Month: August 2023

Today marks the third time we made it to float down the Shenandoah River on tubes, or scientifically speaking, tubing. It’s about 40 minutes from home. There are two options, flat water and white water. The first – is essentially a typical river near Moscow. You float, drink beer from a bottle labeled as juice. Boring, but it attracts the main bulk of people. White water – is a mountain river of about a third-degree difficulty, in some places. On site, they really scare people about the difficulty, causing them to downgrade their bought pass to flat water. In our presence, two families switched. Well, about the difficulty. There are no clear movement lines, just rocks, the rapids are generally not dangerous, but still, a tube is a tube. It’s poorly controllable. There is no helm. Your arms are the paddles. I capsized from it about three times. I quickly learned that when falling, you need to raise it high above your head because any touch of the water sweeps you along with it to the rocks, but this way, you can manage, albeit barely, to stay in control and consciously do something. Overall, I got a few bruises and bumps, but the family seems not to have fallen even once, and it looks like I’m the only one who got sunburned. In general, a wonderful way to spend three hours floating five kilometers.

Yucca and pit

I’m looking at the new history textbook for 11th grade. An interesting artifact. Found page 415, which contains three paragraphs to which there are not even any questions. That’s exactly how it will be.

Interesting to listen to. Alexander Kraynov, Director of Artificial Intelligence Development at Yandex. From the category of “didn’t say anything new, but the thoughts and reasoning are worth the time spent”. That is, for the average engineer the video is not about knowledge, but about contemplation.

A brief excerpt from the middle:

“…But, as it seems to me, we are looking for [AI danger] not quite in the right place. (…) It is now clearly apparent that many people can hardly drive without a navigator. Someone who turns off the navigator from time to time can still navigate. But those who are hunters, or, for example, hunter-gatherers, were able to move and navigate in space without navigational aids. Our ancestors had these skills. They could go into the forest and, without any tools, navigate by the stars, getting to the needed place. Traveling five thousand kilometers or even ten thousand kilometers through the taiga was normal for them. They possessed these skills. Now, we face difficulties. If you put a person in the forest, how many of them could return? Previously, this was not a problem—people traversed distances by orienting themselves with the environment. Now we can no longer do this.

Now it often happens that people turn to basic questions such as “How to cook dumplings?”, “How to cook sausages?”, “How to boil eggs?”—they look for answers on the internet. We are accustomed to relying on the internet for any information. This is characteristic not only of humans but also of other living beings. We strive to minimize the effort to achieve a result. If we have no need to think, we turn to the internet. This tendency is only getting stronger. As a result, our lives increasingly depend on the internet. Suppose, because of this tendency, our lives would be fully oriented towards the internet.

Suppose. I imagine, the most likely apocalypse, if it occurs, it would look like this: on a not-so-lucky day, some admin rolls out some faulty, conflicted balancer and takes down the entire internet. Then, it just completely stops working.

And then, it needs to be fixed somehow. Another admin, who should fix it, where is he, the administrator? He’s at home with a laptop, what he can do remotely, as he always did, but he cannot do it remotely because there is no internet. He needs to somehow travel somewhere, he doesn’t know where to even ask because there is no internet, you have no connection, the navigator doesn’t tell you where to go.

Well, let’s assume he knows where to go, and he even has gasoline in the car. Also, gas stations do not work because their payment systems are down. Then he arrives at the data center gate, the building hatch. And next, how he can’t enter, his magnetic badge does not work. Okay, let’s assume, he somehow entered, hacked in. There you are, wanting, this laptop at this center. And at that point, nobody can explain to me at all what can be done. You don’t understand where you should connect, and that’s it. You don’t know how to do it, you can’t fix the internet without the internet, and there are no knowledge and no technical possibilities, nothing.

And now imagine, what if our internet disappears. I’m not sure, by the way, that humanity currently possesses sufficient knowledge and can remake it from scratch. We are rapidly losing such skills because we are not peeking anywhere, it is nowhere recorded in books, but it’s in the internet, if it doesn’t work, you can’t get knowledge.

Imagine such an exercise: imagine you’re thrown back 500 years. They tell you, you’re from the future. You have electricity there. Let’s make it. But at the stage of making the wire, you are already breaking down. You’re a normal person, you don’t understand how to make it. We can use it, but we can’t make it ourselves.

And it seems to me that in the future, such problems will progress, and the danger that threatens us is not some artificial intelligence that will rework us, I really don’t believe that, but our instability in facing technological disasters.

Caused by our excessive reliance and incapability, the inability to create things. I would actually conduct such exercises—agreeing that, say, the next three days we turn off the internet worldwide. Entirely. Not just leaving smartphones at home—completely turning off the internet and watching. Just for control and to see if humanity can last three days or not.

now it’s possible!

https://arxiv.org/pdf/2308.01074.pdf

The guys presented a practical implementation of a deep learning model for classifying laptop keystrokes using a smartphone microphone. When trained on keystrokes recorded by the nearest phone, an accuracy of 95% was achieved in determining the text typed by a person based only on the sounds of the keyboard. When trained on keystrokes recorded via the Zoom application, an accuracy of 93% was achieved. These are quite interesting results.

Our two cats

To my surprise, I found that drawing cars is more difficult than drawing faces. I tried to sketch our RAV4 “quickly” but got the proportions a bit wrong. The challenge is that you draw the car standing next to it, and because of this, the perspective distortions are much stronger than with smaller objects at a greater distance. Additionally, the shape of a car is defined by a continuous array of intricately inclined planes and transitions between them, rarely aligning with the lines of perspective. Plus, specific details distinguish the car against other brands, and I still need to figure out which ones, other than the front of the car. Overall, it’s an interesting exercise that is still challenging, but I will keep trying. Luckily, there are always enough parked cars on the street.

It seems that converting from direct to indirect speech is no longer a problem. It’s interesting why it still isn’t being integrated anywhere.

I’m a few years late with videos about robotaxis on city streets, but just in case someone hasn’t seen it yet. There are quite a lot of these little cars. Well, how many… to film them, you need to wait about 15 minutes in the city center – at least one will definitely pass by. Not in the center, it might take a bit longer. Primarily, you see the Cruise (GM) and Waymo (Google) cars. They operate day and night, even during rush hours. You download the app, place an order, and off you go.

Do some groundwork.