![\"\"](\"/static/39f7e13d9759af03a86ecde08091b2ee/85b06/The-Latest-AI-News-in-2024-1920-x-800-px.png)
\n
\n
\n
A new AI service for “parsing” music tracks.
\nThe Mikrotakt neural engine can split any track into vocals, instrumentals, bass, drums, and other audio tracks.
\nIn addition, you can upscale vocal tracks to clean up the voice and adjust the pitch. This feature is ideal for musicians and content makers.
\nThe new Chinese model DeepSeek-V3 becomes a competitor to ChatGPT:
\n– registration requires only your email and password
\n– the neural is quite fast: answers are generated in seconds
\n– Thinking mode, where AI turns into an academic on any topic
\n– the model can read diagrams and generate any texts
\n– files of different sizes can be uploaded
\n– the neuron is good at working with code
\n– absolutely free
\nHailuo AI from China can clone any voice without censorship in a few seconds. Its library includes more than 300 voices in different languages and intonations. The model also allows for full customization: you can change the tone, speed, and timbre and add sound effects.
\nThe Chinese have released a worthy competitor to ChatGPT.
\nThe Qwen2.5-Plus chatbot provides accurate answers, can read documents and images, and is completely free of charge.
\nKrea allows rotating models in its “Photoshop”
\nKrea developers have come up with an incredible feature for their neuro-Photoshop: you can drop any object and edit its position in 3D space.
\nThe Upscayl neuron has received a 16x upscale, so now it can save even the most hopeless photos. The download is free.
\nThe Wegic neural network has become even cooler.
\nNow websites are generated in seconds, and all your edits can be made with simple queries.
\nYou just throw in a promo and you have a perfectly(?) designed website.
\nA working AI meme generator.
\nThe Brainrot neuron follows trends, selects relevant images, and produces memes that are even funny to read. And all this is free.
\nGoogle has released a top free guide to AI agents.
\nIt contains everything: basic knowledge, tools, architecture, training methods, and a step-by-step recipe for creating agents on LangChain and LangGraph.
\nIt is a cool service for learning languages with an AI interlocutor.
\nThe Lingocat neuron imitates the native speaker, gives feedback, and has tasks to improve skills. And all this is free of charge.
\nThe free neural from the Chinese outperforms(?) ChatGPT.
\nA new model called R1 has been released that outperforms the best version of ChatGPT. R1 is twice as powerful as the previous version and it has literally everything you need: programming, math, rewriting, and everything else. It’s worth noting that if you look at the graph of how it “outperforms”, it doesn’t encourage you to switch from OpenAI to the Chinese neural.
\nUniversal AI photo editor.
\nGenerating, replacing, or deleting details, styling, changing the background, and upscaling – and this is not the whole list of functions of the new and free Dzine neural.
\nPrompt generator for ChatGPT
\nPromptEngine works according to the official recommendations of OpenAI. The service also supports other language models: Claude, Gemini, LLama, and Mistral, which can be selected from a list.
\nOpenAI President showed a feature for the o1 model.
\nAccording to Greg Brockman, the o1 model works a bit differently than other types of AI, and to achieve high performance, you need to follow a certain plan in writing promos:
\nthe goal, response format, warning, and context.
\n(The Anatomy of an o1 Prompt attached an example image)
\nChatGPT can now write first.
\nThe neuron will remind you of important things like a real assistant. For example, in the morning, it can provide a ready-made excuse for being late for work, and in the evening, a convincing explanation for a deadline shift.
\nSo far, only 10 regular tasks are available, but by the end of the week, all paid users will receive the new feature.
\nOpenAI offers free tokens for ChatGPT.
\nHowever, there is a caveat: in exchange, they ask for permission to use your promos to train models.
\nThe offer lasts until the end of February. Pro users can get up to 1 million tokens per day, and mini-version users can get up to 10 million!
\nTrump has canceled Biden’s decree on control over AI development.
\nThe document, signed in 2023, required developers to share test results with the government to reduce risks. However, Republicans insisted that this restricted freedom of speech and hindered progress. Now, AI development is unrestricted again.
\nChina will organize a competition between robots and humans
\nIn April, China will host the world’s first half marathon where people will compete with robots. The race will involve 12,000 runners and dozens of androids. Technology versus biology is a real race of the era.
\nAI tokens soar after the Stargate project announcement.
\nOn January 23, Trump announced a $500 billion investment from the private sector in the development of AI infrastructure in the United States. As part of this project, OpenAI, SoftBank, and Oracle will create a joint venture called Stargate. Immediately after this announcement, the AI and big data token market showed significant growth:
\n– AI16Z and AI Rig Complex (ARC) grew by 30% per day
\n– GRIFFAIN and ZEREBRO tokens doubled in price
\n– MetamonkeyAi (MMAI) showed an increase of 152.25%
\n– FLock.io (FLOCK) – by 57%
\nThe smartest o3-mini model will be free!
\nSam Altman said that this neural will be able to solve complex problems in programming, math, and data analysis – all for free.
\nFeatures of o3-mini:
\n– advanced analytical abilities
\n– accelerated data processing speed
\n– ease of use for everyone
\nThe release is planned for the near future.
\nOracle founder supports AI totalitarianism
\nBillionaire Larry Ellison stated that the total surveillance system with artificial intelligence based on cameras, drones, and police recorders will force citizens to “behave in the best way possible”.
\nOracle has recently entered into a partnership with OpenAI and Softbank. Digital dystopia is becoming a reality.
\nThe first Chinese base for robot training was opened in Shanghai
\nThey will learn to help around the house, play football, and assemble cars.
\nThe data obtained in the process will be shared with various companies.
\nCurrently, the center has more than 100 androids and plans to expand to a thousand by 2027.
\nChatGPT can now create any kind of application – a new Canvas feature has been released that runs the code right in the browser.
\n– There are no limits to ideas – we invent and ask AI to draw up a detailed project description.
\n– Copy and paste the request into the chat with the neuron, but add the “/canvas” command at the end.
\n– AI will produce a fully working application in React and HTML in a second – you can test it right away!
\n– There are no limits or restrictions – you can even clone YouTube in your browser.
\nThe feature is free.
\nTop 5 promos for working with any amount of information – the neural will squeeze out all the “water” from the source and give you only concise, verified facts:
\n![\"\"](\"/static/f0c3865203f0eb84451f04f5379021f7/85b06/A-weekly-inspirational-newsletter-1080-x-1080-px-1920-x-800-px-3.png)
![\"\"](\"/static/11ec55989bfb8fbe249559f786b70c9c/85b06/AI-NEWS-2.png)