immediately recall my podcast about what reality is.
And I say, look, do these people realize that they misled everybody to think that Gemini Ultra would be like, you record a video, and Gemini will start explaining the video as you record because this was the connotation that the video and the demo were showing. And then they said, no, it was not like that. It was more static. We wrote some of the prompts, so it was not spoken. And then I said, come on. So this is completely different. So it was a little bit frustrating. And for me, Google did not need to do that. But what is happening is that people are so anxious to show movement on AI that they are doing the wrong things, that they are guiding people to think in a different way because that video was not true. And then you can see on YouTube and on X, the former Twitter, hundreds of comments saying, oh, Google is now fostering fake news. And this. So what the point here is that we need to understand what is reality and what is not reality because people like me who are waiting to be practitioners of that can raise an expectation, and then I can decide, for example, not to develop, project or develop a different project with an expectation that these would be available on that.
Way. And this is not true, at least based on the video that was recorded. But this, in the end, also gives me a perception that things are moving so fast that probably in a matter of weeks, not months or not years, we will truly be seeing tools like you. Record a video of your project, for example, of your construction site, and Gemini will start saying, oh, this is this, this is happening. Change that. Oh, this is late. Because look, it was supposed to have ten bricks, and it has only two. So I think this is the travel of our journey and this will be extremely positive for the performance and the work we are doing. What is not nice is this race. Don't mislead people because it backfires against you. So just be mindful of that. It's like when I'm developing the PMO tool, a lot of people ask me, can you release a motor for me? And this, and I'm being extremely conservative on releasing. Why? Because I do not want to mislead anyone that PMO is doing something when this feature is not ready yet. So think about that and see you next week. And if you have time, take a look at this video. It's very nice. If this becomes real, it will be really disruptive. It's sad that this video does not represent the reality, at least the reality we have now. I hope you enjoyed this podcast, and I will see you next week with another 5 Minutes Podcast.
