Semantris
When was the last time you played Tetris?
This is a fun and interactive project that allows you to play word association games with artificial intelligence. Recently created by the Google-AI team, Semantris focuses on processing simple human speech and differentiating the logical connections between words. This is an analogue of Tetris, but instead of blocks it contains words. The game has two operating modes: Blocks and Arcade.
In the first mode, the AI sets a word and asks you to find the closest combination to it, such as “sleep” and “bed.” It then compares the relationship between your answer and the original word with the other words in the list. The more matches you have with other words, the more points you get. A correct word between blocks causes that block to be deleted. The second mode is implemented as a simple timed association game with scoring points.
A neural network is a class of computing system. In this application, it establishes the meaning of individual words and finds semantically similar ones. It is created from very simple processing nodes formed into a network. Neural networks are designed in the image of biological systems, like the brain, but many orders of magnitude less complex. They are pattern recognition systems and are more useful for pattern recognition tasks.
Google uses Semantris to train vector models for neural networks and test artificial intelligence.
What is artificial intelligence?
Many different definitions of artificial intelligence have emerged over the past few decades, but John McCarthy's 2004 article (PDF, 106 KB) (link is external) defines AI as: “The science and technology of creating intelligent systems, especially intelligent computer programs.” . Artificial intelligence is concerned with the task of using computers to understand the workings of the human mind, but is not limited to the use of techniques observed in biology.”
However, the topic of artificial intelligence first appeared much earlier: several decades before this definition. In 1950, Alan Turing published his landmark scientific work Computing Machinery and Intelligence (PDF, 89.8 KB) (link is external). In this article, Turing, often called the “father of computer science,” asks the question: “Can machines think?” To test this, he proposed a test that everyone knows today as the “Turing test”: the experimenter tries to determine from written answers whether he is interacting with a person or a computer. After publication, the results of this test caused heated discussions, but despite this, the test still remains a significant milestone in the history of AI and even to some extent a philosophical concept, since it works at the intersection with linguistics.
Some time later, Stuart Russell and Peter Norvig published the book “Artificial Intelligence: A Modern Approach” (link is external) , which became one of the most famous and popular textbooks on AI. The authors classify AI into four main categories that characterize computer systems based on rational thinking and action:
Human approach:
- Systems that think like people
- Systems acting like people
Ideal approach:
- Systems that think rationally
- Systems that act rationally
Alan Turing's definition could be categorized as “systems acting like people.”
At its most basic, artificial intelligence is a field that combines computing technology with robust data sets to solve problems. Along with the concept of “artificial intelligence,” two of its subsections are often mentioned: machine learning and deep learning. These disciplines are a collection of AI algorithms aimed at creating expert systems capable of making predictions and classifications based on input data.
Today, there is still a huge excitement around the development of AI, which is inevitable when any new technology appears on the market. According to Gartner's Technology Maturity Cycle Theory (link is external), every innovative product, such as self-driving cars or personal assistants, follows a "pattern of development curve: from over-enthusiasm through a period of disappointment to the eventual understanding of the significance of the innovation and its place in the market or in a particular field." As Lex Friedman notes in his lecture (link is external) at MIT in 2022, we are now at the “peak of inflated expectations” and approaching the “precipice of disappointment.”
As the debate over the ethics of AI grows, we may be seeing the first signs of approaching the “chasm of disillusionment.” More information about IBM's position on AI ethics can be found here.
This Person Does Not Exist
This girl looks beautiful, but she doesn't exist. It's just another AI creation. Don't try to find any of the created people online, even though some may be very beautiful.
Software Developer (OCR, C++, ML)
ABBYY, Moscow, can be done remotely, From 220,000 ₽
tproger.ru
Jobs on tproger.ru
Many people have probably already heard about this service. This is the first project of its kind and it is not surprising that it has gone viral. Here you can create fake images in a second. All you need to do is go to the site and refresh the page.
The author of the project is Uber engineer Philip Wang. He used an algorithm (originally written at Nvidia) to make hyper-realistic portraits. This algorithm is known as a generative adversarial network (GAN). All images are generated from an extensive library. The system sequentially uses two neural networks: one creates the picture, and the other evaluates its realism.
Wang initially used this as a way to convince several friends to join him in independent AI research. He then decided that a wider audience could benefit from the potential of GANs. Wang also argued how important it is to educate people that this type of technology can be both revolutionary and dangerous.
This is how it started...
After the faces project launched, other developers used its code (available publicly on Github) for other sites like thiscatdoesnotexist.com and thesecatsdonotexist.com to create non-existent cats. They look pretty realistic, but you may notice some strange bits. Sometimes it looks very creepy.
Artificial Intelligence and IBM Cloud
IBM is a leader in AI technology for the enterprise and has pioneered the future of machine learning across many industries. Based on decades of AI research, years of experience working with organizations of all sizes, and lessons learned from more than 30,000 projects with IBM Watson, IBM has developed a step-by-step AI process for successful AI adoption:
- Data collection: Simplify data collection and provide efficient access to it.
- Organization: creation of a turnkey analytical platform.
- Analysis: Building scalable and reliable AI-powered systems.
- Implementation: integration and optimization of systems across the entire enterprise environment.
- Modernize: Move AI-enabled applications and systems to the cloud.
IBM Watson provides enterprises with AI tools to help transform business systems and workflows, dramatically increasing automation and efficiency. To learn more about how IBM can help you on your AI journey, see IBM's portfolio of managed services and solutions.
Register for an IBMid and create an IBM Cloud account.
AutoDraw
This is a tool that captures your doodles to get the perfect image.
Another exciting and entertaining service, and for some, very useful. It helps turn sloppy sketches into clear images.
You can draw any object you want. Just draw a few lines and the AI will predict what exactly you wanted to draw. It can easily recognize rough sketches, be it a flower, a cake, or a smiley face. It prepares pre-drawn alternative sketches created by artists to replace your designs.
This service is based on AI and machine learning. AutoDraw uses neural networks to compare images against an extensive database and select the most similar options. The more details you add, the better and more reliable the recommendations will be.
Smart cars
Another area where AI is increasing its presence in our daily lives is in smart cars.
Not only are companies like Tesla using automation technology in vehicles, but many car manufacturers are also considering integrating AI with cars to ensure seamless maintenance of the various technologies you use.
Information is shared and transferred between vehicles to better maneuver in traffic conditions. Input traffic data, real-time road block information, etc. Information is instantly transmitted to alert other vehicles on the network, allowing route redirection.
Cyborg Writer
This experimental text editor can add continuation to the words and sayings you write. Just enter any words and an artificial neural network will complete your sentence as if it were written by Shakespeare, the US Supreme Court, Tupac Shakur, or anyone else.
The entire system is based on a trained model. It is used to predict the most appropriate response to the entered text.
Sometimes Cyborg Writer's options are far from the meaning of the entered text. But the creators did not promise perfect accuracy from the very beginning and called it an experimental project. Perhaps the service will become more advanced over time. In the meantime, it’s quite suitable for just playing around and seeing what answer the AI will choose for you.
Tesla [automotive]
The electric vehicle manufacturer develops and implements artificial intelligence to control cars. Elon Musk claims that Hardware 3 digital vision will process up to 2,000 frames per second. This is the company's own product. Previously, Tesla used Nvidia Drive, equipment that was less efficient (200 fps).
Elon Musk's predictions are fantastic. The billionaire says that within 10 years, artificial intelligence will surpass humans in the safety and reliability of driving a car.
Talk to Books
Another AI-based application from Google AI. It clearly demonstrates what this innovative technology is capable of. Talk to Books answers questions through quotes from books. Unlike standard search, this service is not afraid of even the most abstract questions, for example, “what is the meaning of life?” or “what does it mean to be human?” So you can count on honest answers to even the most unusual questions. Take a look at the results for the query: “what is Cthulhu?”:
This service uses vector language learning to find books with contextually relevant answers to questions. All this works without matching keywords. Vector word language modeling matches synonyms, antonyms, related terms, context clues, and other natural ways of using language to find relationships between statements.
The project database contains about 100,000 works. The recommendation engine tries to find the most relevant excerpts in terms of meaning. You can also filter books by category and select only the genres you want.
How to install Assistant Dusya on PC and laptop
After reading our review, you probably have only one question left: how to get Dusya on PC? Everything is quite simple here: you will need a Droid4x emulator, with which you can create an imitation of another operating system on a PC or laptop. In our case, Android is native to “Dusya”. After you have downloaded and installed the program, you can download the necessary application through the built-in Play Market. By the way, we strongly recommend that you do just that, because there will probably be something else in the App Store database that will interest you or prove useful.
Quick, Draw!
According to the creators themselves, this game was created using machine learning. You draw, and the neural network tries to guess what you drew. This doesn't always work, but the more you practice, the more the network will learn. So far, this resource has been trained on examples of several hundred drawings, and the developers hope to add more sketches in the future. The application is designed as an example of how AI training can be used in a fun way.
You will be asked to draw 6 different objects in a limited time. As you draw and add more detail, the neural network tries to guess what you were trying to draw.
At the end, the results will be summed up which of the items were guessed. By clicking on any of them, you can see on the basis of which drawings the network was trained in each subject.
Let's note the important points of installation
How to register a google account
There are two solutions to this issue:
- register a new account in the droid4x interface.
- create a new account on google.com and register.
How to install using .APK file
It may turn out that you did not find applications/games in the catalog or you decided to launch a modified application that is not available on Google Play, which we do:
- Download the .APK file from a reliable source on the Internet.
- Drag the file into the window of the running droid4x program or open the emulator and go to “Settings” - “Other Setting” and click on the Apk Install button.
How to set up keyboard and mouse controls
The emulator developers have provided for this point. When you first launch the application on a PC, control will immediately switch to the keyboard and mouse using the standard “layout”. The user can change and reassign controls as desired.
AI applications for smartphone
Replika
Replika is an app that will help you expand your communication.
AI-powered virtual assistants are everywhere. Siri can set an alarm and Alexa can call an Uber. But there is no one to really have a heart-to-heart talk with. Replika fits exactly this idea. This is an artificial intelligence that scans your social networks and asks a lot of questions about you. Replika is naive and curious, like a child, but this is only at first glance. The application is learning, so a lot depends on you. Each answer will be saved in memory and can be used later. In this way, you can create your own personal copy of yourself, and then enter into dialogue with it.
Jarvis Artificial Intelligent
Jarvis was created by Mark Zuckerberg, who in turn was inspired by the virtual assistant from the wonderful movie Iron Man. Jarvis can control lights and temperature, select music, and set up security systems in your home. This virtual assistant constantly learns to recognize voices and faces. All these features are available using the mobile application. To be honest, the app still needs some work, but it's definitely worth a try.
Youper
Youper is a free and convenient AI-powered assistant that will help you solve your problems. It works as a fitness app that also contains cognitive behavioral therapy (CBT), meditation and Acceptance and Commitment Therapy. Along with this, you can use it as a stretching application. Youper tries to take care of you by giving you various health tips.
System requirements for installation
Since this application is designed to work with mobile devices, it runs without problems even on fairly weak laptops and personal computers. The only limitation is that you must have an operating system of at least Windows 7 installed. Otherwise, some rather unpleasant compatibility problems may arise. And, of course, since we are talking about a voice assistant, it would be a good idea to check in advance whether your microphone and speakers are working.
But the Internet is not required for stable operation of the application. It already “contains” all the necessary resources, including a dictionary.
Minimum System Requirements | Recommended system requirements | |
OS: | Windows XP, 7, Vista (32-bit) | Windows 10, 8 (64 bit) |
Processor frequency: | with virtualization support, AMD Phenom II X4 80, AMD Athlon, Intel i3 Dual-Core, from 1.8 GHz | with virtualization support, Intel i3 Dual-Core, from 2 GHz |
RAM: | from 2 GB | from 4 GB |
Hard disk space: | from 2 GB | from 6 GB |
Video card: | Direct X 9 Vulkan API support, memory from 1 GB | Direct X 12 support, Vulkan API, memory from 1 GB |
Net: | broadband internet access | broadband internet access |
Signum.AI (yes, we are not being modest)
Signum.AI helps improve the efficiency of advertising campaigns and reduce the cost of customer acquisition through precise targeting, synthetic content and automation. The platform combines a marketplace of user data divided into categories based on interests, an AI content generator and comprehensive functionality for launching advertising campaigns on Facebook.
Price:
on request.
Yva
If someone on your team is close to burnout, the Yva system will identify an emerging crisis at an early stage, help analyze the causes and solve problems before the employee decides to quit. Thanks to short weekly surveys of teammates, you can monitor the atmosphere in the team and resolve conflicts in a timely manner. The platform also identifies informal leaders and analyzes their attitude towards managers.
Price:
on request. Demo version available.