Summer is here! Temperatures on the rise (we almost reached 30ºC for a couple days), and rain seems to be over (not saying that too loud, I do not want to jinx it). Lots of interesting news this week, so let’s get to it.
A dinner with Myxomycetes
Really nice macro photographies of slime moulds. Not much else to say honestly, if you like either photography or biology, there is a big change you will enjoy them. Go take a look!
Slime Moulds (Myxomycetes), on Barry Webb Images
AI taking over?
Lots of news related to AI this week. I think they can be bundled into two main groups: AI problems/misuses and people trying to fix AI. On the problematic side, we have how AI systems have already taken over so many crucial aspects of our lives, such as giving childcare benefits to people who need them. Apparently, the Dutch had been using machine learning tools to detect fraudulent applications, but the system was rejecting people in need of help that were not trying to do anything wrong. To make things worse, it seems that the algorithms were tagging ethnic minorities and low-income families as fraudulent due to some kind of training bias. We also have another example of evil use of technology in the case of the San Francisco police, that started using autonomous driving vehicles (which carry a high-tech bundle of sensors to navigate the world) to surveil people. Last, we have another exhibit on what happens when these kind of systems tackle unknown problems: they terribly fail. In this case the people at the MIT tech review report on how business tools to control stock and trade have been messed up due to our change on buying patterns after the Covid pandemic.
However, there are some green sprouts coming (if you feel optimistic). The European Union has proposed a new law that will try to regulate and protect its citizens from the most dangerous uses of AI (such as facial recognition in public spaces, training biases, etc.). Many people (myself included most of the time) always claim the EU to arrive quite late to the party, but I think that for once, they have been one of the first to worry about the topic and are trying to do something to tackle it. Let’s see how the thing evolves (and how fast), but I am crossing fingers 🤞🏼
A quick guide to the most important AI law you’ve never heard of, on the MIT tech review
The Dutch Tax Authority Was Felled by AI—What Comes Next?, on IEEE Spectrum
Our weird behavior during the pandemic is messing with AI models, on the MIT tech review
Looking at the void… with algorithms
This week we got the first images of the black hole in the center of our galaxy, and some of you might be thinking “this image rings a bell”. Actually, it is quite similar to the one previously acquired by the Event Horizon Telescope, but at that time they peeked at the heart of the M87 galaxy. In both cases, the images have been acquired by using telescopes all around the globe, and merging the data using computational tools that allow to enhance sensitivity and resolution to levels high enough to see such distant objects. There are a couple nice infographics on the links below, in case you are interested.
Behold: The first picture of the black hole at the center of our galaxy, on National Geographic
More modern warfare
News keep appearing on different aspects of the Russian invasion on Ukraine. This week I saw this article about how, before the invasion started, Russian hackers attacked an American satellite company in order to diminish Ukrainian communication capabilities. It’s been already 80 days since the invasion started, and despite the sanctions (which some companies like the French Thales might be evading), the conflict does not seem to be coming to an end soon.
Russia hacked an American satellite company one hour before the Ukraine invasion, on the MIT tech review
Save me Obi-Wan
I have seen dozens of displays trying to produce holographic images, and all of them show how difficult it really is to achieve real 3D displays. This one is no exception, though the results are quite impressive and the technology its based on (acoustic trapping) could be orders of magnitude faster with just a few years of work. Moreover, using pressure waves also allows to have haptic feedback, which could be quite useful in many applications.
Genetic algorithms learning to beat Jump King
I found this quite cool, to be honest. I am used to this kind of algorithms being used to solve phase retrieval problems in optics, and some other inverse problems in the field of optimization. However, I never saw such a cool video explaining both the underlying ideas of the technique, and also how to use this amazing tool for such a daunting task as beating a video game. Anyway, if you know both fields, just looking at the training stage is also kind of mesmerising.
AI Learns to play JUMP KING, on Youtube
And that’s it for the week. Stay safe!
Featured image: Sagittarius A*, at the centre of the Milky Way. From the Event Horizon Telescope project